[jjoonathan]

Translation: "this shit is completely fucking hopeless."

Multi-drug combination treatments are where studies of ineffective treatments go to die. "X didn't work, but maybe X+Y or X+Z will" is one of those low-quality ever-present followup leads I was talking about in the sibling post. To a researcher, seeing attention shift to combination treatments means that the primary hypothesis failed and people are throwing Hail Marys. It's not impossible that one of them will work, and that's why nobody will actually publish a paper saying "this shit is completely fucking hopeless," but in a situation where this shit really was completely fucking hopeless I would expect to see exactly this: a couple of positive "hail mary" results that squeak over the significance threshold with low N and low effect size.

[johnrgrace]

Simply Throw enough Hail mary tests with 95% confidence intervals and one will come up as good simply due to randomness.