Procurement under Trump
If Trump wants to improve regulation, here’s where he should start.
I remember when I first really understood how screwed up government technology procurement is. It was 2011, the first year of the Code for America fellowship program, and one of our fellows had just launched a website for the City of Boston. The city had changed the public school selection policy to promote more kids walking to school. The process was now essentially a mapping problem; the primary factor was the distance from your home to various schools, with some other factors such as whether the child had a sibling in another school or special education needs. But to communicate this change to parents, the Department was still sending a 28 page printed brochure with descriptions of each school written in seven point type. The brochure contained a lot of information, but it did not — and could not — tell you which school YOUR kid could attend. Parents were beyond frustrated.
One of our fellows, Joel Mahoney, took on the job of making a website that let you enter your address, the age of your child, and the school(s) any siblings attended, and returned a map showing the schools your child could go to. He had part time help from one or two of the other fellows. It took him about ten weeks to launch a beta version of the site, which worked remarkably well, and was, as another fellow said, “simple, beautiful, and easy to use.”
Joel and the other fellows were working in Boston courtesy of some very forward-thinking public servants who had a lot of political backing from Boston’s then mayor, Tom Menino. These city employees, Nigel Jacob and Chris Osgood, had contracted with Code for America for the program, and under those terms the fellows could work on a wide variety of projects. This meant that the work they did wasn’t subject to procurement rules. Chris and Nigel taught me about procurement, starting with the fact that if Joel’s project — which took one and a half people about ten weeks — had gone through a regular procurement process, it would have taken at least two years and cost at least $2M.
I remember thinking at the time, “Well, I guess we know what the enemy is now. We’d better change procurement.” I became something of a collector of government technology procurement horror stories. No, it should not cost $2B to share documents between sites of the California court system (and to subsequently scrap the attempt). No, it should not take 19 years to do the same for Massachusetts (still not shipped). No, it should not cost $11B to manage medical records at the Department of Defense and the Veterans Administration (yes, this procurement came out while I worked in the White House, to my significant dismay.) I wish I could say these were outliers, but they weren’t. I’m happy to say that’s changing, but I’m getting ahead of myself.
Over the course of the subsequent five years, I discovered that I was probably the ten millionth person to have realized that procurement is messed up, and maybe the 100,000th person to have the audacity to think I actually could change it. The starting place, naturally, is to want to change the rules. It turns out that under the circumstances I’ve worked — in city, county and state governments as an outside partner and in the White House as the Deputy Chief Technology Officer, serving an administration whose party was not in the majority in Congress — changing the rules has been either impossible or merely unthinkable. So I and many others have focused on a variety of different strategies to help governments do good digital work within existing law, policy and regulations. It is possible, as evidenced by the hundreds of procurement hacks we’ve seen over the past few years and recent examples of procurements designed to meet user needs. But it is hard. And changing procurement regulation sure would make it easier.
Now a Trump administration is coming. I hope procurement will be in its crosshairs. A Trump White House will govern with Republican majorities in both the House and the Senate. It’s possible that a Trump administration, with the help of folks on his transition team like Peter Thiel, who has founded companies that have been publicly frustrated with government procurement, could do what President Obama and governors and mayors across the country could not. If they go after technology procurement, can they do it, and will they get it right?
So, can they? Let’s assume for a second that the Trump team goes directly at the Federal Acquisition Regulation. The FAR is an 1895 page document (as downloaded from acquisitions.gov) that’s enormously complex. It also appears difficult to change, though perhaps less so than I had assumed, per a Congressional Research Report from 2015.
The language of rulemaking is sometimes difficult to decipher, but if I read it correctly, it says that if the Department of Defense (DOD), the General Services Administration (GSA), and the National Aeronautics and Space Administration (NASA), get together, acting on behalf of the FAR Council, or the Administrator of the Office of Federal Procurement Policy (OFPP) “they can issue proposed and final rules amending the FAR under the ‘notice-and-comment’ procedures of the Administrative Procedure Act (APA).”
There’s more to it than that, but there’s also more than rulemaking at play. There are the interests of those who benefit from the rules as they exist today, and that deserves greater discussion. But the bigger question is, will Trump appointees want to amend the FAR, or fundamentally blow it up?
Since Thiel is leading the transition effort on digital, the analogy to code will be resonant. Regulation functions in some ways like computer code does, executing rules towards an outcome. Typical computer code is amended far more often than regulation, to add features, fix bugs, or improve stability. But regulation and software both accrue a wide variety of changes over time, and eventually the code, whether computer code or procurement code, becomes bloated and slow. Software programs need a total rewrite when they suffer the resulting performance and usability issues (the former being an issue of speed of the program and the latter being an issue of speed of the user when confronted with too many features, many of which are not needed and/or rarely used.) The notion that complex government regulation like the FAR is in need of a total rewrite for greater speed and usability predates my career in government technology.
I don’t exactly know how one would completely reboot the FAR, and I don’t actually know that the Trump camp intends to do so. What I would say is that I know hundreds, possibly thousands of people, many of whom served previous administrations, who have long wanted this. Those of us most recently in the trenches have been working to fix the larger problem of digital competence in government through other means, because a complete overhaul was not an option in our world. Many of these people are the employees of the USDS and 18F, many of whom have come from “metaphysical Silicon Valley” (aka tech companies who live and die by meeting user needs.)
Many of them have, as I do, concerns about many of Trump’s proposed policies and are grappling with how to engage with his administration. I’ve written about this elsewhere. But this group knows federal technology procurement, and they know it primarily as an obstacle to serving the American people as they deserve to be served, not as the protection against cronyism it was designed to be. They also know how the Paperwork Reduction Act, the Privacy Act of 1974, countless meaningless security protocols and other forms of compliance for the sake of compliance, and a risk-averse bureaucracy — all make it difficult to execute on the best practices that they have learned in cutting edge technology companies. They’ve been succeeding in making technology that works for the public in spite of these obstacles for the last few years, but that doesn’t mean they are satisfied with the scale of change, and it doesn’t mean they accept the status quo. On the contrary. They know how much still needs to change to fulfill the promise of our movement, which, as we say at Code for America, is to make government work by and for the people in the 21st century. And procurement is one of those things that need to change. But what I worry about is this:
The law of unintended consequences is a bitch, and nowhere does it wreak more havoc than in law, policy and regulation related to technology.
I’m not talking about the risk that changing procurement rules could remove protections against cronyism; those protections don’t work now. I’m talking about the fact that the Paperwork Reduction Act, the Privacy Act, the FAR itself — all of these were implemented by well-meaning change agents for the benefit of the American people. They were intended to fix problems of the very same nature as the problems we face today. And each of them has had largely the opposite effect. It’s the nature of lawmaking and rulemaking (today, at least) that you tend to get one chance to propose your idea for how to fix things, you get little or no chance to test it, and then it runs unchecked. Technical folks: I don’t even need to make this analogy for you.
Clay Shirky had a great line in his piece on the launch of healthcare.gov:
[Waterfall development is] a perfect method for a culture that communicates in the deontic language of legislation.
That line sent readers everywhere (including myself) to look up deontic, which apparently means relating to or expressing moral duty or obligation, but comes from the Greek dei, or it is binding. It left me with a vivid image of all the legislation and regulation to which we are bound today and the ways in which they hold us back from meeting user needs, even when we have the best of intentions.
We are not consigned to a fate in which legislation and regulation forever suffer from the inevitable limitations of “once and done.” Nick Sinai has written about “regulatory sandboxing,” in which constraints are relaxed in order to allow for activity from which regulators can learn while they debate and finalize rules. Tim O’Reilly (in the interest of transparency, my husband) has written about a new model for algorithm-based regulation. His chapter in the book Beyond Transparency lists characteristics of successful regulation:
· A deep understanding of the desired outcome
· Real-time measurement to determine if that outcome is being achieved
· Algorithms (i.e. a set of rules) that make adjustments based on new data
· Periodic, deeper analysis of whether the algorithms themselves are correct and performing as expected.
Perhaps a precursor to this model is to be found in a story of “iterative regulation” I told at the Code for America Summit a few weeks ago. A Digital Service team working at the Department of Health and Human Services worked side by side with regulators to create an MVP (“Minimum Viable Product”) of the digital service that would implement a new regulation, then tested it with users, discovered its weaknesses, and revised the draft regulations multiple times based on results with real users. That’s groundbreaking. The Consumer Finance Protection Bureau has similarly tested disclosures with users before finalizing language. Case studies like these are no longer theoretical, and the people practicing these methods give me hope that we have a chance against the law of unintended consequences when we take on the work of reforming government itself.
Procurement reform is a place where the Trump administration and a Republican Congress could have a meaningful positive impact. The teams in USDS and 18F have enormous insights in what many of problems are, what has been tried, and where the pitfalls lie when contemplating reform. And they are some of the smartest, most determined people I’ve met. If the administration wanted to tackle this issue, these teams could be incredibly helpful.
Speaking of people, while I and many others will welcome procurement reform done right, there is no successful strategy for making government competent at digital that doesn’t put talent first. There are too many government contractors who are successful because they are good at procurement and a wide variety of flavors of meaningless compliance (the latest maddening compliance example is here) but who lack any consistent ability to build software that meets user needs.
While procurement reform would help, the people inside government have been trained in the current system, and it is central to acquisition workflows. Crafting a procurement that meets user needs and delivering on the procurement with the vendor who is selected (both jobs that sit inside government) require a very different set of skills and practices. Getting better vendors doesn’t fix the problems of actual service delivery, because those better vendors won’t get the job, or if they do, can’t deliver better results if the people leading the process from the inside don’t operate from a framework of “meet user needs” instead of “meet government compliance needs.”
The work Code for America and 18F have done with the State of California on their child welfare program is a great case in point. Helping the California Department of Health and Human Services transform what would have been a $500M six- to seven-year waterfall project into a modular, agile process, the first award of which went to an existing solution with a proven track record, did not alleviate the need for recruitment, retraining, and realignment of the internal team. In fact, it accelerated that need. The State of California, both its leadership and the departments, has done a fantastic job stepping up to the plate. They are clearly on a path from a long line of expensive IT failures to digital services that meet user needs at a reasonable price. That path is always challenging, but it would be impossible with a narrow focus on the procurement rules alone. (And by the way, they are hiring. You should apply to work there. Seriously.)
There is a lot more to say about the range of focus that’s needed to effect real change within government. David Eaves has written insightfully on this topic, pointing out (correctly) that IT projects are fundamentally adaptive problems, not technical problems, and no process can guarantee the right decision no matter how tightly or smartly you design it. His conclusions mirror those of everyone I’ve met who first encounters government from the inside; most also cite the clear need for civil service reform, without which progress on the talent issues stagnate. (The Partnership for Public Service has good recommendations here.)
Moreover, the presenting problem is often bad technology, which can be fixed by a better interface, but in digital services, following users past the initial interaction leads you to a whole host of operational issues. We’ve learned that in great depth at Code for America over the last six years, and have ultimately evolved our organization beyond simple technology fixes to address the larger issues, with great early success. Any new administration will ultimately have to travel the same path once it really takes on the messy business of governing. As Dan Tangherlini, the former GSA Administrator who championed and nurtured 18F, said:
…our acquisition system does not suffer [only] with problems of policy or regs, but of systems and organization — the latter are entirely the realm of the executive.
We’ve spent years wishing for a less risk-averse federal government. Perhaps now we will have one, courtesy of a man who shows little regard for protocol in the White House, backed by a Republican-controlled Congress. In my six years of running Code for America and playing a role in federal government, ideas like reform of the FAR were unthinkable. Now they are not.
FAR reform could significantly accelerate the kind of progress we’ve been fighting for. Sure, there is reason for caution. If done badly, it could saddle us with an entirely different set of problems instead, and if not done iteratively over time, it’s likely to have unintended consequences. It must work in concert with changes in systems and organizations, not just rules. And it must be done while keeping faith with the ideals of our nation, to serve all the people. But if done right, it’s worth doing, and the good news is there are hundreds of USDSers and 18Fers who’ve been showing what an agile, risk-tolerant government looks like. There’s no substitute for the scars they’ve earned fighting this fight, and no question that they’re here to get better results for the American people. If they share that goal, I urge the Trump team to learn from them.