Week 34 — Taking Stock
Taking a moment to reflect on a busy half year in a new job with a new baby
Before I launch into the weeknotes, I want to acknowledge that I did not post anything last week. As I mentioned in my my weeknotes from two weeks ago, I am heading into an insane number of major presentations and events which is causing 9–11 hour workdays to become the norm. As a result, I have been prioritizing what little free time I have to my family. I’ve also been working on a new computer upgrade. It is the first time in 10 years that I’ve added parts to a computer so I’m a bit rusty and out of date on the latest and greatest specs. With the computer now finished, all I can say that is my limited free time is so much more enhanced with a lightning fast computer which can run the latest and greatest games on “ultra” graphic settings. I have to remember that investing in my hobbies is worthwhile if it supports my mental health and happiness in other areas of my life.
Tonight was also my daughter’s first (real) Halloween. I don’t count last year since she was only 2 months old. She went out as the world’s cutest owl. Despite the heavy rain, there were a lot of little ones out trick or treating. It was nice to see neighbours come out to enjoy Halloween even if we all got wet. While on the subject of Halloween, can we have a conversation about cancelling Halloween because of the weather? Montreal, we need to have a talk. A lot of parents say they want to build resilience in their children. Parenting philosophies have begun to realize the folly of helicopter parenting and the need to teach children to be self-sufficient. Yet, Montreal, you decided to cancel Halloween because of the weather (and let’s be clear the only weather problem on Halloween was the rain). I don’t get it. How does this help anyone? What does this teach our kids? Rant over. The main topic of the weeknotes is the upcoming storm (avalanche?) of work that is coming up over the next three weeks. Since joining the Canada School of Public Service in March, the work has been keeping me busy (see last week’s weeknotes) but when I looked ahead for the next three weeks, there are a number of big events coming up: two Steering Committee meetings, a major Regulatory Innovation showcase event (200 people attending) and the regular work that can’t slip. It’s a challenge I’m excited to take on although I must admit the anxiety and stress it is causing.

The focus of this week is taking stock of the last 6 months. Why 6 months? You might notice by the week count at the top that I’m a few weeks past 6 months. I didn’t do a 6 month review right at 6 months so there is one reason. Another reason is because the first few weeks I was finding my feet having come back from parental leave and starting a new job at the same time.
Looking back on the last 6 months, it feels both shorter and longer than 6 months. That might sound cliche but it is true. I feel like it’s been shorter because time has flown (where did summer go) and longer because it feels like I’ve been on my projects for longer than 6 months. However, in those 6 months here a few things that I’ve led/been a part of:
- Bring one regulatory AI project (Incorporation by Reference) into prototype stage and start the move to production phase
- Launch and award the first ever AI procurement using the PSPC-TBS Source List for AI
- Manage 8 weeks (and counting) of dual engagement between 16 Departments and two contractors (who are still in competition with each other)
- Launch the first ever Rules as Code Discovery project in Canada
- Establish formal partnerships with 16 Departments and Agencies while cost sharing (fund raising) $1.1 million to cover project expenses
- Present at 13 different governance tables, 4 conferences, 3 events and numerous other smaller meetings about the Regulatory AI projects
- Present to my DG, ADM, DM and to DGs and ADMs at a number of other Departments
- Record a podcast on the work I am doing (coming soon)
- Plant the seeds for a lot of exciting work to come in the next fiscal year (more AI work, more Rules as Code work, more data maturity for regulators work)
- Collect (and one day) disseminate the lessons learned from working on a portfolio of Regulatory AI projects. For example, more is always better when it comes to data and narrow scope with small dedicated teams will always beat wider scopes with larger more distributed teams when it comes to AI projects
Writing out a short list of some of the things I’ve done this year has made me realize that talking about the overall lessons learned could be an interesting blog post in itself. I’ll note it for a future weeknotes when I have time to sit down and collect my thoughts.
Reflecting on the last 6 months has also made me realize other things. None of that list is possible if it wasn’t for strong partnerships built on trust and credibility It isn’t possible without strong upper management support who is willing to have my back and punch using their full weight to clear roadblocks and open doors. None of this is possible if people don’t buy into what we are trying to do with our portfolio of Regulatory AI projects. It’s an odd thing that AI projects live in the School. After living through the experience for 6 months+ I can say it is the right choice. A model of shared risk and cost, the ability to create a safe space to fail and by nature of being a learning environment, the School offers a unique blend of attributes that creates the perfect space for this kind of work.
AI Demonstrator Projects (Incorporation by Reference and Regulatory Evaluation Platform)
Regulatory Evaluation Platform: With 8 weeks of engagements complete, we are about one month away from when the prototypes are due. It is an exciting milestone to be approaching. Both contractors have started to talk about how to validate presentation of results and soon we will also be reviewing their data to make sure we are getting useful insights from what they have built.
We are also starting the reflection and learning process with this project. In particular, we are noting what we are hearing from our vendors, the other Departments and our own experience. Data consistency and quality comes up a lot. The need for a narrow scope is another top issue. In particular, the way regulators perform tasks, the processes they follow and how they collect data (and store it) are not standardized which makes it tough to teach an AI model. AI likes consistency, quantity and quality. When you have 16 Departments who all have slight variations to how they count something, it becomes challenging to train an AI. These insights are also driving ideas for future projects around making more regulatory data machine readable like the Regulatory Impact Analysis and Canada Gazette Data and working on data governance for regulators.
Incorporation by Reference: Phase 2 has kicked off for the project. We are focusing on landing the fundamentals: identification of a true/mere reference, ambulatory/static reference and an “all in one” data export. If we can perfect these three features, then we will have a successful prototype. A few weeks ago, we demoed to our DM who found the single thread in the ball of yarn and yanked it. He noted how some “Statutory Instruments” were in the data set while others were not. Ultimately, we decided we don’t need to display Statutory Instruments at all in the tool as it adds no value to regulators. With that being said, it speaks to a larger lesson learned around framing your work before presenting to senior leaders. It is one thing to understand the project and even its purpose. It’s another thing to set the frame of what’s in, what’s out, what is it supposed to do, what problem is it solving and what are its limitations.
Rebuilding the Public Service From The Ground Up: Week 20
For week 20, my idea is to re-think approvals by introducing a common sense quotient.
Idea 20: Common Sense Approvals
You may be familiar with the Meeting Cost Calculator developed by Sean Boots. It’s a great tool to calculate the cost of meetings in the public service. I want something similar for approvals but specifically for the value of approvals to be part of the calculation on deciding who needs to approve something.
Here’s the thing about the public service. We’re not very good at estimating value. I need a DG level approval for a $500 trip or even $100 of coffee for an event. At the same time, I can set-up a meeting with 20 people for a full hour (or more) with no approvals. Even in jest, I can’t say the solution is to add approvals for meetings. However, it speaks to a larger problem of seeing staff cost as a fixed cost. By that I mean, people assume that since we have the bodies in positions, we have to pay them anyway regardless of what they are doing. Travel and hospitality are not fixed costs (e.g. they are discretionary). Staff costs (assuming a person in a box) is a fixed cost. You can’t (easily) dump that salary cost.
So let’s introduce a new way of doing approvals. What is the value in doing the approval? Will it cost an amount equal to the value it delivers? Will it cost more to do the approval than the cost of the transaction? Is it worth $1000 of staff time to approve a $2000 event? I don’t know the answers to those questions but I do feel it is time we start asking how our time and the time of our senior management is better used.
That’s it for week 34. See you next week!
