I agree there are myriad issues with research and the current peer-review process. It's still, however, our best possible means of getting a less-biased peek at the world.

I think incentivizing researchers to do more studies replicating previous studies (and verifying their results) would be a step in the right direction. Researchers are currently incentivized to do unique research, so they can make a name for themselves by getting exciting results. Maybe a rule stating, "You've got to do two studies replicating a previous study, then you can do your own research" would help. I'm not sure who would enforce that rule - peer-reviewed journals, perhaps? But I think that would help 1) keep researchers honest when doing pos- hoc tests etc so attempts to skew data aren't buried deep in the paper, 2) help verify (or invalidate) existing/accepted theories.

Anyway...