At the highest level, I have learned that school reform is much more complicated than I had imagined, and it had been pretty complicated already in my imagination. This project has reinforced that notion as I have tried to visualize some small part of it in a quantitative way.
I have gained a much greater awareness of the distinction between problems and dilemmas. Certainly much political debate, on school reform and other topics, would benefit from making this distinction. I have not given up the search for solvable problems, but there are few that don't come with hidden dilemmas. For example, if (hypothetically) ability grouping could make school better for 70% of students, there is still the dilemma regarding the other 30%. Especially if that 30% is a typically underserved population, that dilemma is hard to manage even if numerically the "solution" to the problem is obvious.
Before the purpose paper, my concept of reform focused on the "what" without much regard for the "why". I had imagined that adding a keel would change some of my other views but, I guess because the mind can be rational even when operating unconsciously, or maybe the mind can rationalize any irrational thought, most of those other views have survived, if in more nuanced form. One of those is that I still feel that some form of ability grouping is better for most students (I can "prove" it with this project, at least with the right assumptions!) and that I feel even more strongly that forced busing hurts cities and the people in them, both minorities and the majority, and should be abolished in favor of the neighborhood schools they displaced, with equitable funding. In fact, the project I now wish I had done was some kind of fictional "history" of the next twenty years of education on the assumption that, in the near future, Brown gets reinterpreted definitively as forbidding only de jure segregation.
This project was a killer but it was also a lot of fun. It reminded me why I still love playing with computers but also why I need to get out of the business of programming them. As I said in the conclusion of the guide, programming something makes you really understand it, and even though the thing I really understand is frightenlingly oversimplified, going through the exercise of simplifying a small piece of this complex topic enough to commit it to code has made me understand some of it better. Even simple questions of treating the levers as mathematical functions forced me to think hard about what sorts of functions they might really be even as I mangled them into linearity for the sake of my sanity.