> I feel like evidence dis-confirming a hypothesis is as valuable as confirming it.

In particular as that would mitigate problem #3 above ("3. The third, and hardest to solve, problem is not factoring in the whole population of experiments.")