I don't know anything about how Presto is programmed, but I have considerable painful experience with custom programs for the payroll module of a certain well-known brand-name enterprise software system, trying to automate calculations of some arcane pay rules arising from collective agreements. The price for automating said rules was frequently astronomical. It's amazing how the human brain can handle these things easily, whereas laying the decision rules out in a linear, programmable fashion has a million "if" conditions and exceptions that have to be programmed and then validated. With these big systems, the more custom programming you create, the higher the cost both to create the function and then to maintain it as the base software is upgraded. Plain vanilla is almost always cheapest in the long run.
I have a suspicion that the TTC transfer table is pretty arcane in places. I sure hope they take a good long look at whether the 2-hour rule might be simpler and cheaper for the long run.