There may be different ways to more or less model the same thing. The trade-offs may be machine efficiency, human grokkability, prove-ability (mathematically an

There may be different ways to more or less model the same thing. The trade-offs may be machine efficiency, human grokkability, prove-ability (mathematically analyzable), and accuracy (sometimes approximations are good enough, and gain on other features). Perhaps related: Factor tables for AI: https://github.com/RowColz/AI https://github.com/RowColz/AI