Sure, but
> You can't extrapolate this stuff.
Why not? I'm not saying it's like physics where we can make nearly exact models, but can't we at least make educated predictions based on previous examples, taking into account the different circumstances?
After all, in your hypothetical you already implicity modeled that a minimum wage a certain percentage of a "reasonable" wage won't hurt employment, while one much higher will.
> The question is not "Are minimum wage increases good in general?" the question is "Given the potential tradeoffs, is this specific minimum wage increase good?"
This only underscores the importance of extrapolation -- lawmakers need to be able to predict these tradeoffs to make good decisions (if extrapolation is possible).