Week 51 — What A Year of Innovation Has Taught Me Part 1
Why Innovation In Government Is A Relay Marathon
I’m sitting in a hotel in Montreal as I write this blog post. I’ve been away from home nearly all week delivering presentations to a crowd of Health Canada staff and inspectors in both Toronto and Montreal. Somehow, I’ve managed to time my trips around major snowstorms. This winter alone I have traveled three times and all three times I have avoided major snow storms back home in Ottawa.
Last week, I wrote a post with a substantial update on the Rules as Code project. After a good chat with Jason Morris over the weekend and some further reflection on the post, I want to clarify a few parts of the post. In the post, I wrote how we were trying to address the question of how much vacation pay an employee is entitled to. That question became our focal point because we wanted to see if we could answer the question simply by encoding the vacation pay entitlement rules. However, I want to add some clarity to the post especially around what impact our focus question has on the Rules as Code process. Our scope was to do a translation of the existing vacation pay entitlement rules. We did not set out to test co-drafting in code nor do we want to test automation of the rules. We are still deriving the meaning behind the rules, mapping out in a logical way and encoding those rules using computer code. Our intent in setting the focus question was to limit scope, demonstrate what could happen after Rules as Code (e.g. the code is used in a practical application) and to give us focus. Will we have an encoded version fo the rules which can be used so a computer can derive or make a determination based on the rules? Yes, we will. Will we also test whether we can build an application using the code to answer a basic question about vacation pay? Yes, we will. It’s a fine line between Rules as Code and automation. We might venture into both during the project but our intent remains to test Rules as Code.
Public Sector Innovation is a Relay Marathon and Not A Sprint
This week I presented to Health Canada staff and inspectors in Toronto and Montreal. You can read my presentation here.
After a year of working on the demonstration projects, we’ve learned a lot of lessons about public sector innovation, both at the organizational and individual level. This week, I’ll share one at the organizational and one at the personal level. I’ll share more next week but these are the two ones I love the most.
As a public sector innovator, you have to remember that innovating in government is a relay marathon, not a sprint. Why a relay marathon? First, it’s a relay because you don’t do it alone. You work with others to win the race. You pass the baton onto others who are part of a team which finishes the race. You don’t win a relay race by yourself. For the public sector innovator, it’s a reminder that innovation is a relay race, it’s coop not single player (to use a video game analogy) and so you need to remember the importance of finding your allies and working with others. Secondly, it’s a marathon because innovating in government is not a fast process. You have to be in it for the long haul. You aren’t going to see an overnight transformation of a major entrenched system with a rich history and ingrained processes. When I worked at Transport Canada, I always liked using the analogy of a large ship; as in a large ship takes a long time to turn around. Well, so does the public service. You have to understand that innovation is about the long game.
When it comes to lessons learned for organizations, I’d like to focus on one of our data lessons learned. Government generates a lot of data but it’s more than the quantity of data, it’s also about the quality of the data. As the old saying goes, “garbage in, garbage out”. So as an organization with lots of data and wanting to do something with it (maybe use AI?) you have to remember that just because it feels like AI can solve your problem doesn’t mean you have the data needed to use AI to solve the problem (of course I am under the assumption AI is the right solution to your problem). Here’s an example that many in government will understand. If I asked a room of public servants what a full time equivalent (FTE) is, how many different answers would I get? Chances are I would get a different answer from every single person in the room. Now imagine, we are (or have been) collecting data on how many FTEs our Department has. Each team has been tracking how many FTEs they have and eventually it gets rolled up at the Departmental level. You can see the problem if we don’t agree on the definition of an FTE. This means everyone in the Department is counting their FTEs differently. It means we can’t trust the data (e.g. no data integrity). It means that any kind of analysis won’t be useful. It means that introducing AI or advanced analytics won’t help us because our data is hot garbage. Fixing these kinds of problems isn’t sexy. You won’t get a lot of Twitter buzz for doing it. At the end of the day though, this kind of slogging tough work is the backbone of a future where you can derive insights for the organization which support decision making.
AI Demonstrator Projects (Incorporation by Reference, Regulatory Evaluation Platform, Rules as Code)
Regulatory Evaluation Platform: We have started to explore different ways of enhancing the ability of the tool to measure similarity of regulations. We are considering using a Word2Vec model but we need to explore whether it is feasible or not.
Incorporation by Reference: The prototype for this project is under testing. We will be measuring its accuracy and whether the reduction in manual labour facilitated by this tool reaches the levels we want it to/as promised.
Rules as Code: This week was an off week for this project as I was traveling all week. Next week, we will be sharing the first pseudo code with the broader group of subject matter experts and drafters. For those watching, we expect to have a report to share on our work by mid April.
Future Demonstration Projects: We have a number of demonstration projects planned for next year. I’m thinking it will be apt to start sharing these projects after week 52 (one year after I started at the Canada School of Public Service).
Week 51 is in the books. I will continue sharing more lessons for week 52 before transitioning to a detailed look at new projects starting in week 53. Have a great week!
