The Health Connector Autopsy Report

It was pride, not CGI, that doomed the Massachusetts Health Exchange

Released August 26, 2014, 8:27 a.m.

This report contains 31,000 words and is licensed under a Creative Commons Attribution 4.0 license. Reuse and remix with attribution.

A report summary is available here.


This comprehensive, independent report was written to document all the significant events of the failed Massachusetts Health Insurance Exchange, and explain how and why it happened.

The public got glimpses of the story, but never understood the harm it caused for so many people who needed health care. They were not aware that the government knew long before October 1 that the exchange was likely to fail. The government deliberately hid that information from the public over and over again, while tens of millions of dollars were being squandered. Worse than that, senior government officials engaged in misleading spin, and even outright dishonesty, about what was going on.

I do not issue those indictments lightly, but reluctantly, and only after doing months of research, connecting the dots, and consulting experts.

This report contains a lot of new information. It also aggregates media reports to paint a comprehensive picture, such as showing the different kinds of harm to people who needed insurance. I also compare what was known about the exchange project at various times to what was being communicated to the public.

The most valuable part of the report is probably my analysis of the state of the exchange in the year before it was released. I am a seasoned IT professional who has many years of experience, and I have helped to save large, troubled projects. I read all the audits done from the start of the project until after it failed. The taxpayers paid $9 million for these audits, and they showed that the project was doomed from the start.

What motivated me to write this report was the conviction that someone had to tell the entire, harrowing story: someone who saw a close friend harmed by this; someone who was a writer; someone who could understand all the audits and technology documents, and use them to question what government officials had been saying.

This report is lengthy. But isn’t it worth it to spend time to know how and why our government wasted two hundred million dollars and harmed thousands of people? And you heard that right.


Tens of thousands were scared and frustrated, but a significant percentage of those had far more severe consequences: lost payments, loss of insurance, delayed and canceled appointments and procedures, people paying for care with their own money when it was supposed to be covered, and doctors turning away children for lack of insurance. Desperate people who couldn’t wait any longer, abandon their Connector applications, and buy full-priced private insurance to keep their health care. Other people enrolled, paid for insurance, had their checks cashed, but they discovered at the doctor’s office they had no coverage.

How many people had serious problems? According to two references about customer service call records of Health Connector and Medicaid tracking problems that need “escalation”, we know the number is most likely in the thousands. These calls are documented, so the Health Connector leadership knew what was happening in the months after the exchange launches.

But despite having evidence to the contrary, and their knowledge of the severe problems with the exchange before the launch, government officials continue telling the public that everything is fine, that everyone has insurance, that people should keep going to the exchange, and that some are merely experiencing temporary inconveniences.

They know they are not telling people the truth. They never were. They have known for months the exchange is broken. They canceled features that they kept telling people were going to be there. The independent auditing firm on the project has diligently written weekly and monthly reports showing severe problems for a year. Yet the government hid all this from the public because they were afraid of making the policy, and the people who believed in it, look bad.

The public had no idea of the severity of the problems until the end of January 2014 — four months after the failed launch of the exchange. At that time, we learn that the Health Connector Board, including one of Governor Patrick’s cabinet secretaries, had a memo in July of 2013 saying failure was “likely.” Worse, three weeks after that memo, the executive director of the Health Connector talks to the Boston Globe about the upcoming launch, and lies about the great experience users will have on the website.

In September, they get a testing report saying that most exchange functions don’t work. Core features like eligibility determination and selecting a plan — don’t work at all. Without these features, the exchange is useless for more than 200,000 people who need subsidized insurance. Even the few features they do release have little chance of working well because the entire infrastructure is so unstable. Testing is cut back to save time, and on the night before the launch, there has been no testing done with actual users.

But they launch it anyway on October 1. The meltdown is instantaneous. No one can get their eligibility checked or shop for insurance, people who don’t need eligibility checked still have problems as the site is so unstable and can’t handle the amount of traffic that was expected for the launch. In the immediate aftermath, the Executive Director of the Health Connector goes on television, smiles, and tells the public all is well, even though people are jamming the customer service lines at the Health Connector saying they can’t do anything.

Still think this story was just about a website, as the government has been saying? No — it was about people who needed health care and couldn’t get it, like my friend Anna.

The broken exchange infrastructure was never fixed, despite an army of technical people working on it. People were moved from extension to extension, and into temporary plans, at enormous cost to our state. By the spring, over $170 million was wasted on the exchange alone, and the Massachusetts “model” became the embarrassment of the nation. We were ranked last among all state insurance exchanges.

Yet somehow, no one was held accountable for the largest, most spectacular failure of state government in many years. On February 13, at a public Health Connector Board meeting, Executive Director Jean Yang broke down and cried about what had happened. Yet she, along with everyone else in the exchange project, never accepted responsibility, and she remains in charge to this day.

How could all this have possibly happened? How could a state so rich in political, technical, and health care leadership produce such a disaster? Why has the media struggled to explain this story to the public? Why have our political leaders and pundits had so little to say about it? Why didn’t our governor or state legislature act sooner? What should have been done to save this project, and when should it have happened? I will work hard to answer all these questions.

This is a long report, and you don’t have to read all of it. But don’t worry; the tale is certainly not boring. Like some classical stories, it will have a simple root cause: people who had too much pride.


  1. Terminology and Organizations
  2. How to Use This Report
  3. Part One: Why and How I Wrote This
  4. Part Two: Anna and All Who Were Harmed
  5. Part Three: BerryDunn, The Vendor You Don’t Know
  6. Part Four: Crucial Decision Points
  7. Part Five: The Complete, Annotated Timeline
  8. Part Six: Why?
  9. Part Seven: Unanswered Questions

10. Acknowledgments

11. A Little More About Me

Terminology and Organizations

While the public thinks of this story as “The Health Connector” what we are really talking about is a project that was called the “Health Information Exchange and Integrated Eligibility System, or HIX/IES.” However, for the sake of simplicity, I will say “the exchange” when I am referring to this project.

What exactly is the “Health Connector?” It is an independent state agency that helps people find health insurance. It is operated by the Commonwealth Connector Authority, though no one outside the government uses that term. You may see “CCA” in documents, so now you know that refers to the Health Connector.

Also, while the Health Connector is the public face of the sponsorship of the exchange, this project was also supervised by MassHealth (which is our state’s Medicaid program) and the University of Massachusetts Medical School. That is important, as the differing visions of these sponsors were a major problem for the exchange. I will try to make distinctions, but when I am asking aloud what the Health Connector leadership knew, I am usually also referring to MassHealth and UMass. That being said, it is usually Health Connector Executive Director Jean Yang and Health Connector Board Chairman Glen Shor who speak publicly about the exchange. Other important sponsors are Kristen Thorn, who is on the Connector board representing Medicaid, and Dr. Jay Himmelstein, of UMass Medical.

How To Use This Report

For the general public:

This report would take well over an hour to read. If you want to read only the important things, read Part Two to get a sense of the harm this caused for a lot of people. Then read Part Four to see the six times the exchange could have been saved or repaired. Then read parts Six and Seven to understand why this happened, and what we still don’t know.

For the media:

There is a lot of information in this report that has been previously unreported or unknown. There is news in here. Some of that news is in Part Three. Few have seen the audit reports done during the exchange project, and even if they had them, perhaps they couldn’t understand them. I have spent many years in IT, and I read them all. They are damning. They show the truth of the exchange project was always available, and serious problems were known long before July, as previously reported.

The other newsworthy part of this report is how problems with this costly project were deliberately hidden from the public; and how government officials were not honest with the public before, during, and after the launch of the exchange. Even Governor Patrick says things that are in opposition to what is known at the time.

Scan the entire timeline in Part Five to see many examples of this, including several I am sure you don’t about. When I mention something government officials were communicating that is at odds with what was known internally, I use the icon above.

Lastly, Part Seven asks precise questions about what we still do not know. This will be useful for further inquiries.

For legislators, candidates, and reform-minded activists

If you are unfamiliar with the harm exchange failure caused, then read Part Two. Part Three is useful because it shows that even with the truth available, people don’t fix the big problems because they are chasing too many small ones. Check out Part Four to see when this project could have been saved. Part Six will also be interesting because it explains why many actors didn’t act on the problems and why the media struggled to tell the story. Part Seven outlines what we still need to find out.

(As for IT project reform, that’s a whole other issue. However, what the federal government just did is an excellent reference.)

For all:

Links to source material will be inline, and also at the end of this report. There are also notes on the side, which are primarily additional thoughts on the paragraph they are next to. (Click on the one next to this paragraph to see an example.)

Don’t forget to stop by the acknowledgments at the end — many journalists wrote great stories that informed this report, and others helped me get the story right.

Part One: Why and How I Wrote This

My name is Ed Lyons. I am a writer; I am also a software architect who has many years of technical experience; and I have been a part of many large IT projects. I am even one of the people they fly in to help save a project that’s in a great deal of trouble, as I have done several times.

A good friend of mine is chronically ill, and has been part of the Commonwealth Care program for a few years. She went through hell in this process. I am her authorized representative, and I have often handled her needs with insurance companies and MassHealth. I was with her through this process, and spoke to MassHealth often. I used the old exchange and the new one.

After the nightmare subsided in late January, and my friend finally felt safe going to appointments again, I was angry. I wanted to know who did this to her, and how it happened.

I read everything that was out there: the media stories, all documents from the government, and what the government was saying in traditional and social media all along. I decided I would not get lost in the weeds. There are important questions around health care policy and figuring out the full cost of the disaster. I will not explore those things in detail.

I will remain focused on the exchange project, as a project, that was to deliver something people really needed: health care.

For a bit more about me, see the very end of this report.

Part Two: Anna and All Who Were Harmed

First, my friend Anna.

Anna and I were sitting on a curb outside of Children’s Hospital in October. After waiting for 30 minutes on the phone, she hung up the phone and started crying. She said through the tears, “I’ve worked so hard at this. Why are they doing this to me?”

I just looked down and whispered once again: “We will figure this out.”

I had tried hard to help her renew her insurance, but I had failed, despite my experience working with MassHealth on her behalf, as her representative. (She has often needed my assistance, as I am good at language, paperwork, and diplomacy; crucial for anyone dealing with the health care system on a regular basis.)

Anna has several health problems, and had major hip surgery a few months before at Children’s Hospital, in order to fix a birth defect that had become problematic. She had follow-up appointments, gynecology needs, circulatory problems, and she needed several kinds of drugs for terrible chronic pain, an anxiety disorder, and other conditions. Her healthcare was crucial to her, and from October until late January, it seemed like she would fall through the cracks in the foundation of the exchange. This was terrifying for her. For instance, if she temporarily lost her prescription coverage, her pain would be terrible.

Things would get worse after the first week in October. More calls. Letters. Continued frustration with the website. The two primary hospitals Anna used weren’t sure if her care would be covered. The Health Connector wasn’t even sure what kind of plan she would get for a while. They said they would figure it all out. She simply didn’t believe them. Even when the extensions came through, Anna was nervous.

She got billed $900 for a January emergency room visit at a large Boston hospital she had been to several times. They told her plain and simple, “You have no insurance.” It didn’t matter that MassHealth said she had insurance a few weeks earlier. It didn’t matter that state officials said this wouldn’t happen. It was happening. Anna was in a panic. It took us three weeks to fix the problem with the ER visit.

After that bill, Anna couldn’t take the chance with her next two appointments. She canceled one and was able to move the other to February. It wasn’t until mid-January that the people at MGH and Children’s Hospital were confident about the situation, and could finally tell her that she would continue to be covered there. Anna began to calm down, finally.

All the Others

Seeing how bad the site was, and very long wait times to talk to customer service that I had never seen before, I knew others had to be going through this. After all, Commonwealth Care alone contained more than 200,000 people, and that program was ending because of the Affordable Care Act. Everyone was told to go to the exchange.

I am now going to include several cases so readers understand some of the human cost of the meltdown:

On December 7, the Globe writes a story about someone undergoing the fear of losing desperately-needed coverage like Anna:

Gina Kamentsky of Somerville has been trying to enroll through the Massachusetts Health Connector for weeks. Her insurance plan, bought through her partner’s former employer, expires this month. Because of expensive medications and a doctor’s appointment she cannot miss, the 54-year-old artist from Somerville said she needs coverage in January. Plus, the landmark 2006 state health insurance law, the model for the national law, requires her and most others to have coverage.
Kamentsky was first stymied by the widespread technological problems. At the end of October, she filed a paper application. She has been stuck since then in a shuffle of documents and on long calls with customer support staff, but she feels no closer to being enrolled.
“There’s no feedback,” she said. “There’s no way to tell if anything has been processed. Time is ticking away.”

That’s bad enough, but this is the only kind of hardship the government acknowledged was going on. But it gets far worse.

In January, a man wrote an op-ed in the Springfield Republican about “the nightmare of trying to purchase insurance through Massachusetts Health Connector.” His wife is pregnant. She is due April 4. He is forced to renew through the Connector, he doesn’t succeed, and he and his wife are put into a temporary MassHealth plan none of their providers take. Will they have to cancel all their appointments and find new doctors? Pay for everything out of pocket? Here is what he wrote after speaking with a customer service representative at the Health Connector:

She knew of no time, and was under orders not to estimate a time, when we would be able to get the information so we can buy a policy for our network. We are in limbo waiting to pay for insurance coverage that covers appointments with doctors who we trust and know our medical history.
So today we sit, with my wife is seven months pregnant at this point, with no usable health insurance, and we are being delayed trying to get insurance that we will pay for.
We cannot be the only hard-working, middle-class family in this boat, frustrated by not getting answers and being unable to sign up for health insurance.
To date, we have no answers regarding insurance and our temporary insurance expires at the end of January. All we want to do is pick our insurance, pay the premiums, and continue on with our necessary treatments.
In the meantime we have temporary insurance, paid for by taxpayers of the Commonwealth, to pay for the little bit of medical coverage that we have. This makes no sense.
We are just over nine weeks away from the birth of our first child and we have no idea if we’ll have insurance coverage. We cannot even work on picking a pediatrician until we know what insurance plan we will have and who will be in network.
This is a continuing nightmare for which the government seems to be offering zero relief.

On January 9, the Boston Globe tells the story of a man who’s wife is in pain, but can’t get insurance, has to pay $1,000 out of pocket, and delay some of her health care:

For Kamel, the uncertainty has been stressful. His wife, Carol McCorkindale, has two sports injuries requiring specialists and physical therapy.
“We’re kind of in a bind,” he said Wednesday morning. “She’s in pain, so we need to do something. We postponed the really big [appointments], but we do the smaller ones, because we can’t wait until February to get treatment.”
Kamel said he received a call Dec. 20 from a Connector representative who said that his insurance application had been accepted. Kamel, who was not applying for state assistance, specified which plan he wanted and hand- delivered a check for the first premium payment to the Connector’s Boston office the very next day.
The check cleared on New Year’s Eve, but that didn’t stop Kamel from getting a disturbing call at this Medway home. A Connector representative told Kamel he still needed to select a plan. The information had gotten lost, but Kamel could not simply choose again — that might result in duplicate enrollments, he was told.
In the week that followed, even as his wife postponed some necessary medical appointments and ran up about $1,000 in charges for others, Kamel received little clarity about the status of their coverage.
He called the Connector three days in the past week. Each time, he was told there was no record that he had selected a plan, while he was urged to be patient and was promised a supervisor would call him. No one ever did.
“They don’t live up to their promises,” he said.

Also in January, I noticed two friends of mine in Facebook saying they had purchased insurance, gotten back the canceled checks from the bank, and yet found out from their providers they had no insurance. A friend of mine who is a lawyer said she had a friend going through this also.

In fact, the Boston Globe on January 31 writes a story about several people who picked plans, mailed in the payment, and still had no insurance. They are scared and angry about what has happened. Health Connector Executive Director Jean Yang admits there are “40 or 50” cases like this.

Here is an excerpt about a woman having to put off surgery for her hip:

After paying a January premium but not receiving coverage, Heather Foley, 37, of Brewster was told by Connector representatives that she had to pay a February premium, too, or her plan would be canceled. Her detailed log of calls made in an attempt to verify her coverage spans four typed pages.
That the Connector has asked her to make payment on a service she has not received is frustrating, said Foley, a chiropractor who has been putting off surgery to repair cartilage in her hip. In her business, she said, “that would be fraud.”

In February, more comes out. WBUR has a story that describes someone in trouble who merely wanted a private plan:

“I’ve put in all this effort and I don’t have insurance and I really don’t know where else to turn” said Dan Ginsburg, who made more than a dozen calls to the Connector in December and January, trying to renew his private health insurance plan.
Ginsburg, who is from Southborough and runs a small software firm, says he bought a family plan on the Connector’s old website last year with no problem. He was finally able to renew last month and sent the Connector a check for more than $1,200. It was cashed, but Ginsburg’s insurer says they did not get the money and cancelled his plan.
Ginsburg says he supports state and federal efforts to expand health coverage.
“I want to see it work, I really do,” Ginsburg said. “But it just seems fundamentally wrong to me that you could be accepting payment for people’s insurance and then not providing insurance. That, to me, borders on fraud.”

On February 10, the Senate president Therese Murray tells the Boston Globe of the harm she is seeing, and mentions a woman who needed a heart transplant, but lost her insurance for it over a matter of $14. She says her office is swamped and that she doesn’t believe Governor Patrick’s assurances that people won’t fall through the cracks.

At the February 12 oversight hearing, legislators say their offices have fielded many calls from their constituents, who didn’t call because of a website with glitches, but because they can’t get health care. For example, this from Representative Jay Barrows:

“What do I tell the constituent who calls and says: ‘Representative Barrows, I’ve submitted everything, my check’s been cashed, in December. I don’t have my ID card. My doctor won’t see my sick child.’ What do we tell them?” he asked.

More from the hearing via the Boston Globe story:

“We have had people fall through the cracks,” exclaimed Representative Marjorie C. Decker, a Cambridge Democrat.
“We’ve got people out here that can’t afford to wait while we fix this thing,” said Representative Gloria L. Fox, a Roxbury Democrat.

Also, in February, health policy expert John McDonough relays a terrible story in the Boston Globe that he hears from someone “on the ground” in the exchange disaster:

“There is currently no solid process to get urgent medical need cases processed in a timely fashion. Uninsured consumers that have applied and are waiting an eligibility determination have two options to get needed health care right now: 1) Delay your appointment or 2) Pay out of pocket. To give you an example, a man with cancer had a kidney and part of his liver removed in December at a major hospital in Boston. He had a post-op appointment scheduled in early January and was told by his surgeon’s office that he couldn’t come for the appointment until he ‘got his MassHealth figured out.’ The ‘solution’ he was given was to delay his appointment or pay out-of- pocket. We’re hearing stories like this regularly at the advocate and provider tables.
“We have been given ‘contacts’ at the state to help us ‘resolve’ issues and get people coverage, but the system is so, so broken, to the point where these contacts take many days to find applications and then many days to process them. Also, we are seeing erroneous eligibility determinations come out of the new clunky systems they’ve built.”

There were other reports of harm to people. Most of these cases weren’t special in any way. Which means there were surely many more.

(How many people are having serious problems? We don’t really know. In a February 27 Connector Board meeting, we learn on slide 8 in a presentation that 300 people a week are having very serious problems, known as “escalations”. This is 16 weeks since October 1. That means we could be looking at 4–5,000 people. The problems didn’t even end at that point. And not everyone with a serious problem calls customer service. So it is probably safe to say thousands of people had serious consequences as a failure of the exchange.)

I was bitter after what Anna went through. A state government that would spend a lot of energy in 2014 saying it cared about women’s access to health care didn’t give a damn about Anna’s, or that of thousands of women like her.

Months later, I told Anna I was writing this report. I said, “I found out that they knew all along that the insurance exchange was broken. What happened to you happened to other people. They weren’t even prepared to process your application by paper. It was going to be a disaster, and they knew. All the way up to the governor’s staff.” She was livid, and replied, “Of course! That’s because we are the ‘poor people’! This would never happen to rich people and their insurance plans. Those of us on MassHealth? Screw them! No insurance? You figure that out. Nobody notices. Nobody cares.”

And it didn’t have to be this way at all. The government had several chances in two years to make sure Anna got what she needed: a working exchange. You will see that in Part Four. But first, it’s important to know that the truth of the project’s problems was always available to everyone involved, including the Health Connector Board.

Part Three: BerryDunn, the Vendor You Don’t Know

When the launch of the exchange went so badly, you had to ask, “Wasn’t someone paying attention to how things were going?”

It turns out that someone was.

This project had an “IV&V” (Independent Verification and Validation) vendor. This means that someone is watching for quality and compliance with sound project processes. They aren’t caught up in the politics of the project or the mad scramble to get things done. This role is often demanded on large government and private sector projects, because sometimes the client and the vendor can unintentionally cooperate in releasing a system that has all kinds of problems. Someone has to be objective.

On this project, that someone was a firm called Berry, Dunn, McNeil, and Parker. (They call themselves BerryDunn, and so will we from now on.) They got paid $9 million, a handsome sum, to track how this project was going. They wrote weekly and monthly reports, which are now public documents you can peruse. (Here are monthly and weekly archive links.)

So, even though you may have been told that a Canadian company, CGI, was “the vendor” here, there was another one whose job it was to figure out if this project was on the right track.

The audit reports are detailed and comprehensive. They begin in October 2012, talking about how the furniture has been set up for the offices, and they end in November 2013, a month after the exchange fails, with haunting and desperate criticism of a project that has been devastated. One of their final remarks is that CGI is no longer reporting progress and the government isn’t even checking anymore. A month after the launch, after all future releases are now in doubt (the ones that were supposed to save the exchange), and the project has even lost the ability to be properly audited (page 55):

CGI does not regularly report project schedule metrics. The Commonwealth does not routinely monitor the CGI project schedule. There is no process in place for CGI to report changes to the schedule and impacts of those changes.

You might ask, “When did things start to go wrong?”

Almost immediately.

That’s right, the second report, on November 30, 2012, starts with:

This HIX/IES Project is a time-constrained project and the system design phase is behind schedule. CGI and the Commonwealth have acknowledged this issue and are taking steps to understand the root causes, mitigate the impacts, and update the project plan and schedule.
Reasons for the delay are numerous. They include: extended and multiple deliverable review cycles caused by the submission of poor quality deliverables (risk R02E.2), ineffective communication between individual reviewers and agency approvers, insufficient documentation of reviewer comments and vendor responses, lack of availability of Commonwealth resources to attend JAD sessions and complete reviews, and complex material that requires the involvement of many business stakeholders.

Who is mostly to blame at the beginning? It isn’t CGI. It’s mostly the government side.

The project gets worse all the way to October 1. The audit reports document the increasingly complicated problems and complicated responses. The number of issues and the responses to issues become so overwhelming that you can see that both sides of the project have lost their way. The audit reports have a comic feel to them, as BerryDunn remains emotionless amid the chaos.

So you get this item at the end of April where they point out that CGI is not even reporting when they miss deadlines:

Or this one, in late May, where BerryDunn says the data layer is fragile and dangerous and has 143,000 coding violations in 50,000 lines of critical code. (This absurd number means coding standards are simply not being followed.) They say this is a big deal and a high-risk item from now all the way to the launch. But the government side decides this isn’t critical and puts it on hold.

Stepping Back

I will mention the audits often in this report. If you take time to read them, and especially if you have technical project experience, you will see that this project was always in trouble. Did the government ignore these reports? It’s hard to see how, as the reports themselves often mention problems reported to senior leaders and “executives” in the Health Connector, Medicaid, and UMass.

So if anyone in charge of this project pretends that they didn’t know what was happening, remind them that we paid $9 million to a firm who told them weekly and monthly how things were going.

Lastly, you also have to wonder why the federal government would keep giving extraordinary amounts of money to a project that is obviously in the hands of people that couldn’t manage it.

Part Four: Crucial Decision Points

This abbreviated timeline focuses on six moments in the two-year project where something straightforward could have been done to turn things around. The analysis here will not be in hindsight, but what an experienced, competent IT project manager would have done in those circumstances, not knowing what was to come.

These decision points will also appear on the complete timeline and be noted as such. However, they will not be discussed in detail there. Here, we want to capture the context of the situation.

As someone who has been in computer software for many years, I have seen many projects that went well, some projects that had problems, and some projects that failed, despite desperate efforts to save them. I am well aware of what kinds of things go wrong, when you can turn things around, and what kinds of strategies are successful. I am also familiar with how vendor contracts are written, the processes that dictate how things get done, and all the ways clients and vendors can sabotage a project.

I have identified six crucial decision points in the project when sound management practices and good government would have made a big difference and greatly increased the chance of saving the project. At the minimum, the public could have been informed, and help might have arrived.

Here are the points:

  1. The Start of the Project (Fall of 2012)
  2. The Appearance of Severe Governance Problems (End of 2012)
  3. CGI Can’t Deliver Contract Scope (January 2013)
  4. The Failed Launch (October 2013)
  5. The Success of the Federal Tech Surge (December 2013)
  6. The Exchange Shakeup (January 2014)

Decision Point 1: The Start of the Project

Fall 2012

This decision point will have the most detail of the six, as how this project began made the most difference in what was to come. I have divided the content into five separate areas:

  • CGI
  • Management of the contract with CGI
  • The contract itself
  • Too much money
  • A lack of technology expertise on the Connector Board


While many things were being done to prepare for the Affordable Care Act, the building of the exchange was scheduled to begin in March of 2012. Four months later, they selected CGI as the primary vendor and signed a contract in October. (Yes, that took far too long.)

Part of the reason they were chosen was because they were also working on the federal health exchange. The thinking almost certainly was that it would be an advantage as they were building the federal data services that state exchanges would be talking to, and they understood what everyone was trying to do.

However, it could be a problem to have the same firm on both ends of the federal/state exchange divide. Having CGI in control means that these interfaces could keep changing, as there is less pressure from outside to create a durable contract early on.

CGI was also chosen to work on exchanges for six other states: California, Colorado, Hawaii, Kentucky, New Mexico and Vermont. A serious liability. Just how many good software people can one firm have for one type of project? Big firms can quickly hire lots of people, but how good will they be? In my experience, you can’t go hire hundreds of people and get quality.

Lastly, what if the firm had major problems? If CGI failed, you’d be stuck with both the federal and state side of the infrastructure in trouble. And what happens if one of their other state exchange projects has trouble? Would they pull their best resources from our project to help that one? (Moving people around happens all the time in this industry.)

CGI should not have been hired.

Management of the contract with CGI

The first thing to know is that the contract with CGI was handled by the UMass Medical School. Specifically, the job was given to Dr. Jay Himmelstein, a professor at the medical school and Director of the New England States Collaborative for Insurance Exchange Systems. He was assisted by another professor, Dr. Michael Tuttle, listed as the project manager on the contract with CGI.

Why was UMass Medical in charge of this contract and CGI? It’s unusual and an important part of the story. The first significant installment of federal money for our health exchange, $35 million, is given to UMass in February of 2011. They have an idea for a regional insurance exchange where states in New England would establish a collaborative effort to build excellent exchange technology and processes. Himmelstein was an innovator here, and wrote a paper about the challenges involved. His organization gets the $35 million to build out this vision, even though the other New England states don’t participate. (This matters because in the spring of 2014, the governors of Connecticut and Rhode Island say we owe them millions for their share of a regional grant they got nothing out of. This remains unresolved.)

The big federal money for the state of Massachusetts doesn’t start coming until more than a year later, so it must have seemed like a good idea to have UMass hire CGI and manage them. However, by the time the contract is signed in October of 2012, the state has $60 million of its own federal grant money. Here is the key thing here: Once it is clear there will be no regional exchange and the state of Massachusetts has more money in the game, the contract should have been handled by someone else in state government with more appropriate experience.

In fact, when Governor Patrick removes UMass from the project entirely in February of 2014, he says that the medical school is not in a position to manage a vendor who is having problems and should be “on a short leash.” But the vendor is having problems from the start, and UMass is never right for this job.

The contract itself

40 hours of lawyering would have saved $200 million.” — John Miller, candidate for Attorney General, on the final settlement with CGI

My boss on my first real job once told me, “You only need a contract for when things go wrong.” When you look at the timeline and all the problems CGI had, and then see how we paid them millions for delivering a system we couldn’t use at all, you have to wonder what was in the contract.

If you want to see the contract, it is available at this link. Warning: the contract and all related attachments is 1,182 pages, and is 87MB in size. It will be difficult to get any value out of reading that packet if you aren’t accustomed to these kinds of documents, or don’t have a good reason to review them.

Looking at the contract settlement announced in June of 2014, where CGI gets most of the money it wanted, it appears the fault is with code ownership in the contract, as Governor Patrick says on February 6 at a press conference:

Looking at the contract settlement announced in June of 2014, where CGI gets most of the money it wanted, it appears the fault is with code ownership in the contract, as Governor Patrick says on February 6 at a press conference:

Patrick did not rule out ending the contract with CGI, but said that would be complicated due to the complexity of the website and CGI’s ownership of the computer code.

But in fact, if you read the contract, there is nothing in there posing intellectual property problems. Had CGI owned the code, that would have been a problem. But they didn’t. They were allowed to use components they owned, but they had to get permission from UMass in order to do that. There is a specific place in the contact to list “Contractor Property” that CGI would keep and license to us, but that section (page 43 of the enormous contract document packet) is empty. So unless UMass gave CGI permission to do this during the project — and who knows what went on during the chaos near the end — this isn’t the issue.

So what was the problem?

The contract has a lot of standard language for software contracts. But it refers to a statement of work where CGI is paid for “code drops” along the way. These milestones are generic and are not related to specific functions or passing any kinds of testing. This is great for the vendor, as they can get paid just for putting in work. Of course, the last phase of the project before launch is testing all of this, so it is in their interest to do a good job, but in this case, CGI never gets the code finished.

They did deliver a lot of code. But it was of terrible quality. The contract should have made it more obvious what needed to be completed at what time, and what the expected quality was, instead of using generic code drops that aren’t related to success.

(Where do the idea of “code drops” come from? Dr. Himmelstein’s big vision for a regional health exchange in 2011. They are not a standard software development lifecycle feature.)

Try thinking of the problem this way: imagine you hired a wedding planner who merely committed to provide batches of wedding activity along the way. But you don’t make sure that the dress is ordered way ahead of time. Or that the church is booked six months out. Or that the guest list is more important than the table centerpieces. If you get three months into the wedding preparations and you don’t have the reception hall booked, you are going to have big problems, even if 25% of other planning is done. The CGI contract should have identified what key pieces had to be in place early on, to lower the risk.

Too Much Money

When I heard in mid-2013 about how much the state was being given to build their exchange, I was stunned at the number. More than $69,000,000? I was vaguely aware of what functionality they were looking for, and that they would have to connect to federal data services. It seemed like 5 times what was necessary, based on my experience in large projects. (Even later, when I learned more from the audit reports and contract documents about the eligibility system and lots of other back- end components, it struck me as at least three times as much money as was needed.)

I then heard it cost them many millions of dollars to build the existing exchange site, which was little more than a brochure that advertised an entirely paper-based system for people who needed subsidized insurance, with links to insurance companies to people who didn’t qualify for subsidies.

I had used that site often to find out about the insurance carrier options for Commonwealth Care, and to find out where to mail all the paperwork that went back and forth. That they spent millions to create such a simple site should have been a warning to all.

I knew from years of IT project experience: too much money for a project is a bad thing. You can get distracted by things you don’t need, and forget why you started the project to begin with. Here’s a useful metaphor: Imagine that you are sending your son off to college, and want to give him money to handle all his expenses so he has a great college experience. Many parents arrange a regular deposit of money. How many parents would say, “I have put $50,000 in your checking account. Have a great time in college!”

I don’t think anyone would do that. It isn’t hard to see why.

A lack of technology expertise on the Connector Board

If you’re going to build a website for your daughter’s soccer team, you don’t need a seasoned IT professional to manage it and the risks. If you’re going to spend $70,000,000 over two years to build an enormous system to handle the health insurance needs of more than 250,000 people, you have to have someone who has experience doing that.

No one on the Health Connector Board, at UMass Medical, or at MassHealth, had that kind of experience. All the expertise at the top level was inside CGI and at the validation firm, BerryDunn. So the leadership at the Connector, MassHealth, and UMass had no way of weighing the risks or seeing the clear signals that someone like myself would have seen all along the way. They should have appointed a special person, who wasn’t on the project day-to-day, to watch the project and attend board meetings.

An interesting aside: Did the Connector have internal technical people available who knew what was happening? The salary schedule obtained by the Boston Herald in mid-2014 shows at least three on staff at the Health Connector, with good salaries, who seem to have been the right people to figure all this out:

  • Scott Devonshire, Chief Information Officer $164K
  • Jason Hetherington, Director of Technology, Strategy, & Services $124K
  • David Lemoine, Director of IT Implementation $107K

Did they see the board memos? What about the audit reports? Any IT director of anything is qualified to read the BerryDunn reports, which are disturbing. Someone should ask these people what they knew and when they knew it. For instance, Scott Devonshire was seen by Dr. Himmelstein, the visionary behind the exchange, as the candidate leader from the Health Connector. We don’t see him in the audit reports, but he does speak at the November 10, 2013 Connector Board meeting about problems with the exchange. (However, he does not seem to understand the severity of the problems, according to the presentation and the discussion in the minutes.)

Regardless of their roles, the taxpayers paid these three people almost $400,000 during the time of this one-year project. What did they know?

Decision Point 2: The Appearance of Severe Governance Problems

End of 2012

If you want to know what was the primary management problem of the Health Exchange, just listen to how Health Connector Executive Director Jean Yang explained things on television in April when she was asked if the meltdown was her fault:

“I am always happy to look at myself and everyone around me, critically. That’s always a good thing to do. But I would not say that this is an individual issue, at a broader level, it is a governance model weakness we have had over this project because this is a multi-agency effort… this is a committee-driven governance model, which is predicated on consensus, and when there is no consensus, there is no authority, there is no effective decision-making. ”

It certainly sounds like no one was really in charge.

A Boston Herald story about the exchange referred to a serious clash between the Health Connector and MassHealth. They couldn’t agree on things, and it was decided CGI would meet with the two groups separately. By November of 2012, there were dozens of action items on the client side that were unresolved. The audit reports from BerryDunn confirm the problems in decision making from the beginning.

However, for reasons I cannot understand, this situation wasn’t remedied until the end of January of 2014, four months after the exchange blew up.

Decision Point 3: CGI Can’t Deliver Contract Scope

January 2013

The BerryDunn weekly and monthly audit reports, which start in October of 2012 and go to the beginning of November 2013, show that this project had problems from the start on both sides. (Yes, the government should have started the project in March and not October.) But CGI’s failures early on are unmistakable. They were unable to keep any promise they made.

The most important breach is in early 2013. From news accounts, in late January, CGI makes it clear it cannot make the launch date — one that was not flexible — with the required functionality. In February 2013, Dr. Himmelstein and the Health Connector Board agree to reduced functionality and a shorter schedule. (They end up finding out in April that CGI can’t even deliver what they agreed to in February.) Kudos to the BerryDunn people, who frowned upon this new arrangement.

What should have happened in January instead? A few things:

1. Inform the public. There are nine months to go, there is plenty of time to fix what’s wrong. But they needed to alert the public, tell us that they were not on track to succeed, and that a lot of people would be in big trouble if they failed.

2. Bring in new people. There is time to bring in new management and a crisis technical team. This, not a year later, is when a special manager should have been appointed.

3. Take the contract away from UMass. It’s time for a more seasoned manager to play hardball here.

4. Audits. The technical audits that happened a year later (MITRE, Microsoft) should have happened now. Yes, the BerryDunn reports are useful for seeing problems and plans, but we can’t tell exactly what was complete at this time, and someone needs to be more blunt about the problems.

5. Prepare for a paper-based process. There is a need for “Plan B” here and now was the time to get ready for a large amount of paper-based submissions.

But while we are here, let’s go back to the issue of the contract.

In late January, CGI says they are way behind. (Some of this is the government’s fault as they have been bickering about what they want to build and they have delayed the project.) They come to Himmelstein and say they need to reduce the functionality, combine “code drops”, and tighten the schedule. At this point, they are 3 months behind on a 12-month project. Himmelstein agrees to amend the contract. (Though I am sure he was very unhappy about it, as his grand vision is failing.)

At this point, the contract should have been completely rewritten, at a minimum. It should have been taken away from UMass now, and not a year later.

So while the contract should have been written differently, the bigger problem is the people managing the project. First, the government is causing lots of problems in what is asking CGI to build. Then, CGI isn’t fielding the right people to get the job done. The auditing firm in place (much more on them later) is documenting the chaos all through the project. A seasoned IT project manager in February would not have continued on this path. In fact, CGI falls further behind every month until the fall.

When you read everything all the way to the finish, you understand why the settlement was so bad for taxpayers. It’s because the government was a major reason why the project went poorly, the contract didn’t have the right protections, and the people who managed it ended up enabling the failure instead of preventing it. (For more on the contract settlement, see June 20, 2014 in the complete timeline.)

Decision Point 4: The Failed Launch

October 2013

Everyone supervising the exchange knew they were rolling out only a small number of the exchange features they needed, inside a very troubled project. Once again, the BerryDunn audit reports are illuminating. The project is on fire. Testing time has been cut to almost nothing. There is no user testing. There is no performance testing. At all.

The code CGI is writing during the death march is awful. From the September 30 audit report:

The following coding violation statistics are based on the ACR Report submitted by IV&V on 9/30/2013:
— Coding violations increased by 53% during the past month — The # of Critical violations increased from 785 to 1,619 during the past month — The # of Major violations increased from 13,023 to 21,133 during the past month

Now do you see why the exchange could not be fixed? They were inserting more and more problems as they were trying to finish their work.

I can’t explain all the problems in the pre-launch audit report to a non-technical person, so I took a screenshot of the middle 20 pages of it from a bird’s-eye view. Every bright red box you see is a critical issue that hasn’t been resolved (20 in these pages alone):

Sept 30, 2013 Monthly IV&V Report

So if anyone wanted to know if there were any “red flags” visible before the launch, well, let’s say that there were.

It also failed the formal testing done in early September, where only 11 of 52 conditions of ‘readiness’ were ‘satisfactory’. That means they weren’t ready at all. Not only that, but we have to understand what wasn’t working prior to the launch. Here is the startling reporting from the Springfield Republican in early January 2014:

The automated determination process, through which the connector determines what subsidy someone is eligible for and what insurance program they should be in, has not worked, forcing the connector to process applications by mail.

This isn’t a nice-to-have feature! Not having this means the exchange is dead on arrival for most visitors, who need subsidized insurance!

Because there is no higher-level testing, the auditor says they will go live based on the strengths of lower-level component testing instead of higher-level testing or user testing. (I simply don’t have words to condemn this level of malpractice.) This is like saying that a car is safe because the individual components worked in a laboratory. Nobody has tested an actual car that uses them. No one has ever driven the car, either. But 250,000 people are going to start driving this model tomorrow morning.

What did they think they were launching? As it blew up, no one could know for sure unless they saw the scope buried in one of the final audit reports, (page 4 and 5 of the October 2013 report, dated November 7.).

Everything that Anna and the people who need subsidized insurance require is not there. All the exchange has is a few reporting tools, anonymous browsing (you can see all the plans you don’t know if you can get or afford), and unsubsidized customers can find plans. All the “e-commerce” stuff that was promised before, that Jean Yang says is there on PBS on October 1, is not there. (Some of these tools are slated for an October 29 release that gets moved to November 17, and then never happens.) The site doesn’t even tell people that these features are not there or not working.

Never mind all the other stuff that’s wrong with what they thought they were launching. The overall system is so unstable that even the small stuff that is supposed to work fine — suffers. (So, for instance, two weeks after the launch, I visited the exchange again in the middle of the night when no one else is hitting it, and I did see health care plans listed on a page. I wasn’t logged in and I had no idea if Anna qualified for any of them — it was a generic list — but at least I saw them. During the day, the site was so broken I never even got to that page.)

Let me be clear: No competent architect or IT project manager would have agreed to launch this project under any circumstances. Open enrollment lasted months. It should have been delayed at a minimum.

But they didn’t delay, and it blew up.

Worse, there is no sign that CGI is getting better at this project or that they can keep their promises to add the features after the launch. Even worse, it is soon realized that CGI has screwed up the federal side of the exchange, as isn’t working either. Add to that, other state exchanges CGI is working on have also failed, including in nearby Vermont.

It simply is over for CGI at this point. There is no way they will get things working in time for the end of open enrollment, not when the system is this unstable, and other exchange projects they are working on are on fire.

Other than delaying the launch of the exchange, what should have been done?

1. Move to end the relationship with CGI. What happened months later should have happened now. There is no point in having them continue. A savvy project manager would have reached out to his or her counterparts in the six other states CGI was working for. I would want to know if those projects were in trouble, and to what degree. They would have been, considering some of those launches didn’t work out, either. I would have made the case that there is nothing they can do now to fix such a broken system.

2. Get a new vendor, focus on the consequences of paper applications. (This is what happened with Optum and hCentive in February. It should have happened after the failed launch.)

Decision Point 5: The Success of the Federal Tech Surge

December 2013

In mid-October, the White House, frustrated at their own failed launch, decided on what would be known as the “Tech Surge” — bringing in top-level technology people to fix problems and get things working again. This plan was made public soon after, and Time magazine did a nice cover story about this. The Health Connector should have done the same thing, as we have that level of talent here. (I know some of them personally. A call from the Governor and a few drops from the infinite pool of federal dollars would have made it happen.)

In fact, the White House and its health care people even came to Boston on October 30 — well after both the federal and our state exchanges were known to be failures — and had a huge celebratory event in Boston at Faneuil Hall. The Health Connector staff celebrated it in social media:

You would think that two humiliated teams would be swapping notes on how things were going, and the White House people would have mentioned how they were handling their very similar problem, no?

(In fact, it takes until the February 13 Connector Board meeting — 10 weeks after the completion and success of the federal tech surge, where the board discusses trying the same approach.)

Would it have worked? Even if things couldn’t be fixed, a highly qualified team would have made it clear very quickly that the Connector had to give up on what was there and move on, rather than wait four more months to do that.

According to an article in MassLive, Yang said she was casting about for local IT expertise in late January. But this idea never goes anywhere.

Decision Point 6: The Exchange Shakeup

January 2014

In very early January, the Health Connector, and even Governor Patrick, finally understand they need to change course. They arrange an audit from the MITRE Corporation, to be delivered sometime in late January. Governor Patrick and Executive Director Yang see an early draft. There is nothing in it that should be a surprise to anyone who has seen the audit reports.

Also, in late January, Microsoft does an additional audit. (That is hidden from the public for a month.) The Health Connector is evaluating a new vendor to take over (Optum) and they are looking for someone to lead the recovery, which will be Sarah Iselin, from Blue Cross Blue Shield Massachusetts, to be appointed February 6, when the official shakeup begins.

But everyone knew enough in January to act, especially as it is known that CGI has had serious problems in other states and at the federal level. In an article on MassLive in mid-March, Jean Yang says that the functionality promised shortly after delivery never happened. (The BerryDunn audit reports document this.) This means that they are just fighting fires in October rather than building the missing features.

So, here is what should have happened at this time: CGI should have been dismissed now, and not two months later; a new vendor should have been brought in at this time, and not in early May, when hCentive was hired. The two reports that were done should have been released to the public immediately. The consequences of waiting are that almost all of Sarah Iselin’s tenure is wasted getting to these decisions, rather than moving forward from them.

Part Five: The Complete Timeline

December 2011 — August 2014

This is a comprehensive account of progress, problems, and communications from the Health Connector, ordered chronologically. There is much information in here that will be surprising, even to those who have been following this story. Only by reading the whole tale does one finally see the tragedy and insanity of it all. (Even I found myself laughing at some of the things I discovered.) The items in the timeline are factual, and they are all sourced inline and later in the endnotes. I provide a lot of commentary that goes with them.

Important milestones in the story will be tagged with this icon. If you don’t want to read all the events and descriptions, you can just jump from milestone to milestone to see what matters the most in this story. I will also use a variety of other icons to characterize events along the way, and I will include the decision point icons from the previous section to mark those six occasions.

2011: A Man With A Vision

The best way to understand how this story started is by seeing the vision of Dr. Jay Himmelstein of the UMass Medical School. He envisions a regional health exchange where everyone shares practices and technology components. His ideas, which you can see in this presentation, are quite good. So good, that the federal government gives him $35 million to build a regional health insurance exchange. He invents the governance model that even Jean Yang later admits doesn’t work. Its born on slide 18:

He views the leader from the Health Connector in this project as CIO Scott Devonshire, who I heard was involved with the exchange project. He also sees the project manager as Janice Baker, from the Executive Office of the Department of Health and Human Services, with assistance from a UMass colleague, Dr. Michael Tutty. Tutty’s name is on the contract and in project audits at the beginning of the project, but his name fades away. Baker’s name is there all the way through, assigned to an endless series of crises. (Her name has been absent from all the news reporting. She appears to be the main project manager for the exchange. Someone should find her and ask her about this project and all the audits she starred in. She appears to be a long-time consultant and project manager who works for EOHHS.)

I liked Himmelstein’s vision. If only it could have been implemented without software. He even has the idea of Integrated Eligibility (where many programs use one system to figure out what someone is eligible for) even though it makes the first version of the exchange more complicated. (The idea is great; the implementation looks hard to pull off. And not for software engineering reasons.)

However, his vision seems uninformed by what happens with committees and especially what happens with large software projects. His simplistic notions of “reuse” in software components were abandoned many years ago. These days, we meet the same shared-infrastructure objectives in other ways.

In fact, the more I learned about his vision, the more I saw the choice of CGI as disastrous. However, to his credit, at least he sees that the coding of this project should begin in April of 2012, not in November. (Interestingly, in slide 17 of his presentation, we see where the mysterious idea of “code drops” comes from.)

What matters to us here: The exchange project began as an academic collaboration exercise. By January, this model of accomplishing a health exchange had broken down, and needed to be changed dramatically.

March 2012: Exchange Project Begins

The federal funding of this project begins in 2011, but building the exchange doesn’t start until 2012.

The University of Massachusetts Medical School is put in charge of hiring the primary vendor and managing the contract, as they got the first big installment of federal money.

First Decision Point

The beginning of the project is the first of six crucial decision points in this project. Details are in the previous section, but the key points are: they should never have hired a vendor on the hook for so many other health exchange projects and the federal exchange; they should have written a contract with a different delivery structure; and they should have had someone overseeing the exchange at the board level with experience in large software project management.

August 2012

The Health Connector requests a change in requirements that adds six weeks to original scope of work. It also tells CGI that it wants to build “ the absolute Rolls-Royce of any health exchange.”

November-December 2012

Second Decision Point

This is the second occasion when a big change would have made a difference.

There is great disagreement between the Health Connector and MassHealth about the project. CGI has 80 outstanding items on their to-do list that the client needs to resolve. There are many duplicate meetings and delays already, and the different organizations in charge of the project (Health Connector and MassHealth) are strongly disagreeing on the project direction. Things are so bad that project manager asks CGI to meet with MassHealth and the Connector representatives separately.

The collaborative approach that Dr. Himmelstein envisioned is not working out very well. It is time for a more disciplined approach.

January 2013

Jean Yang becomes Executive Director of the Health Connector. Glen Shor was the previous Executive Director. Shor becomes Chairman of the Connector Board and accepts an appointment to Governor Patrick’s cabinet as Secretary of Administration and Finance. Also at this time, John Polanowicz becomes Secretary of Health and Human Services, which oversees Medicaid. (His mysterious absence from this story is discussed later.)

Third Decision Point

CGI can’t make the launch date with the functionality needed. At this time, the Health Connector and MassHealth should have done three things: inform the public, bring in a new management team, perform the audits that happened a year later, and prepare for paper-based submissions.

February 2013

The Connector decides on reduced scope. A key event here, that the public learns in January 2014, is that CGI and Dr. Himmelstein (who was managing the contract for the Connector side) agree to a contract modification because of CGI couldn’t get enough done in time. The fact that the contact was up for discussion is striking to know in hindsight.

April 2013

CGI says in late April that it has doubts it can even deliver the reduced functionality that they agreed to in February. They ask to push back big parts of the application. But according to a memo obtained by the Boston Globe, the medical school (who is managing the contract) refuses, saying it would have “a major disruptive impact” on plans for the federal law.

This is the death-knell for the exchange as envisioned. Himmelstein, unfortunately, can’t separate the launch date from what CGI can deliver, which he doesn’t understand are not related. CGI, even under great pressure, can’t keep any of its promises. If I were the manager, I would have quickly found out how the other states were doing with CGI. (Some of those projects must have been in trouble also.) At this point, new people should have been brought in. CGI can’t bring in new people as they are struggling elsewhere.

May 2013

According to memos obtained by the Boston Globe, CGI and the UMass Medical School agree to a risky timeline, less testing, and an exchange with less functionality than what was even agreed to in the amended February contract negotiation.

Cut testing time for a vendor who is already struggling? Inexcusable. They needed additional testing. This is a recipe for disaster, and no competent IT manager would have agreed to this. Certainly this is another occasion to ask why a medical professor was handling these negotiations.

July 1 2013

According to a July 1 Connector Board memo, 10 days earlier, CGI said they are “2–7 weeks behind” where it needs to be at this time. Worse, the memo concludes, “there is a substantial and likely risk that CGI may be unable to deploy into production the scope of HIX/IES functionality on October 1, 2013.” This memo also lists a timeline with all the deadlines that CGI has missed until this point.

Imagine if the Boston Globe had seen the July memo then. Why was this hidden from the public? Why wasn’t this disclosed at the upcoming Connector Board meeting?

Dr. Himmelstein, handling the contract, talked to CGI, and said this, according to the Boston Globe:

Himmelstein asked the company to make a plan for fast-tracking the project and explained that a failure to launch would hurt the reputation of state agencies involved, run counter to federal law, and cost the state more money. It would “prevent the citizens of the Commonwealth from accessing programs that they are entitled to under the Affordable Care Act,” he wrote.

Let’s stop and think of what this means. CGI can’t make the reduced scope. They are weeks behind. And now it looks “likely” that they can’t get what they are supposed to up and running at all. They know this from a Board memo that everyone has seen. 250,000 are now “likely” to go to an exchange that doesn’t work, and will be unable to get insurance. And the Connector has no way to process that many paper applications. There is no “Plan B’ available.

This Is An Emergency!

Is the public told? No. In fact, the opposite happens: the roadshow begins. The Connector has to start doing lots of work to publicize the launch date and tell everyone in print, online, and in news interviews — that things are going to be fine, when they knew things were not going to be fine at all.

Does the Connector Board know what’s happening now? We can’t even tell. In the Globe story, one board member, Nancy Turnbull, says she knew of the website problems all along. Another, Jonathan Gruber, says the problems with CGI were not brought to the board’s attention in July. But certainly Himmelstein’s Board memo was explosive, and would have been noticed.

Mid July, 2013

The Health Connector sends out its summer newsletter to its broker community with statements that are not only false, but are the opposite of what the Connector Board knows about the state of the exchange and the integrated eligibility system.

We get this language from Executive Director Jean Yang:

Our technology will be state-of-the art as we build a new integrated system of eligibility determination and shopping, which will serve both the Health Connector and MassHealth, the Commonwealth’s Medicaid agency. But the Health Connector will also continue to serve small businesses, brokers and non-subsidized individuals – and our new state-of-the art platform is being designed to support the unique needs of these users as well.

The project has already been massively scaled back and is now “likely” to fail. The integrated eligibility is a mess and will not work for at least a year. Why did she say this?

It gets worse. At the end of the newsletter, under “completed milestones” we get this:

Leveraging federal support, custom build a new integrated system of eligibility determination and shopping, which serves both the Marketplace and MassHealth.

This is utterly untrue.

It gets worse.

Even though the Connector Board knows that things are going terribly from multiple sources, the agenda and meeting minutes for board meetings from July to October do not mention problems with the exchange.

July 19, 2013

The state government (the BerryDunn audit reports doesn’t make it clear who) gives a presentation to the federal government (CMS) about contingency planning if certain features are not there. They create three tiers of features, from very important (Category 1) to not as important (Category 3). They present plans for workarounds for lots of little things going wrong.

It is important to note that none of the Category 1 features listed were delivered on October 1.

July 21, 2013

Health Connector Executive Director Jean Yang visits the Boston Globe to talk about the new exchange, three weeks after she and the Connector Board are told in writing that failure is likely. Is there any hint of this? No. She talks a lot about policy, makes everything sound great, and says, “Our goal is to offer a significantly better user experience, an easy-to-navigate process.” However, the launch is only 10 weeks away. The reporter should have pinned her down and said, “Goal? Aren’t you about to deliver that?” Because after all, she knew she was absolutely not going to deliver that.

Yang could have said, “We have having big challenges in getting the exchange ready for all those who need insurance and we are working around the clock to be ready.”

But that would have resulted in the Boston Globe asking questions she had no interest in answering. That might have alerted the public of what was to come.

July 29, 2013

In the original contract, this is the date when CGI is supposed to be done with all coding, leaving the next two months for testing, monitoring, and preparation. If they were still working off the original scope of the project, they would be, in my view, about six months behind. (On a one-year project!)

Early August, 2013

The Death March Probably Begins

When a software project is hopelessly behind, and everyone is angry, and the programmers have to work around the clock and on the weekends in a desperate attempt to catch up, we have a specific name for it in our field: The Death March. Part of the name comes from the fact that programmers work too many hours, don’t sleep enough, become paranoid, and become desperate. They make mistakes. And in most cases, they don’t succeed.

I don’t know at what point CGI decided that it had to tell its people to do whatever it took to save this project, as millions of dollars and their reputation were at stake; at what point the programmers were at their desks more often than they were not. I know that it must happened at some point, in light of the audit reports and the way the launch went. My best guess would be that the Death March must have began in early August.

I spent time on this issue for good reason: IT project managers know these marches don’t succeed unless they involve fresh people, and that code written under these circumstances is often full of new errors. (The audit reports show this is happening on the exchange.) You think you are making progress, but you aren’t. This is one of the big reasons the exchange couldn’t be fixed after it blew up. There were defects being added deep in the infrastructure for months.

August 27, 2013

The launch is 34 days away. The exchange software is still in terrible condition, and an audit done in a couple of weeks will show that. But what does the Health Connector do? Launch a dedicated webpage, still there today, that tells everyone about “The Better Connector” that will be available October 1, with all kinds of great features that won’t actually be there. People are told to go to the exchange to apply for insurance, and the exchange only.

Anna and I saw the “Better Connector” page as we were preparing. We assumed it was all going smoothly.

September 10, 2013

Website Testing

With only a few weeks to go, the exchange undergoes a suite of functional testing that really should have happened weeks ago. Only 11 of 52 functional areas get a rating of “satisfactory.” The rest get “partial” or “missing.”

After this frightening result, the expectation was that the missing pieces would be added in late October. Worse, as we learned much later, (and I wrote about in Decision Point Four) the components that determine your subsidy level and what kind of plan you should get — don’t work at all. They never end up working. The site simply cannot go live without these things, as this is what people like Anna must have.

This was a “readiness” test that showed they weren’t ready, with only 20 days remaining. The audit reports also confirm how bad things are at this point, meaning that even things that “work” are in jeopardy, as the whole infrastructure has fundamental flaws. At this point, any competent project manager calls off the launch. It’s over. There isn’t time to stabilize the overall system, even if you decide that only a few features are going to be there initially.

September 11, 2013

Health Connector Publishes Consumer Guide

The day after most of the tests fail, the Health Connector publishes a 35-page document called, “The Massachusetts Consumer’s Comprehensive Guide to National Health Reform.”

Of course… everything is awesome! Go to page 16 to see what will be happening to the website:

Massachusetts already has this type of Marketplace, and it is known as the Health Connector. However, the national health reform law will mean changes to the Health Connector such that it will look and feel somewhat different than the Marketplace consumers may have known previously.
Generally, the changes that are coming for the Health Connector will mean more choice for consumers (e.g., more participating health plans, dental plans available for purchase), coupled with additional decision support tools and a more consumer-friendly shopping experience.


But it gets worse:

Beginning with the Open Enrollment that starts as of October 1, 2013, the Health Connector at will serve as the Commonwealth’s access point for individuals and families to determine their eligibility for all subsidized health insurance coverage (including MassHealth, ConnectorCare plans, and Health Connector QHPs with tax credits) as well as for unsubsidized QHPs offered by the Health Connector.
To that end, the Health Connector, in collaboration with the Executive Office of Health and Human Services (EOHHS) and UMass Medical School, has developed a new electronic system for eligibility and enrollment with enhanced functionality including real-time eligibility determination. Specifically, the Health Connector will have the capability to assess an individual or a family’s MAGI, as required by the ACA, which will be used to determine eligibility for certain MassHealth populations and for ConnectorCare plans and QHPs with tax credits.

“Has developed?” As if it is done? Ridiculous! The Integrated Eligibility System — that figures out your eligibility for many kinds of programs — is the hardest part of the entire project, has had nothing but problems, isn’t in scope for the October 1 launch at this point, doesn’t work now, and doesn’t work in future releases. Yet the public is told it has been accomplished.

September 12, 2013

Connector Board Meeting

The Board gets a presentation on how the Affordable Care Act implementation is going for October 1.

The presentation, which you can view here, was created by three directors at the Health Connector: Roni Mansur, Ashley Hague, and Jen Bullock. As for key priorities before the launch that is imminent, you get:

Complete IT work to support individual eligibility, shopping and enrollment for the first phase of our updated system

This is a red flag, if you’re a tech guy. This work can’t be an action item 19 days before launch. It should have been finished weeks ago and be in testing. This hints that these people really know what is going on, rather than the theory that they simply haven’t been told.

“Project Status” is slide 6 and is worth looking at:

This slide is a combination of truth and lies. First, there is no user acceptance testing happening now and it does not happen before the launch. Also, performance testing never happens. (We know these things from the BerryDunn audit reports at this time.)

The second main bullet is the fascinating part. It says that there will be a code freeze within the next seven days. (A code freeze is when you stop writing new code so you can test the existing code for reliability and not worry that the new code you are also writing is messing up the system. Every project has a freeze at some safe distance from a release. This is not a safe distance.)

That the presenters know of a late-in-the-game code freeze means they know the status of the project from those working on it. They aren’t out of the loop.

The idea of a “code freeze,” just days before such a major and important launch, is also a red flag. New coding should have stopped weeks ago. They should be testing and fixing problems. But the truth is, CGI is writing lots of new code as they are desperately trying to get things working. This is another example of the need for technology experience at the board level. Someone needed to say, “What do you mean they are still writing code? Are they crazy?” In fact, CGI’s initial schedule from the contract documents in late 2012 showed that new coding should have stopped weeks ago.

The presentation ends with this on the last slide:

Our highest priority, through our IT, operational and outreach efforts, will be providing a strong and reliable member experience, which will help maximize our success in transitioning members and preserving coverage for the Commonwealth


September 13–30, 2013

Final “Go/No-Go” Decision(s)

The audit reports list September 13 as the final date to decide whether or not to release the system on October 1. However, that report says that CGI has not provided the proof that they are ready. Here is the language:

The Commonwealth did not have the advantage of exit criteria from user acceptance testing to make the final go/no-go decision for MA HIX/IES Release 1.0. Without formal approval of all schedule changes by the Commonwealth, this process is likely to continue for Release 1.1 and impact the quality of the deployed solution.

Everyone else, here on Planet Earth, decides whether or not to approve a launch according to user testing showing the system works well. Here, the lack of user testing “resulted in the Commonwealth relying heavily on CGI’s component testing results to make a go/no-go decision for Release 1.0 for October 1, 2013.”

So, because we think the parts of the car all worked in the lab, yet no one has driven the car that uses the parts, we are going to assume that the driver is going to be OK. Even though so much about constructing this car has gone badly. Absurd.

One has to wonder what would have led to a “no-go” decision.

The Real “Go Live” Decision

While September 13 was officially the decision date to decide whether or not to deploy on October 1, projects in serious trouble don’t really make this call until the last possible minute. Whatever the committee decided two weeks ago, there was another decision that was made the day before the launch that was the real “go/no-go” call.

But someone decided, that despite open enrollment lasting for months, the exchange could not be delayed until it was fixed, but it had to launch for October 1. Even though they knew people couldn’t apply for subsidized insurance policies. (The audit reports show that they believed they were going to do at least two more releases, in late October and November. They never happened. Of course.)

Why did they release it? I think that they saw that some features worked, such as anonymous browsing, and submitting an application for the non-subsidized population. They thought that they could “make the date” and then add those features afterwards. Let’s put aside for a moment our knowledge that these future releases in October and November don’t happen. The real problem is that the overall stability of the system is poor, there has been no proper testing, and they don’t have the tools in place to know precisely how well things are working or what the problems are, after they go live.

OCTOBER 1: 8:27 a.m.


No surprise to anyone who has been reading the timeline up until this point.

The theme for the rest of 2013 should be called, “And the band played on…” in reference to the famous quote about the live music continuing on deck while the Titanic was sinking.

You will see lots of communications and statements from everyone involved — even Governor Patrick — that things are fine and everyone should keep going to the exchange that does not work properly, and never will. How they could have maintained this public posture while they had such damning internal evidence that this would not work — is something to think about.

Fourth Decision Point

In the Decision Point section, I described all that should have happened when the launch blew up. They should have ended their relationship with CGI, gotten a new firm to help, and focused on the consequences of a paper-based system for tens of thousands of people.

Yang and Gruber appear on the PBS Newshour at the Launch

Connector Executive Director Jean Yang and Board Member Jonathan Gruber appear on the PBS Newshour on October 1 to celebrate the release. They are asked about policy and also about the exchange. Yang lies about the exchange: “The experience you get is not going to be much different than what you typically get when you go to an e-commerce website where consumers can easily browse, compare options, and complete a transaction.”

Gruber is not at the Connector like Yang, but is at his home. He shows a few pages of the exchange (he’s doing anonymous browsing, which even I saw sometimes worked if no one else was using the system in the middle of the night). He talks positively about everything, and the shows the reporter his tropical bird collection. It’s very serene.

It’s amazing that while they are telling everyone to go to the broken site, and indicating it will be something used “easily,” there are thousands of people failing to get their insurance. What were Yang and Gruber thinking?

Back at ground zero, the Health Connector people, who should be able to see in their application logs that no one is successfully signing up for subsidized insurance (and little of anything else), hear of problems over the phones, and tell people the traffic is the problem.

Now, it’s possible the system was so incomplete that the non-technical people didn’t even know people weren’t completing any applications for subsidized insurance.

In fact, we find out at the February 13 Connector Board meeting from Sarah Iselin (page 3 of this document) that Optum is going to put in the missing monitoring tools for our exchange just like they did for the federal exchange. (Also, we see from the October 10 Connector Board Meeting minutes that they don’t know how many of the completed applications are from Commonwealth Care members. Another red flag.)

Even without tools, CGI’s technical people could have easily looked into the database and found out no one was completing applications for subsidized insurance, which is the bulk of the target user population. (I know they did. All of us who launch projects always check what’s happening under the hood so we know what is going on. I have the feeling that a message like, “No one is successfully getting subsidized coverage and the logs are full of tens of thousands of error messages” would travel up the hierarchy pretty fast.)

In addition, we know from the MITRE report in February that the call center and staff of the Health Connector and MassHealth were “overloaded” by people who couldn’t get through.

Perhaps the technical people didn’t report what they were seeing. But the call center people knew, and they surely reported what they were hearing about from so many callers, as that’s part of their job.

October 4, 2013

Executive Director Yang appears on a television news program, “Broadside: The News with Jim Braude,” a program where a savvy political analyst (Braude) asks questions of a guest. He interviews Yang for four minutes.

Jean Yang, who was inside the call center for the PBS NewsHour special, is certainly hearing from them that people can’t get subsidized coverage, which is what most of them need.

Braude, who must have visited front door of the site (or didn’t go at all) says, “I went to your website, and it’s terrific.” Yang smiles and says: “Thank you. We do believe we offer people a process that’s easy to navigate,” and she smiles again.

“Easy to navigate?” Is this why the phones are ringing off the hook at the Health Connector? Jean Yang lied to Massachusetts on television. (And how would she know what people would find easy? There was no user testing!)

At this time, Anna and I have had no luck getting her new insurance. We called the customer service line twice since the launch and they told us to try again a couple of days later when the traffic was lower.

I got up before sunrise the next morning to try it before the “traffic” was there. It was fast enough, and it didn’t work. I did a surface analysis about how the pages were set up. I was disturbed. Everything was amateurish and there were some fatal flaws that were easy to see, but were right there in front of me. I was stunned, and knew that serious problems lay beneath.

October 10

Connector Board Meeting

This is the first time the board meets after the launch. The minutes of that meeting are informative.

Executive Director Jean Yang gives her report and “thanked everyone for contributing to the success of October 1 as the first day of open enrollment and stated that the journey is far from complete.”

A report is given showing traffic, page views, and visitors. They admit there are problems, but that these are expected with a large rollout. (Um…no, that’s not what’s expected.)

They say that there have been 1,704 applications submitted. Crucially, when a board member asks how many are from Commonwealth Care, the presenters say that they don’t know. That’s important because these could all be unsubsidized applications, leaving out the bulk of people who need the exchange. But what is more important is that the eligibility and plan selection tools that the Commonwealth Care people need are not in place at this time. The audit report confirms that they were not working for this release. Do the presenters not know this? How could this not be mentioned?

Mid-October, 2013

It is clear the federal exchange at doesn’t have a problem with “glitches” or “traffic” but that CGI did a poor job and it is broken. At this time, President Obama orders a “tech surge” to bring in great minds from the technology world to assess, overhaul, and get the federal exchange working.

Once I saw that CGI had botched the federal site, and that other states were having problems also, I began to believe the rot went deep inside our state’s exchange. Since things on our exchange didn’t improve in two weeks (I visited the site twice a day), I knew things were very bad. Every troubled launch I ever saw always had lots of fixes in the first few days. But nothing? Oh my God. I began to believe that it couldn’t be fixed at all.

October 25

Finally, Himmelstein writes to CGI, says they have not delivered, and asks for a mediation with them. He clearly has an understanding of the severity of the problem that is invisible in the board meeting two weeks earlier.

He then writes a memo to the Connector Board saying “extensive costs” have been incurred for working around the problem (probably call center staff) and they are worried about harm to the state’s reputation. Senior leadership at the Connector know about the issues. Page 5 of the October 2013 monthly audit report confirms this:

The Commonwealth project leaders and executives made effective decisions regarding what functionality should be placed into live operations and what functionality required further development, testing, and training efforts. Contingency and mitigation plans were updated to address the functionality made available to Commonwealth users and stakeholders during the month of October. These efforts are expected to continue during the month of November.
The Commonwealth and CGI have begun to monitor and report on operations and maintenance (O&M) statistics

They — Commonwealth project leaders and executives — know what is in the exchange, and what is not. They are also looking at the operations and maintenance statistics. Also in October, Himmelstein says, in writing, that CGI has failed. So when Governor Patrick says months later that they don’t know how bad things are until late November, this is utterly false. (Never mind that by the end of October, the call centers at the Health Connector and MassHealth are logging an enormous number of customer situations that are serious. Is this spike in volume of problems not being reported through either Yang and Shor or Thorn and Polanowicz to Governor Patrick? I mean, this is people’s health insurance, right?)

October 30, 2013

President Obama visits Boston to Celebrate with Governor Patrick and Connector Leadership

President Obama came come to Boston on October 30 to celebrate the Affordable Care Act in the place it all began: Boston. Governor Patrick and the leadership of The Health Connector and MassHealth were all there at his speech to celebrate with him. They posted several photos of the event online and were very proud.

As I said elsewhere, the federal “tech surge” was already two weeks underway to fix the federal site. Our people should have talked to the White House staff about this effort and started a similar initiative here.

November 2013

Communications Breakdown

At this point, after a terrible month for Anna, I began to notice that the people doing communications for the Health Connector were living in a different world: one where the launch went well.

At pharmacies all over the state, there were these little brochures at the pickup counter, advertising the benefits of the “new and improved health insurance marketplace” and telling everyone to go to the exchange. They advertise features coming soon like online chat, a message center, and checking your balance by phone. (Those features were some of the first ones cut as the exchange project fell behind.) These brochures would remain at pharmacies permanently.

In social media, it’s all fun and games. No mention of the problems. Nothing about their overheated call centers. Just lots of photos of smiling people happy to have coverage, and silly public service messages like this one, telling people “not to underestimate the power of floss.”

Worst of all, at around this time the Health Connector puts out their 2013annual report to the public at this time, which celebrates the launch and advertises advanced new features to come in the next year, beyond the improved exchange there now. The report isn’t dated, but Jean Yang tells the Connector Board it is complete at the November 14 Board Meeting. The fact that the 2013 Annual Report is titled, “Building on Our Success, Preparing for the Future” is really all you need to know about the disconnect between this organization and the citizens outside of it.

It’s insanity. The Health Connector is not an enormous organization. How can everyone there not know the exchange is in trouble?

November 14, 2013

Governor Patrick tells reporters, “The site gets better every day.” But this isn’t true. The audit reports from the end of October and the end of November show that things are getting worse. I know because I kept trying it and there are now some news reports that people continue to fail and that wait times on the customer service lines have increased to a level where one-third of people are giving up. Also, the Connector Board meeting held on this day finally has the members discussing the problems.

Connector Board Meeting

There is another slide presentation, but this time Scott Devonshire’s name is on it. He’s the Chief Information Officer at the Health Connector, and he discusses the problems. The first slide is: “Where We Are Now: IT Functionality.”

A bullet on the end of page 3 says:

Real-time eligibility determination and complete end-to-end shopping and plan selection for subsidized products to be phased in beginning later this month and onwards through December.

This is quite serious. It is an admission that everyone who needs subsidized insurance, which is most visitors, has not been able to do that.

Problems are discussed, as we see in the minutes. Devonshire talks about the exchange. He points out that some of this is due to problems with the federal data hub (true enough). He says that the exchange problems have had “a significant impact on individuals coming to the website.” He says the Health Connector is frustrated with these problems, and that they are working around the clock to fix them.

Even though the Connector Board is finally hearing about problems, they are minimized and, of course, more functionality is on the way. But the last BerryDunn report, dated 10 days earlier, had crucial information that wasn’t in this presentation or the minutes of this meeting.

That audit report said that the second release, which was supposed to happen October 29 got moved to November 17, and even that was in doubt. The presentation given at this meeting talks about needed features being rolled out in December. But the audit report points out continuing problems CGI is having fixing things and making deadlines. Not only that, the exchange has numerous grave problems in overall stability. Let’s do the bird’s-eye view trick again for the 20 pages of this audit report where current issues are listed. Every bright-red box is a critical problem:

There are seventeen different critical problems in the system at this time. In a live system!

How does the Health Connector CIO, who is qualified to read this report, not know about what is in it? This report makes it clear that the rollout he is telling the board about is not going to happen.

It’s around this time Anna and I are getting increasingly nervous. The website still doesn’t work, despite regular attempts. We filed a paper application. One of the customer service representatives told us on the phone that she was going to be in something called, ‘Neighborhood Commonwealth Choice Plus.” or something like that. They said it would be similar to what she had now. At this point, all we really cared about was whether the providers she used were going to take whatever this insurance was. Anna called her providers. MGH said they were pretty sure they were going to take what she had. Children’s Hospital didn’t know. Two other small providers had no information they could give us. We called the Connector back, and they said that everything would be sorted out by Christmas, and Anna would get a letter in the mail at that time with more information, clearing things up.

November 23, 2013

According to figures published in the Boston Herald, “only 23,894 applications had been completed — many by bypassing the site entirely and signing up over the phone or by paper application. And just 1,062 applicants have successfully enrolled.” (The 1,062 are almost certainly unsubsidized applicants and subsidized applicants who applied outside the exchange.)

November 26, 2013

The Health Connector finally begins telling the public what it has been telling people on the customer service lines: download a form and apply by mail.

The strange tone, “Did you know…?” sounds like this is just another fun way to do this. But this is now the only way you can apply if you qualify for a subsidy, and far worse, the Health Connector has no practical way to process any significant number of these paper applications. (We would find out in a few months that there are 50,000 unprocessed paper applications in a stack and they don’t know what to do with them.)

Early December 2013

Decision Point Five

The federal exchange “tech surge” has worked and stabilized the basic functions that users need. Monitoring and release tools have also given them the ability to manage the ongoing minor issues effectively. This approach, a surge and better tools, should have been adopted by the exchange here.

Mid-December 2013

For all the people in subsidized plans, roughly 250,000, they get put into one of two buckets. Above a certain income level, they keep their plan for a while. This was about 124,000 people. Below a certain level, they get put into MassHealth, due to a federal expansion of Medicaid. This was about 130,00 people.

December 18, 2013

The Connector is not just suggesting that people apply by mail, they now say that this is the preferred way. But their main website still tells people to go to the exchange for insurance.

December 19, 2013

This is the first time online that the Health Connector acknowledges problems with the website for the previous 11 weeks.

Late December 2013

Much happens in late December. People on subsidized plans get physical letters saying they will be covered (one way or another) until March 31. The federal data services that the state exchanges talk to are now in good condition, making it impossible for the state exchanges to blame problems on the federal end from this point forward.

Anna gets a letter saying she has been put into MassHealth, even though her income is above the cutoff level. She was scared about this, as she liked the program she was in, and MassHealth had a smaller network of the kinds of providers she needed. She called her providers again. Some said things would be OK, some didn’t know.

December 26, 2013

WBUR writes a story today to help struggling people get answers on what to do. They even ask Jason Lefferts, the spokesman for the Health Connector, how people should begin an application. He admits that their customer service reps are fielding 9,000 calls a day and yet, Lefferts tells people that the exchange is still the best option?

December 30, 2013

According to reporting by the Republican, as of today, only 497 people have gotten new, subsidized insurance plans through the Health Connector.(But not necessarily though the exchange.)

So, of the 250,000 people who need subsidized insurance, only 0.01988% have succeeded by the original enrollment deadline.

58,864 have completed applications, but those applications were unprocessed. (Anna’s was one of those applications in the pile.)

Also, we hear that the Health Connector customer service lines are so overloaded that a very high number of callers are dropping off before talking to anyone. (This means that the internal call logs at the Health Connector aren’t even a reliable guide to the severity of the problems being caused.)

January 2014

Decision Point Six

This month is when Governor Patrick decides that it’s time to do things differently. What should they have done at this point? As I said above in the Decision Point section: CGI should have been dismissed, a replacement vendor should have been brought in now, and not in May (when hCentive returns).

January 6, 2014

The Health Connector Board asked the federal government for a company who could do an audit of the exchange infrastructure. CMS offers to have MITRE, a private company, do this. (I know MITRE. They are good.) This audit begins today. According to that report, MITRE finished their review on January 10. Why the final version wasn’t delivered until February 4 is a mystery. That wastes weeks of crucial time. (From my experience, it is likely that there was great debate about the content, and several versions of the report went back and forth between MITRE and the Health Connector. But the report is so damning, the details were not important.)

January 8, 2014

Governor Patrick and Attorney General Coakley, in different places, are asked today whether the state should consider suing CGI. Patrick says he has made inquiries about this. Coakley says she would if “there’s an ability or a need to do that.” She then says the government is working to fix the site, and “I know that process is under way, and we can and will stay involved in that.”

Coakley’s statement, along with another sentence or two afterwards, shows that she isn’t following what is going on. If Himmelstein believes on October 25 that CGI hasn’t delivered on the contract, and writes them asking for legal mediation, why are we in January without a detailed statement from our Attorney General on the contract and this massive failure?

January 9, 2014

Connector Board Meeting

The Republican has the story about this key meeting. People at the Board meeting say they have commissioned an independent firm to assess what’s wrong. (This is MITRE, which began its work January 3.) Glen Shor, who is the Health Connector Board Chairman and a member of Governor Patrick’s cabinet, is there, and says, “It will inform us about CGI’s continuing roles and responsibilities and, among other things, it will help inform us about the level of damage done and the appropriate approach to accountability.”

Connector Board Member Jonathan Gruber wants to know if the report is going to look at the failures that he, and others in charge of the project, were responsible for: “The review presumably has to review not just CGI’s fault but the institutional fault in the Connector for not realizing this went wrong.”

Yang answers, and says the review is technical only.

A lot of the information they say they need from the MITRE report is already available to them through existing audits and reports. Later on, Jean Yang would admit that the features that were supposed to be delivered by CGI in late October were never delivered. It is clear that CGI must be fired, and that could have happened now, and not a few months from now. Jonathan Gruber — asking for him and his fellow board members to be reviewed — was amusing. Why not write his own report? He has all the information he needs already. His social network through MIT provides many people who could easily analyze what happened.

And it’s clear that Yang is not interested in what the Health Connector did wrong. She never would be, as she will prove throughout the winter in written testimony and media appearances, which you will see later in this timeline.

January 25, 2014

According to a story in the Boston Herald, Health Connector Board member Jonathan Gruber writes a memo to the rest of the board that says that while CGI is responsible, the Connector Board has been “disinterested” in the exchange as a technical project. He admits that he wasn’t listening at board meetings and was using his iPhone instead. He says that too little has been shown to the public in board meetings. He goes on:

“This has been great for projecting a harmonious image for the board, but it has, I think, contributed to our falling down on our jobs,” he wrote. “I think we need to stop being afraid to surface hard issues and disagreements in public at board meetings.’’

Gruber also asks that the Connector Board start getting reports on how the fixes to the exchange are going. That the Connector Board has not been getting updates on what is getting fixed is alarming.

January 28, 2014

In a Boston Globe health policy blog post, John McDonough, a former Senate aide to Ted Kennedy who worked on both the Massachusetts and federal health reform laws, rips into Governor Patrick:

If there has not been a conspiracy of silence in leveling with the public about what’s gone wrong and how it will be fixed, our responsible state officials, including Governor Patrick, are all doing a fine imitation of one.
This is unacceptable. I can’t name a single Massachusetts official who is leveling with the public on this national embarrassment. From what I hear, second and third hand, four months into this disaster, the Patrick Administration has no plan.

When Governor Patrick was asked about McDonough’s comments, he said that McDonough was in no position to know what was going on, in this Lowell Sun article:

Gov. Deval Patrick on Thursday fired back, calling McDonough a “great guy” but suggesting he was in no position to assess the administration’s response. “He’s a long way away to be able to tell anything,” Patrick told the News Service.

But it gets worse. Patrick says that the exchange isn’t the issue here, but the policy and coverage:

“The folks who are worried about the website, yeah, it’s concerning, but the website is not the main event and I’ve said that over and over again.”

But this isn’t true. The “website” is the main event. The exchange was the only way people could continue to get health care. The exchange is why the federal government gave us millions of dollars. The idea that the government is going to go house-to-house to sign people up “ with a clipboard” as Patrick says in the same breadth — is ludicrous. Especially as there is no feasible way at this time to even process the paper applications on those clipboards.

This is yet another example of senior government officials loving the policy, but not caring about the implementation.

February 4, 2014

The much-awaited MITRE report — a technical audit — is delivered to the Connector Board. It is clear from press coverage that Health Connector Executive Director Jean Yang and Governor Patrick have seen it in draft form at least a week before this.

The report is labeled “For Authorized Distribution Only” on every page. Why?

What does the report say? It covers two areas: the state of things at the time, and options for moving forward. I am going to discuss it in detail. (I have been assigned to write and contribute to audit reports like this about half a dozen times in my career.)

As for what is wrong, the diagnosis is grim, as anyone could guess from the BerryDunn audit reports. Everything is wrong, and wrong in ways that are simply not allowed to happen on IT projects. Here are some highlights with my thoughts:

The system is so poorly constructed that it is corrupting information about people and their policies. This means that even if the system gets fixed, people’s information will still be wrong and there will be problems with their insurance. (In software development, data integrity is so important that it is unthinkable to launch something that corrupts the data.)

There is no technical person at CGI making sure all the components of the system work properly together. I have played this role on large projects. It’s crucial. When you have someone build you a house, someone has to coordinate the plumbers, carpenters, masons, and roofers to make sure it all works out. That’s what this is. Imagine building a house where everyone just does their thing without coordination.

There is no single “entity” in state government in charge of the technology.

The number of features kept being reduced throughout the project, yet the ever-shrinking set of functionality never was not delivered on time.

Core features that had to be there still do not work. Things that had to work for going live still do not work as of January 6. That’s right — basic things like creating an account or searching for a MassHealth provider — never worked. This is like Amazon launching without the ability to log in or searching for what to buy. It is unthinkable.

The government wasted too much time in the beginning of the project figuring out what it wanted when it should have begun the work.(You’re going to have to trust my translation of MITRE saying, ‘The requirements development process took more time than was reasonable.”)

Nobody ever had actual people try using the system before it launched.(Again, translating from “There has been a lack of credible user testing.”) I knew this the second time I visited the exchange site when I came across a page error that would have never have been missed by any tester.

They knew one of the most important features of the exchange did not work, a showstopper in fact, but they didn’t change their plans to fix it, and instead kept working on other things. (Translating from, “Problems with eligibility determination were identified but not raised during the development life cycle.” What they meant was that the coders weren’t even sure how to solve the problem properly.)

Much of the technical work is of unacceptable quality. It would be too hard to explain to non-programmers how serious the problems are.

CGI didn’t have senior enough people available in many areas and couldn’t handle the work. No surprise here. This is why I wouldn’t have hired CGI. They were working on too many exchanges and there was a great risk they would be spread thin and couldn’t get enough good people on any one of their projects.

MITRE advises how to move forward:

  1. Get one person in charge of this. (Governor Patrick said he had a draft of this in mid-January, and went off to find Sarah Iselin.)
  2. Pick one of three options: fire CGI and start from scratch with new vendor, fire CGI and improve what is there now, bring in new vendor to work with CGI.

So what does the Governor decide? (He is involved now, and by picking Sarah Iselin as his assistant, all decisions are going through his office.)

First, let’s get his take on the report: In this WBUR story, Patrick says he disagrees with MITRE’s finding that the governance structure didn’t work. He said everything would have been fine if CGI had done it’s job.

All the evidence is contrary to what Patrick says. Putting aside the issue that a governance process is supposed to be able to handle a failing vendor, the BerryDunn audit reports, along with other testimony in newspaper reports show that the governance structure was a severe problem right from the start. In fact, the governance model is why the project started late, and differing visions within the fractured leadership team were the reason why CGI couldn’t get things done for the first few months. For example, the first monthly audit report says:

The current Deliverable Review/Approval cycle is long and, in practice, does not support project objectives. (High impact)

That’s a diplomatic way of saying, “If the government takes this long to review and approve everything, we are never going to finish.”

Patrick then says, “It turns out that this vendor has required and will require a much, much shorter leash. And that’s hard to do by committee.” But this vendor needed a shorter leash a full year before he says this.

The governor goes further and says that the exchange is “a largely fixable system.” There is no basis for this type of statement, other than to mislead people about how serious the problems are. Because the MITRE report says the opposite of that. The audit reports in the end of 2013 show that the number of defects is increasing, and that serious structural problems have not been addressed. For the record, I am not saying Governor Patrick came to this conclusion. Someone else told him this. But the governor should have said to that person, “But it blew up four months ago. If it is fixable, why hasn’t it been fixed by now? The federal exchange was fixed in two months. The records say we have 350 people working on this. If that many people can’t fix it in four months, how can it be fixable?”

Which of the three options does the government choose?

Strangely, it doesn’t pick just one, it does all three: first we get option three: bring in Optum, who fixed the federal site and has the expertise. Then option two: we decide to fire CGI. Lastly, when CGI’s code can’t be fixed after weeks of more work, we do option one with a new firm, hCentive, which has working components from other successful exchanges. (Strangely, hCentive was one of the original authorized sub-contractors on the project.)

We should have just done option one at the end of January, rather than waiting until May to do it, wasting valuable time and millions more. The MITRE report is so scary, it is clear that neither the vendor, nor the codebase, could have been redeemed. But it isn’t until March 17 that the government says that they are going to part ways with CGI, a process that takes weeks.

February 6, 2014

Governor Patrick holds an event to appoint someone to take over the troubled exchange: Sarah Iselin. Her title is, “special assistant to the governor for project delivery.” Iselin is a well-respected leader in the insurance industry, and takes four months off from Blue Cross Blue Shield Massachusetts to take this post. She has a staff of two people, neither of whom have technology experience, either.

Patrick also removes UMass Medical from project and contract oversight at this time.

Governor Patrick says that one of the problems with CGI is the issue of intellectual property. There is no basis for this statement as a primary issue in the contract, as the language there made sure that there would not be a intellectual property problem upon termination.

Governor Patrick invokes an emergency provision in procurement law to give no-bid contract to Optum to help with the exchange and figure out how to process all the paper applications. It is clear any fix is far off. They will be paid $9.7 million for the first 30 days of work. This comes out to an eye-popping $326,000 per day.

While we are here, is Sarah Iselin the right person to take over? Of course not. This program was severely compromised by the fact that the people in charge of it were concerned about policy and not about technology. (Connector Board member Jonathan Gruber says this explicitly in a board memo in January.) Sarah Iselin is another insurance policy person with no technology experience. (There are plenty of people in the insurance industry who manage IT projects as their business is almost entirely software-driven.)

At the February 12 legislative hearing on Beacon Hill where she testified, Iselin was asked about her lack of technology experience for this project. Her answer was that they were hiring a new firm, Optum, to handle the tech side. Once again, the government outsourced their technology judgment. Considering we were about to start paying Optum $326,000 per day from that day forward, I am confident they were going to tell Ms. Iselin whatever she wanted to hear.

(It is important to note here that this was advertised as a 30-day deal. The good news, later on, is that Optum will come in under budget for this month at $5.3 million. But the bad news is that they will stay on in March for a projected additional $11.1 million.)

Governor Patrick apologizes

For the first time, we hear a direct apology to the people who were harmed from Governor Patrick. But what is said?

“To the people whose transition has not been smooth, I join my colleagues from the Connector and MassHealth in apologizing to you for the inconvenience, and I want to thank you for your patience”

This mischaracterizes what happened. This spin is contrary to what has been documented for months in customer service call logs about people losing insurance, missing appointments, and paying for things that should have been covered. This is a persistent, deliberate effort to shape the public’s understanding of the harm caused, despite clear evidence at the Health Connector that something far more serious has been happening to people.

Governor Patrick Explains to Globe Why He Didn’t Act Sooner

The Boston Globe reports that Governor Patrick says he didn’t act sooner because the extent of the exchange problems were not “clear until the end of November.”

The Governor is not telling the truth. His cabinet secretary — Glen Shor — was informed of the problems throughout all of 2013. When the site blew up at the October launch, it would have been known in a few days how bad it was from the customer service call logs. It did not take eight weeks after the launch to know what happened. Himmelstein writes CGI on October 25 and tells them they are in breach of their contract. And the BerryDunn audit report at the end of October is frightening.

Sarah Iselin Promises Transparency

Iselin says at the hearing that there will be weekly briefings to keep people informed about what’s going on. But these dashboards are really about policy issues, the application backlog, and customer service issues. Most of what the legislature and the press want aren’t there. I was frustrated to see how limited these reports were. There is no information on the state of the exchange or the costs. (See the dashboard section in the report references for links to all of them.)

A Small Number Succeed with the Exchange

In that same story, the Boston Globe reports that, four months after the launch, a small number of people could submit applications and find out if they were eligible for subsidies.

This is the heart of what the exchange is supposed to do, and it’s news that, four months after the launch, a few people can do it.

February 10, 2014

Harm to real people, lack of confidence in Patrick from Murray

In an interview with the Boston Globe, Senate President Therese Murray says her office has been swamped with calls about the exchange, and that she doesn’t believe Governor Patrick’s assurances about everyone having coverage. She mentions a constituent who couldn’t get a heart transplant because her income was $14 above the legal limit and now has no health insurance.

Murray says, “and she is not the only one.”

This isn’t news to Anna and I. We now know there are many more people like this, who have fallen through the cracks and are not getting care, no matter what Patrick, Shor, and Yang keep saying in public. (See Part Four of this post, “Anna, and All Who Were Harmed” if you haven’t read it.)

February 11, 2014

The Herald runs the infamous story saying the Health Connector, at the beginning, wanted to build the “Rolls Royce” of Health Exchanges. This story got my attention because of the disconnect between the plans and the implementation. In fact, the governor of Connecticut would make fun of us in the spring, saying they had built a “Ford Focus” that worked, while we had tried to build a “Maserati,” and failed.

February 12, 2014

For the first time, the Massachusetts legislature holds a hearing about the failed exchange — 133 days after the it fails. The forum is the Joint Committee on Health Care Financing. Sarah Iselin, in her job for just two weeks, appears, along with Jean Yang. Cabinet secretaries Shor and Polanowicz cannot attend for different personal reasons, upsetting legislators. (Once again, John Polanowicz escapes a public part in this story.)

Legislators from both parties vent, demand answers, but Iselin claims she can’t speak about the past or the behavior of CGI, upsetting the chairman of the committee.

This meeting ends up being useless theater. The answers the committee asks of Iselin never get delivered, something Democrats lament months later when Republicans want legislation demanding them.

In the hearing and on Twitter, Sarah Iselin repeats that she will provide weekly public briefings.

February 13, 2014

Connector Board Meeting

This is the first Connector board meeting after the Iselin appointment and increasing media coverage of the problems. People are paying attention now in a way they were not before. A lot happens.

The most dramatic thing that happens is that Health Connector Executive Director breaks down and cries about all the problems. Yet she accepts no blame. She receives support from a fellow board member, as reported in the Boston Globe:

Dolores L. Mitchell, a Connector board member, thanked Yang. “A shaky voice every now and then sends a powerful message about how much you care,” Mitchell told Yang. “You’re going to get it right. I know you are.”

Ridiculous. This isn’t about caring, but fixing problems. Yang has had more than a year to “get it right.” Her crying in front of the public and the media should have been her last act as a public official. Yet she remains in her role to this day.

It is interesting to note that there were several state-based health exchanges that had major problems on launch. In all those cases, the heads of those exchanges were fired, resigned, or took medical leave. Jean Yang is the only one who still has her position, despite the fact that our exchange was the worst in the country.

Board member Jonathan Gruber said about the exchange, “There’s clearly been a failure of the actual coding of this beast.”

Also at the meeting, the Health Connector drops a bomb: there are 50,000 paper applications that are unprocessed and they have no practical way to enter them into the system. (The number was actually 70,000, according to a reliable source.)

Optum, the new vendor, is going to bring in 300 people to process them, and work on a new way to speed up the process (which, thankfully, they figure out.)

Finally, at this meeting, the board discusses the idea of a “tech surge” as the federal government conducted from mid-October to early December. (It is never carried out, though.)

Sarah Iselin gives a very interesting piece of information we have never seen before: there are 350 developers from CGI on the Health Connector project. This is an astonishingly high number. I am assuming that CGI kept adding people when things got worse. Still, knowing the kind of exchange software that was desired from the initial contract documents, this is more than double the right number. Let’s not forget, this team delivered a lot less than what was even in the original contract, and by this point, these 350 people still haven’t fixed the exchange, and never will.

Microsoft Report Delivered, Hidden

On February 13, outside the board meeting, Microsoft finishes a technical audit and delivers it to the Health Connector. The Connector hides the report for an entire month, releasing it in mid-March.

The report is damaging. There is much in here the public should have known sooner. For example, it is the first time we know about the tests done on the exchange three weeks before launch that showed there was no way it could operate properly.

March 17, 2014

In the Lowell Sun, Matt Murphy of the State House News Service reports that, according to Glen Shor, that the state is spending an extra $10 million per month on people who were supposed to get federal subsidies.

The article also says that Iselin has decided that the government is going to part ways with CGI.

But there is other news this day: Optum appears to finally be getting control of some parts of the situation. They are reducing customer service call times, getting paper applications into the system, and publishing some metrics about progress; a welcome sign.

However, we do not know the state of the exchange infrastructure, or how well the repairs are going, though Sarah Iselin is quoted on Twitter, admitting that a working exchange is far off:

March 20, 2014

Connecticut and Rhode Island say that they are owed millions in grant money that was wasted by the Massachusetts exchange, as they didn’t benefit from the regional grant, as they were supposed to. (It’s hard to disagree with their logic.)

The status of this liability is unknown.

March 27, 2014

A story in the Boston Herald says that some portion of tens of thousands of people are getting free health insurance who don’t qualify because the state is incapable of checking their eligibility.

How is this happening? Lots of people are getting letters because they may be eligible for temporary Medicaid for a few months, but of course, some are not eligible, as people’s situations often change. But the state has no way to easily determine their eligibility and is putting them all into these plans because they can’t risk people losing coverage who qualify. Here is the example from the Herald:

Bob Landry of New England Benefits in Andover told the Herald one of his clients was going to sign up for a Blue Cross Blue Shield family plan that would have cost $865 a month. Then he got the letter from the state offering free insurance. “He had a reaction of, ‘I can’t believe they’re doing this, but I can’t say no to this,’ ” said Landry, also a board member of the Massachusetts Association of Health Underwriters (MassAHU). “He understandably took the benefit, moved on, and canceled the Blue Cross Blue Shield plan.”

And the longer this temporary insurance is extended, the more of this that we will see.

We need to stop here and make something clear: the Health Connector keeps telling state officials that we are going to reimbursed for the people we are putting into care that is subsidized by the government. But every person we put into this care who does not qualify for federal subsidies is someone whose health care we are paying for all by ourselves. We have no idea how large this liability will be.

April 3, 2014

Health Connector Executive Director Jean Yang must testify before the U.S. Congress, as the House Oversight Committee is worried about data security for the state exchanges, in light of the terrible quality of code being deployed. It’s a legitimate concern, especially as the audit reports on the project mention security vulnerabilities.

Her testimony, available at this link, begins with a few pages congratulating herself and her colleagues on their policy achievements in this state. She is asked about the exchange, but she starts with policy pride, as she is accustomed.

When she finally gets to the exchange, at the end of page 3, we get this:

Even so, while implementing the ACA, we have experienced website challenges. These are mainly due to failures of our system integrator. But with our new team in place, we are on a path to go-live with a functional, reliable Exchange website for the next open enrollment period. Challenges with our system integrator and project management shortcomings impeded our progress in achieving our full vision for the website by October 1 of last year, so we decided to deploy only parts of the new system on that date.

Yang accepts no responsibility, it’s just vague “project management shortcomings.” And it is incorrect to say that they “didn’t achieve their full vision.”

They got none of their core features deployed for people who needed subsidized insurance and had a total system meltdown! She takes a few paragraphs to give an update about the paper applications and call times, blames CGI one more time, and concludes by patting herself and her peers on the back, yet again, for their great policy work.

April 15, 2014

Jean Yang appears on a major television program as a guest for the first time since October 4. She is once again on “Broadside.” (Link to video is here.) Now that it’s six months later, the problems are known, and Braude takes a very different approach with her in this 7-minute interview:

Braude: “You knew before the launch that the vendor had problems.” Yang nods in agreement. He asks her how early she knew there were problems. She doesn’t answer, but says there were serious decisions to be made that only the functionality that was ready was to be deployed. She then says that the functionality that was supposed to come after the launch “never came along.”

This is critical information: she was involved in decisions about what was to go live, and CGI never delivered what they were supposed to after October 1. If CGI didn’t deliver anything after the launch, they should have been fired months sooner. Also, with this statement, she can’t claim ignorance of what was going to be released.

Braude asks her whether she reported the problems above her board “that there were problems with CGI to anyone in the Patrick administration.” Yang looks like she is about to answer. Then, in an awkward set of facial movements, (check at 2:08) stops herself from answering. She then recovers and says there were challenges all along and that the administration was involved.

What Braude doesn’t know is that the Connector Board Chairman, Glen Shor, is in Governor Patrick’s cabinet. He could have asked much better questions. But what matters here is that Yang admits that the “administration” was involved with the challenges along the way. This means that when Governor Patrick says the first he knows about the severity of problems is in mid-November, this is another piece of evidence that this wasn’t the case.

Braude gets at accountability: “Are you part of the problem? Is part of the change that needs to happen, you?”

Yang: “I am always happy to look at myself and everyone around me, critically. That’s always a good thing to do. But I would not say that this is an individual issue, at a broader level it is a governance model weakness we have had over this project because this is a multi-agency effort… and this is a committee-driven governance model which is predicated on consensus, and when there is no consensus, there is no authority, there is no effective decision-making. ”

(Yang makes it clear that no one was really in charge.)

Braude: “Did you offer to resign in the middle of this whole thing?”

Yang: “No.” With a smile.

Jim: “There’s got to be a lot more wrong than just the contractor. Is the board part of the problem?”

Yang: “I am sure we have a lot to learn. We are all learning a lot from the experience to make sure this doesn’t happen again.” Yang adds: “The vendor was very responsible for the issue.”

So Jean Yang is not responsible. She smiles at the idea she should have offered to resign. She blames CGI.

April 25, 2014

The Gateway Pundit posts a great breakdown of federal funding for the five worst exchanges, showing that Massachusetts had received $179 million to build the exchange.

They confirm that $35 million was for us to build technology that other New England states were supposed to get from us.

April 30, 2014

Fixing the exchange is not working out. Iselin meets with the federal government to discuss a possible federal takeover.

May 5, 2014

After many more months and millions, the government gives up on fixing our exchange. They decide on a new two-track plan: spend $100 million more using a new system built by a company called hCentive, while preparing to connect with the federal exchange if that doesn’t work out. (hCentive was an authorized sub-contractor on the exchange project, which left in late 2012.)

Governor Patrick invokes emergency procurement measures again and hires

hCentive for a no-bid five-month sprint. They worked on successful Kentucky and Colorado exchanges. They are subcontractor of Optum, which owns a 24% stake in it.

Let me point out that because hCentive has built working exchanges and can reuse that software, and all the fixes that are coming out of those projects, they are the right kind of firm to use here. They would just need to add Massachusetts-specific customizations.

May 8, 2014

Connector Board Meeting

The Board says it will be seeking $121 million in new money from the federal government.

It is also revealed that Maydad Cohen, Governor Patrick’s Deputy Chief of Staff for Policy & Cabinet Affairs, will take over for Sarah Iselin.

Secretary of Finance Glen Shor after the meeting is quoted telling the world that they are “deeply committed” this time.

But this has nothing to do with commitment.

May 23, 2014

The Boston Herald has a story saying that while the Connector is telling people they are setting up a state exchange, they are simultaneously telling the federal government we will join their exchange. That story also confirms we are asking the federal government for an additional $120 million to pay Optum and hCentive.

June 12, 2014

The state announces that Commonwealth Care and temporary programs extended until 12/31/2014. This means they have given up all hope of doing anything with the exchange before next year’s annual enrollment.

June 18, 2014

A story in the Boston Herald and another in the Boston Globe show there is a worry that all the free Medicaid coverage may cost a fortune, and there is no money set aside for it.

From the Herald:

From May to June, the number of people on the temporary insurance jumped from 201,113 to 227,374. The cost to insure those people rocketed from $51.7 million to $90.5 million — and no one knows how high it could go. “There’s no specific budget for temporary Medicaid coverage,” said Secretary of Administration and Finance Glen Shor. “We have not been making projections about it because it is a new program and we are learning about its costs and learning about enrollment.”

June 20, 2014

The settlement deal with CGI is finally signed by both parties. The Boston Globe has the story. The original contract was for $69 million. With all the changes and additional components, they could have received as much as $89 million. The settlement was that they would get $35 million more on top of the $17 million they had received by the previous fall, for a total of $52 million.

All the money for this is from federal funds.

Lastly, the settlement says that Attorney General Coakley has the right to sue CGI afterwards for up to $14 million. There is no further information on whether her office is pursuing this.

July 1, 2014

Republicans in the Massachusetts House get a provision attached to legislation headed for Governor Patrick’s desk that demands a financial accounting of the exchange disaster. Many Democrats support it. Unlike the previous amendment, it doesn’t stop funding for the Health Connector, but merely asks for a report. When asked by the Boston Herald whether he will veto this simple request, Governor Patrick refuses to comment.

July 10, 2014

The state has decided that hCentive will use what they have from other successful state exchanges, and on this date, that firm shows that their components can talk to federal data services without any problems. From a technical point of view, this test is not very impressive, though it is essential.

The other good news from this Boston Globe story is that the state might not need the entire additional $121 million from the federal government. Also, Maydad Cohen, who took over for Sarah Iselin as the Governor’s manager to run the exchange project, says he hopes the federal government will pay the entire bill to fund the new exchange. But there is no information as to whether they will.

July 11, 2014

Though he had earlier refused to say whether he would veto a bipartisan provision to ask for a financial accounting of the exchange, Governor Patrick signs the provision into law, compelling the Executive Office of Health and Human Services (EOHHS) to report the financial impacts of the exchange problems to the legislature by July 31.

July 26, 2014

Heath Connector gives out salary increases

The Boston Herald reports that the Health Connector gives out salary increases to some employees.

(What do I think of this? While others think this is outrageous, in light of the money wasted, these numbers are insignificant. If this money is needed to retain good people, fine. But, you have to wonder what these people were doing throughout the crisis. As I say below, there were some IT people on staff at the Connector. Where were they? Perhaps successors would be worthy of the salary increases.)

Also interesting, the Boston Herald provides a link to a document showing the salaries of the people at the Health Connector, as of July 2014.

Executive Director Jean Yang makes a not-too-shabby $179,243.48.

July 31, 2014

The financial report demanded by the state legislature arrives. (The link to it is here.)

The first few paragraphs, like every other communication they issue, is full of praise for their policy, even though, once again, no one asked them about this.

The fifth paragraph starts with an incorrect characterization of the problems with the exchange:

“The Commonwealth, has, however, encountered challenges with one aspect of its ACA implementation efforts. The ACA required us to create a new online system for accessing individual and family coverage through the Health Connector and MassHealth by interacting with federal databases to validate eligibility and determine the program for which an applicant is qualified. Previously, any Massachusetts resident seeking help paying for health insurance was in a purely paper-based system, applying through a paper application that was manually reviewed, with notices and other documents being mailed to the applicant.”

“One aspect.” The part where needy people get their insurance. (However it is nice they admit the previous exchange, that everyone praised in hindsight, was, counter to popular understanding, paper-based.)

They go on to blame CGI. They say that system performance was “unreliable” and that consumers were “inconvenienced.” The fact is that no one got subsidized insurance from the exchange whatsoever.

Maydad Cohen and his subordinates at the Health Connector have access to thousands of call center reports of people having very serious problems that go way beyond “inconvenience.” For this report to characterize the problems in this way is disappointing and untrue.

They say that those looking for unsubsidized insurance were able to shop and enroll, but they say that performance problems hindered them also. (They don’t explain how many of these people failed in the process.)

They congratulate themselves for taking months to eliminate a paper application backlog that should never have existed, and a paper process that should have been prepared for months before.

Then they get into how they handled everyone’s temporary insurance needs. They say the total amount of money spent at this time on the project is $108 million. It isn’t easy to figure out how much state money has been spent, especially because some of the costs are going to hit other parts of the state budget. But it is certainly in the millions. The cost of the extra customer service representatives alone was $1.1 million.

They conclude with more talk about how great the policy is. Of course.

August 7, 2014

The Connector Board gets a demo of new hCentive solution. It appears to work OK. However, Michael Astrue, a former general counsel to the US Department of Health and Human Services and a former Social Security commissioner, said he had seen the demonstration and found it “rough and clunky.”

In my mind, clunky is OK at this point. They have time to polish it. (This is where they should have been at a year ago.) Also, between now and October 1, hCentive will continue to get the benefit of enhancements applied to the other successful exchanges they created.

August 8, 2014

The Health Connector decides to stick with new hCentive solution they just saw, and abandon all plans to move to the federal exchange. Coverage is in the Boston Globe here about this major decision.

August 15, 2014

Nearly 400,000 people in Massachusetts, many of whom do not know it, must re-apply for insurance in the next open enrollment period, and the government isn’t sure how to reach all these people. The state says that it will cost $15 — $19million to reach all the people who need to sign up. Of course, our state wants the federal government to pick up the tab. Also, in another bit of news, the state says the cost of the exchange, previously $179 million, will now increase to $254 million, passing the quarter-billion mark. They want the federal government (of course) to pick up the additional $80 million they are asking for.

Considering how bad the project went, you have to wonder why the federal government would give us more money.

End of Timeline


What do I think of the next open enrollment period?

Strangely, after all this, I think the exchange will work. Yes, there is a bad data in the databases from the yearlong disaster. This will cause problems, and I hope that the government makes a separate effort to fix that, which will not be easy. But what is happening now — the exchange leadership team, a few months ahead of time, seeing a demonstration that works pretty well — is what software projects are supposed to look like. What is happening now in August, the mirror opposite of last August, is a sign that things will go well.

Will it work perfectly? I doubt that. But we don’t need a Rolls Royce this time around.

Part Six: Why?

This section, which is entirely my opinion, will try to answer the following questions: Why did this happen? Why didn’t the legislature do something? Why didn’t Governor Patrick do something a year earlier? Why did the media have such a hard time telling this story? Why isn’t this all just CGI’s fault? Why have our politicians and pundits said so little about this?

Why did this happen?


The leadership in our government healthcare organizations cares very much about giving people access to health care. They are very proud that Massachusetts provided access to nearly all of its citizens, and so much of their public communication shows this pride. In fact, if you look at their public testimony in this crisis — to everyone who asked for an accounting of the collapse, the beginning of each of those documents (e.g. Yang’s congressional testimony, legislative cost analysis) is always a lengthy recitation of how proud they are of our health care system, even when no one asked them about that. We also see praise of policy in newsletters, and the words and writings of the leaders.

The image of the policy-makers is also important throughout. In the early summer of 2013, when CGI wants further and dangerous reductions in scope, Dr. Himmelstein warns of the consequences of failure, including harm to the “reputation” of state agencies.

Or how about Connector Board member Jonathan Gruber, one of the architects of the Affordable Care Act, who said that the board was more concerned with presenting “a harmonious image” than owning up to the real problems or putting more information out at public meetings, which don’t even mention the serious problems in the months before the launch.

I believe this evidence shows that what mattered to the leaders, every step of the way, was the righteousness of what they were trying to do, and how they looked while doing it, not the quality of what they were doing.

They were so proud of earlier accomplishments, and so confident in the Affordable Care Act and the additional benefits available in it, and so sure this was the right thing, that it was inconceivable that things could go terribly. Even when told that things were not going to go well, it never occurred to them that they needed to stop what they were doing or create a real fallback plan, not an endless series of small contingency plans that were irrelevant if the whole thing failed. They were going to have their big triumphant moment on October 1, once again showcasing their quality to the rest of the nation. It was impossible that this was not going to happen.

However, there were many people in leadership positions — all the way up to Governor Patrick — who should have remembered that this was ultimately about sick people, and not pride in their policy. Someone should have smashed the glass, pulled the emergency brake, and stopped the train in time. (Governor Patrick finally did, in January of 2014. A year too late.)

I am sure some people will look at the noble goals of the people involved and say, “They weren’t being dishonest. They just didn’t know what was wrong or how bad it was. They didn’t have time to correct the reports and printed materials that went out.” But they had the evidence given to them over and over again from the auditor and other reports. And they didn’t tell the public anything along the way.

Lastly on this question, since it is related to the mindset of the leaders of this project, some people have raised the issue that we should have gotten a waiver from the federal law, and that we should have tried harder to obtain one.

If you think that anyone leading our health care system in Massachusetts was interested in getting a waiver from the Affordable Care Act, you know nothing about these people. They were excited about the benefits, and they were — in a way — getting ready for their big celebration with President Obama, as you can see in the photo below. They were not going to RSVP with a “no thanks.”

Just look at this majestic scene in Boston’s legendary Faneuil Hall — when the federal and state government came together to celebrate the Affordable Care Act. Everyone is celebrating and congratulating each other. Both the federal and state exchanges have collapsed. While everyone is smiling and mingling, Anna, along with many others, is worried that she will lose her health care.

This disconnect, between the policymakers and the people jamming the customer service lines during this ceremony, is at the heart of this story.

Lastly, if I had to write a one-sentence explanation for why the failure of the exchange happened, it would be this:

Our government was so proud of its health policies, and so worried about the image of the policy-makers, that it decided nothing else mattered as much: not good government, not sound management, not telling the truth, and not even the people who needed health care.

Why didn’t our legislature do something?

I believe there were three reasons that our state legislature didn’t act.

First, this was probably seen as a federal project funded by the federal government. It didn’t appear that we had any skin in the game. It wasn’t obvious that there could be such a failure that millions in state money would be spent to handle the collateral damage.

Second, like everyone else, there was the improper “it was a website” framing. It probably was not seen as a large, risky undertaking. The vendor was building lots of other state exchanges.

Also, our legislature probably came to the conclusion that we already had an exchange that worked. They may have just seen this as some sort of upgrade. Of course, unlike Anna and I, they hadn’t used it. Had they done so, they would have seen it was mostly a brochure for a paper-based system with links to private insurers. There was nothing about the original site that indicated that the Health Connector could create a much larger, more complicated system. (Never mind the integrated eligibility idea.)

Third, there were the politics of it all. The Democrats who control the legislature are fully behind the idea of universal coverage, and, in general, supported the federal law. They were less inclined to question the value of the project or how it was going.

After the failed launch, many of them heard from their constituents about the problems. But what were the Democrats to do? Republicans all over the country who were screaming about the Affordable Care Act saw the failed exchanges as evidence the law was terrible. The Democrats had no interest in joining that chorus. I am confident that when Governor Patrick told people that things were going to be fixed soon, it was so much easier to go along with that. Also, the legislature didn’t have access to the proof — as Patrick had — that it was absolutely not going to be fixed soon. Or at all.

Even when things didn’t get better after the hearing, and Sarah Iselin doesn’t give the legislators what they asked for in the weeks to come, what are they going to do? The Republicans in the House propose legislation to cut off state funding for the Health Connector until they get some answers about the nature of the problem. A small number of Democrats join them but the bill fails in the House, 110–40 on May 13, long after the severity of the problems are available, even if not understood. In that debate, some Democrats complain that the answers they demanded from the Health Connector in February still haven’t been given to them. Yet they mostly don’t vote for the bill. The Republicans try again with another amendment in late June that asks for an accounting of the costs. This time, there is strong bipartisan support, and it passes.

So why did it take from October 1 to July 1 for the Democrats to finally demand some answers? Some of it was not related to politics, but just ignorance of what had happened. But wanting to support the Affordable Care Act and a Democratic governor had to be a factor.

That all being said, if a legislature can pass a bill to protect women’s privacy in public places in 48 hours; if it can pass a bill to safeguard women’s access to reproductive health clinics in 34 days; then it should not have waited 4 1/2 months to just hold a hearing to find out why so many women like Anna had problems accessing health care.

Especially a hearing that accomplished so little.

Why didn’t Governor Patrick do something sooner?

First, we need to address whether or not Governor Patrick knew of the problems. In other state exchange failures, governors have said they didn’t know. But did Patrick? Let us look back and figure that out.

This was a project managed by the executive branch, and Governor Patrick was in the best position to know what was going on, and force changes. He finally did in late January, when he appointed Sarah Iselin to take over the project. Did he know earlier there were problems? Absolutely.

On January 8, Governor Patrick says about the workarounds that have been happening for the past few months, “And the number of workarounds, while they’ve been a source of my own pride in the creativity of the folks over there at the connector,” showing he has been following this for at least a little while. But we can do much better than that.

The key to understanding what Governor Patrick knew is Glen Shor.

Glen Shor is appointed to Governor Patrick’s cabinet

Shor is one of the most important figures in the Health Connector story.

He is Executive Director of the Health Connector in 2012, when the Exchange project launches, and has lots of problems in the second half of that year. In January of 2013, around the time when CGI admits it can’t deliver, he leaves his post in order to join Governor Patrick’s cabinet as Secretary of Administration and Finance. (Jean Yang becomes Executive Director, John Polanowicz becomes Secretary of Health and Human Services, which oversees Medicaid — a big player in this project.) Crucially, Shor becomes Chairman of the Health Connector Board. This means he goes to all the board meetings and gets all the memos and reports. Does he remain involved? Yes.

According to documents and interviews obtained by the Boston Herald, in late 2012 and early 2013, the friction between The Health Connector were so bad that cabinet secretaries had to “duke it out” to get decisions made. The friction between cabinet secretaries has to involve him and the secretary of Health and Human Services, which, as of January 2013 is John Polanowicz. (No other cabinet secretaries have any possible jurisdiction over this.)

A small diversion on Secretary Polanowicz: He has escaped all public attention in this story, except for the Herald report that two cabinet secretaries are battling over this project. Since he oversees Medicaid and the rest of the government’s health care furniture, Polanowicz certainly cares about the exchange. He takes over during in January 2013 when the project is in trouble and the Connector people and MassHealth can’t get along. Why isn’t he a major player in this story? First, the director of Medicaid, Kristin Thorn, who reports to him, is on the Health Connector Board. Second, the project manager is Janet Baker, who represents the Executive Office of Health and Human Services. Third, since Glen Shor is both the former head of the Connector, and now the Chairman of the Board, and goes to those meetings, and is in the cabinet, there seems to be little reason for Polanowicz to assert himself. After all, Shor, Thorn, and Baker have this covered. Right? That being said, it would be interesting to know what Polanowicz knew and when. (Certainly in late 2013 when an extraordinary amount of people are shoved into temporary Medicaid plans, he is talking about this issue with his Medicaid people. But, as he skips the Feb. 12 legislative hearing, we never know his involvement.)

Back to Governor Patrick: With Secretary Shor in his cabinet, and problems within his cabinet over the exchange project, are we really to believe that Patrick was completely unaware of the problems? Shor also gets the Board memos all throughout 2013, including the one in July saying failure is “likely.” So when Patrick asks Shor how things are going as they are getting ready for launch, which he certainly must ask as all governors running state exchanges do, as the whole nation awaits October 1, what does Shor say?

In April of 2014, when Jean Yang was on NECN and asked if she told anyone above her about problems with the exchange, including the governor, she refused to answer the question. Why did she refuse? She could have said that the Connector Board chairman, with her every step of the way, was in Patrick’s cabinet. The fact that you can see her stop herself from answering is a sign that this is a sensitive issue.

Certainly, by the middle of October, there are emergency financial measures being taken to cover things like a massive increase in customer service representatives answering phones. Those kinds of measures would certainly get the attention of the Secretary of Administration and Finance, Glen Shor. At this time, the meltdown was apparent to all. Governor Patrick could have acted, instead of three months later.

The only other time we see Governor Patrick admitting anything he knew,as reported by MassLive, was when he said in February that it wasn’t until November that the state knew the scope of the problems, as data being put into the exchange was being recorded incorrectly.

Yes, I am sure this broken system was corrupting the data. But the idea that it took eight weeks to know it was broken is not believable. Patrick knew earlier for sure, as I said in other places above. Also, while we don’t know if the September testing report got to Patrick’s desk at that time, when the exchange melts down at launch, how long does it take for someone to say, “Did we test this thing first?” Even if Patrick didn’t know the exchange failed on October 2, I would think that the history of the system would have surfaced and been told to him quickly.

What should he have done?

Governor Patrick should have appointed an overseer in January of 2013, and not a year later. When the big tests fail in September, something that must have gotten the attention of Shor, Patrick should have asked for the audit that happened in January of 2014 and should have insisted on a backup plan for processing paper applications at this point, and not months later.

He should have inquired in late September whether or not the exchange was ready, and upon hearing it was not, he should have permitted a delay.

As he spoke to President Obama at the joint event on October 30, he should have taken the president’s lead on the federal “tech surge” and ordered one to happen here, just as Obama did two weeks earlier.

Lastly, Governor Patrick should have insisted on transparency along the way. The September 2013 testing report, and the Microsoft report from January 2014 should have been made public immediately, instead of deliberately being kept out of sight.

But why didn’t he do anything until January?

This isn’t easy to answer. My best non-political guess is that his people did not provide him with a clear way to handle this. (I have seen this situation in failed IT projects in the private sector.)

But the fact that Jean Yang and Glen Shor certainly didn’t have good options does not mean the Governor couldn’t have appointed an experienced, outside IT manager for the project; couldn’t have told the press, legislature, and public of the problems; and couldn’t have ordered backup preparations for the system failing.

Lastly, I will leave you with this last piece of evidence of Shor’s involvement, just a simple Tweet from the Health Connector — from May 8, 2014:

We, indeed, Secretary Shor.

Why did the media struggle with this story?

This has been a hard type of story to cover. In fact, in several states where the exchanges failed, the media either didn’t cover the story until very late, or didn’t know how to write about it.

For example, the press in Oregon didn’t start covering the story until well after exchange failed, and the Columbia Journalism Review, one of the nation’s most respected journalism publications, wrote an article about how the press handled it. The article’s first two paragraphs are chilling to those of us here following this story:

Behind Oregon’s failed health insurance exchange is a story of spin that in some ways mirrors federal officials’ management of news about the Affordable Care Act from its passage in 2010 to the present day. For the last three years the Oregon Health Authority and Cover Oregon, the state exchange, played a PR game, hiding myriad IT and contract disputes with Oracle, their contractor; relentless bureaucratic backbiting; and bad business decisions. Gov. John Kitzhaber has claimed he didn’t know what was going on until after the exchange opened, while those tasked with building the site ignored internal warnings, shrugged off oversight reports, and misled the public, reassuring themselves and the press that all was well.
We know this now because, when problems became apparent just before the state exchange was supposed to go live on Oct. 1, reporters in the state press corps started to read between the lines of official pronouncements and began their own investigations, which cracked the story wide open. We didn’t know it sooner because, for months before that, both legislative and media watchdogs were slow to question the official spin. There are lessons for journalists in both the strong recent reporting and the weaker early coverage.

I read that and said, “My God. Just like here.”

The newspapers here ran dozens of stories about exchange. But the public never saw the overall narrative in a story that lasted for months. The problems seemed like anecdotes, and the assurances by the government that everything was going to be fine were tough to pierce, because so much of the evidence was either hidden, late in coming, or buried in technical documentation that was hard to understand.

I believe some of the problem was that we didn’t have higher-level pieces about the problem from major media institutions. For instance, if the Boston Globe, in addition to more than 20 individual articles on the exchange, had a timeline piece showing what was known when, that might have provided a useful framework for the story.

Is there a partisan element at work in the coverage? There are correlations. The Boston Herald, widely seen as having a conservative slant, ran a lot of stories about the exchange failure also, and they tended to use more critical language in headline and story content. They also had some opinion pieces critical of the implementation of the exchange. The Boston Globe, widely seen as liberal, did not write a single opinion piece of any type throughout the entire story. This is quite surprising as it is the only major story in the past year that escaped the notice of all their editorial writers and columnists.

Lastly, I think a big problem in covering this was framing. As it was in other states, the story was framed as a problematic website. It should have been framed as people having problems with health care coverage, as I showed with the examples of people who missed appointments or had to pay out of pocket. After all, when the Supreme Court struck down our buffer zones at abortion clinics, that situation was successfully framed here as a health care access issue more than a free speech case. The media and legislators reacted quickly.

Why isn’t this just CGI’s fault?

The Health Connector, the Attorney General, and Governor Patrick have been quite clear: this was almost entirely the fault of the vendor, with perhaps a little blame for some vague “governance model” problems — a description meant to absolve any specific person who was involved in that governance.

Was it almost entirely CGI’s fault? After all, they were incompetent in their own management, they made promises they couldn’t keep, they provided technical people who were unqualified for the work, and they fought to wring millions out of the taxpayers that they did not deserve.

But there are many mitigations. First, we can see from the BerryDunn audit reports that there were serious problems on the government side of this. The problems there are much worse than things I typically see in my own professional work. CGI has to have separate meetings with the Health Connector and MassHealth? Wow. That’s cancerous.

Second, as I said above, CGI should never have been hired. Too many eggs in one basket, and with so many other state exchange projects, they were spread to thin for them to bring in better people when things were going badly.

Third, there is something a previous boss told me long ago: “Our number one priority is to keep our promises to a client. Our number two priority is that when we cannot keep our promises, that we tell the client, and that we give them enough time to make up for our failures. I have never lost a client for telling them early on that we can’t deliver on a project. I have lost a client because we told them too late.”

CGI told the government in January of 2013 that they could not make the deadline with what they agreed to. Nine months early. Then they told them again in April that they couldn’t deliver what they promised in early February. The Microsoft report about the project says, “Every critical milestone, to date, has been missed.” The monthly audit reports show that CGI continued to fail and fail and fail all the way until the launch. At some point the blame shifts to the government for not getting rid of them or getting outside help.

I believe that the government bears the majority of the blame. After all, a Health Connector spokesman said about CGI: “we kept a very close watch on their work, including taking a number of proactive project management steps.” So it wasn’t that they were unaware of what was going on. (And the audit reports are damning on this.)

I will conclude with this: there are two other people who thought the Health Connector should not have just blamed the vendor, but themselves as well. One was current Health Connector Board member, Jonathan Gruber, who said in a memo in January 2014: “While the blame for the current problems is being place(d) on CGI and government staff, part of the blame should be falling on us the board” Or how about Doctor John Kingsdale, who said about failing state health exchanges in the New England Journal of Medicine, “Yet true to script, state and federal exchanges are blaming their vendors, as though they had nothing to do with selecting and managing them, and switching vendors after the damage is done.” We should listen to him. After all, he was the founding Executive Director of the Massachusetts Health Connector, which he led from 2007–2010.

Why have our political leaders and pundits had so little to say about this?

The problem that our political leaders and commentators had with this story is a mix of things I have already mentioned. They only saw scattered coverage. They didn’t understand the nature of the project. They didn’t see a lot of the harm being caused. It was federal money. It wasn’t clear what to do.

Pundits just didn’t have the information. For example, Jim Braude gets to interview Health Connector Executive Director Jean Yang on October 4. He doesn’t have the history of the problems. He doesn’t know the exchange has failed. Since he’s a pro, imagine how that interview would have gone if he did.

The other issue is the political climate around the Affordable Care Act. From October 1 forward, the media is full of Republicans attacking the new health care law. To Democrats, they appear to be opportunists who would say anything to impugn the Affordable Care Act. Even the ones here don’t have the facts that would make this specific to Massachusetts. Their words are unfortunately dismissed as partisan when there is genuine scandal here in the implementation.

Also, the timing matters. By the time the media realizes how bad things are, which is in late January, everyone is in temporary insurance coverage and the frustration is over for many. Everyone appears to be covered, so there doesn’t seem to be a compelling story here anymore, other than the astronomical cost involving mostly-federal dollars. The press keeps writing stories in the spring and summer, but people were losing interest.

Part Seven: Unanswered Questions

Here are questions, that, if answered, would be a great help in knowing what happened, and what issues remain:

  1. Why was the UMass Medical School allowed to continue to manage the project after the state had the majority of the funding in late 2012? Wasn’t it obvious that the state had more to lose than UMass did?
  2. If CGI was having problems and was late on everything from an early stage, why did we make payments for so long?
  3. If CGI agreed to a contract amendment in late January because they were so far behind, why wasn’t that the opportunity to make bigger changes to protect taxpayers?
  4. On July 1, there is a Board Memo saying that it is “likely” that even the reduced features won’t be deployed. The entire board gets this memo, including Governor Patrick’s cabinet member and Board Chairman Glen Shor. This is now clearly an emergency. Why wasn’t the public informed? Why weren’t plans changed? Why wasn’t there a move to get ready for paper-based submission for subsidized plans?
  5. In mid-September, a testing report is done showing a majority of the functionality is missing or doesn’t work properly. We find out later that core features like eligibility and plan selection don’t work, making the exchange useless for most visitors. Who knew about these test results and all the audit reports? Why wasn’t the launch delayed?
  6. Who specifically, on the government side, approved the “go” decision on September 13 when they had to decide whether or not to launch the exchange? And who agreed to send it to production two weeks later? The audits say they didn’t have the testing needed to make this decision. CGI did not do this on their own. This was a high-level decision. It was either Glen Shor, Jean Yang, someone from EEOHS, or a designated person. Who was it?
  7. Why, after the failed launch, and after obvious technical evidence that no one could enroll, did everyone at the Health Connector, including Jean Yang and spokesman Jason Leffert, continue to tell people that things were fine and that they should keep going to the broken exchange? Why do so many printed materials from the Connector all year say that everything is great? Why does Jean Yang misrepresent the condition of the exchange on PBS and NECN? Why does Leffert tell the Springfield Republican at the end of December that people’s “best option” is going to the exchange? Who authorized this persistent, dishonest communication strategy?
  8. Many people were harmed by the failed launch. Was there any attempt to figure out how many people were affected and to what degree? Many of them called the customer service lines to report their problems. What happened to these reports? Were they logged? How were these cases resolved?
  9. The federal exchange and our exchange both failed on October 1. They were both created by the same company, CGI. The federal government, within two weeks, decides to bring in tech experts right away, along with Optum, and fix most of the problems by the beginning of September. Why didn’t we do the same thing? Everyone in this state involved with the exchange was at the big October 30 event with the federal team at Faneuil Hall. Why didn’t you talk to them about how they were working on the same problem with the same vendor?
  10. The Microsoft report, which contained vital information about the state of the exchange, was delivered to the Health Connector on February 13, but was withheld from the public for a month. Jean Yang’s explanation for the delay makes no sense. It had important new information. Why was it withheld?
  11. In the beginning of February, Governor Patrick acts decisively on the MITRE draft report and chooses Sarah Iselin, who gets down to solving things. Why wasn’t CGI fired at this time? The MITRE report was damning. Optum has the technical chops to know the exchange can’t be fixed and they have also been fixing CGI’s awful code at the federal level. Why did you wait weeks longer? Why did Iselin and Patrick say the exchange could be fixed when it couldn’t be?
  12. Secretary John Polanowicz had jurisdiction over part of this project. He might have let Medicaid Director Kristen Thorn and Janice Baker handle this, along with fellow cabinet secretary Glen Shor. But the Herald has reporting that Shor and Polanowicz had to solve problems beneath them about the exchange project around the time when Polanowicz took over. What did he know about the problems with the exchange, and when did he know them?
  13. The audit reports for BerryDunn are proof that this project was doomed from the start. Since they cost $9 million, these reports surely couldn’t be dismissed. Outside the project team, who saw them? Why were there no apparent consequences for what was in these reports along the way?

  1. In March, the governments of Connecticut and Rhode Island said that our state government owes them millions of dollars as the regional grant money that we got from the federal government ($35 million) was wasted and they didn’t get the technology transfer that was part of that grant. How much money do we owe them? What is the status of this?
  2. Attorney General Martha Coakley has the right in the legal settlement to go after CGI for $12 million under the false claims act. Yet, after saying she was opening a case, not a word from her office about this since. Are we pursuing this money?
  3. There is evidence that data was damaged during the rollout of the failed exchange. This data will cause problems for people in the upcoming open enrollment no matter how well hCentive’s software works. What is being done to handle this?


There are a lot of people I want to thank for this. Most of all, I want to thank The Boston Herald, The Boston Globe, The Springfield Republican, The Lowell Sun, WBUR, and the Pioneer Institute for their work on this story. They provided enough coverage for me to be able to piece together the entire story. I hope everyone goes to the source material I cited to see all the details. Note that just because I used their material, this does not imply that they have endorsed this report. That even applies to those who graciously agreed to review it ahead of time, such as Josh Archambault, who was very helpful in providing valuable details and analysis.

I also am grateful for the existence of The Noun Project where I purchased licenses for all the icons used in this report.

Lastly, I want to thank all the people in customer service at the Health Connector who spoke for so long and so often to help Anna through the months of chaos. There are bad actors in this story, but not you.


There are many sources for this report. I put many of them in the report as links. But I am going to list almost all of my sources here for convenience.

Note that you will need a subscription to the Boston Globe to see their stories, and pay for archive access for the complete Boston Herald articles.

Boston Herald Articles

Directory of All Herald Connector Articles and Opinion Pieces

Infighting delayed state’s ‘Rolls-Royce’ Obamacare site

Connector failures details in ‘not for public release’ report (Microsoft)

Feds to mull Connector takeover

Connector enrollees could lose plans

Deval Patrick refuses to commit to Health Connector disclosure

Connector staffers cashing in

Herald publishes Connector salary list

Springfield Republican Articles

Enrollment in Massachusetts health connector plans falls short due to website glitches

New report details costs of Massachusetts’ Health Connector website failure

New Microsoft report critical of management, technology of Massachusetts Health Connector website

Gov. Deval Patrick hires new management for Massachusetts Health Connector website following critical report

State health officials face July deadline regarding online insurance

Massachusetts to sever relationship with Health Insurance Exchange technology vendor CGI

Massachusetts Health Connector to pay $35 million to sever ties with technology vendor CGI over failed website

Viewpoint: Temporary MassHealth insurance from Health Connector means out-of-network prenatal care

Temporary’ Medicaid group up to 267,000 as state tests new site

Massachusetts to commission independent review of Health Connector website

Gov. Deval Patrick, state Attorney General Martha Coakley considering lawsuit against CGI over Massachusetts Health Connector website problems

State: Revamped Massachusetts Health Connector website passes 1st test ‘with flying colors’

Pioneer Institute Articles

Mass Health Connector 2.0: Limited Function and Unable to Provide Subsidy and Medicaid Information

100+ Unanswered Questions about the Failed Connector ACA Website

Boston Globe Articles

Problems continue to plague Mass. health marketplace

Some who sought health coverage remain uninsured

Mass. Health Connector fraught with uncertainty

Deal reached on botched Mass. health site

Insurance executive to help fix Health Connector site

State dismisses health site firm

State health care site official blasted

Murray criticizes health care site

State knew about troubles with health site

State hires firm for $10 million to fix insurance website

Troubled Mass. health site leads to feud on $45m grant

Mass. paying $35m to drop builder of health website

No money set aside for shift in Medicaid

Mass. may give up on still-failing health site

Mass. sticking with its health insurance website

50,000 filings for health coverage in limbo

New Mass. health insurance website passes key test

New Mass. health website estimated to cost $121m

About 400,000 in Mass. must seek new health plan

PBS NewsHour

Yang and Gruber Appear October 1, 2013 for Launch

NECN Broadside

Jean Yang’s October 4, 2013 Appearance

Jean Yang’s April 15, 2014 Appearance

Lowell Sun Articles

Governor Patrick responds to McDonough criticism

Health Connector revises plans for functioning exchange (SHN)

Getting rid of CGI, $10 million per month in unexpected costs (SHN)

WBUR Articles

About the MITRE Report

Third Party Exchange Reports


Microsoft Report

BerryDunn weekly and monthly report download

Health Connector Materials

CGI Contract and Attachments

The 2013 Summer Broker Newsletter

The Consumer Guide to the ACA, September 11, 2013

Health Connector salary data

Waugh Report notes mention in 2013 annual report on Nov. 30

The ‘How we’re better’ page

Testimony before US Congress | Globe story about it

Some image updates in 2014 about progress

Board Meetings

Main Connector Board Meeting Minutes Page

Feb 27, 2014 Iselin Board Presentation

Annual Reports

2012 | 2013

Weekly Dashboards

Main Dashboard Page

Feb 21 | Feb 27 | Mar 7 | Mar 17 | Mar 27 | Apr 10 | May 8

A Little Bit More About Me

This report isn’t merely an accounting of what happened. Just quoting from documents would leave too much unknown. I believed there had to be commentary and conclusions alongside the facts, otherwise this report could suffer the same fate as the BerryDunn audits: a detached account that doesn’t get across the severity of what went on. Yes, I was very critical — and even outraged — at some of what happened. But I believe such gross negligence that harmed a friend of mine, warranted it.

So here is a bit more about my background, in case you are wondering who is behind the commentary.

As for technology, I have been involved with it almost my entire life, and professionally since 1996. I have held senior technical positions for many years for a few companies. I have worked on large applications like HIX/IES and even for insurance companies. (My opinions here are not in any way endorsed by my current company, or any that I have previously worked for.) I am also involved with the open source community, which is an innovative, global collaborative effort to build the software components that power the world.

As for my interest in government and politics, I also happen to be a Republican activist in Massachusetts. I am not well known outside the party, and I hold no position in the party. (I don’t knock on doors or make phone calls. I am more of a mad scientist who writes about technology and politics, and builds some things.)

As for being Republican, I am quite moderate, and I did not write this to go after the Democrats. I wrote this because I felt I could explain what happened and I was angry about what happened to my friend. If Republican candidates want to use this report to make the case for their election and reform policies, more power to them. If reform-minded Democratic candidates want to do the same, I will be happy also. I just want to see accountability and reforms. I don’t want to see our technology-rich state embarrassed any further or more of my tax dollars tossed into the fireplace.

I also believe in working to improve government, not slash it to pieces. In early August, I spent a weekend working to fix the City of Boston’s permitting system, and created a mobile website to see permit status:

For the record, I was OK with the universal health care law here when it passed several years ago, and I appreciate all that it did for Anna. I have not joined many hyper-partisan Republicans in endless, mindless attacks on the Affordable Care Act. (Of course we should have gotten a waiver here.) But I believe the need to fix government transcends partisanship. After the exchange crash, I even met with my state senator (Will Brownsberger) — a great Democrat — and wrote up a short paper with ideas on how to fix IT project management in this state. It was 100% sincere and non-partisan.

If Governor Patrick — who I did not vote for — called me a year ago and asked for my help, I would have been respectful and agreed, no questions asked.

So, make what you will about my interests in the context of this report. Feel free to peruse all the source material in the references section, and come to your own conclusions. Find a technical architect or two, give them all the audit reports, and ask them how things were going.

I am on Twitter as @mysteriousrook and can be reached at this email address:

This report is licensed under a Creative Commons 4.0 International License. You can reuse any or all of it with attribution and without permission.