LogFAQs > #910556966

scar the 1 posted...

Yeah, but what do you expect? Reviewers work for free, and rarely if ever have resources to re-run experiments and as such validate that data isn't faked. Scientific publishing is - and has always been - heavily reliant on trust. The "study" ITT abuses that fact to discredit particular fields of study.

Peer review, in general, is fucked. There's competing incentives, lack of time spent providing an actual thorough read through the work, this reliance on "trust" which can roughly be translated to the "benefit of the prestige", etc...

There's very little chance that peer review does a good job arbitrating good science. The system needs to be revamped in some way. At the very least, making peer reviews public record would go a long way.

This is all to say, this is almost unrelated to the current topic. An article goes through a desk editor and some number of reviewers. Discrimination of complete bullshit from real work shouldn't be this tough, even in our current system. Good science vs bad science, well that's a tough judgement, requires very careful reading and pretty pure incentives. "Real" vs "pulled out the ass in a deliberately hoax-y way" is a different story.
---

=E[(x-E[x])(y-E[y])]