[swebs]

In science, evidence determines your viewpoints. In politics, its the other way around.

If the results of an experiment support your views, then it was a great experiment and proof that you were right all along.

If the results of an experiment contradict your views, then you can fully scrutinize every aspect of the test to come up with reasons why it got such a wrong result.

[plutonorm]

I don't think science is free from those flaws either. Science advances one death at a time.

[swebs]

There are certainly plenty of fields that have a these problems and are vulnerable to be politically driven. I'd say the further toward the "soft" sciences you go, the more you're able to manipulate results to fit certain viewpoints. 538 has a nice little demo showing off the problem.

https://projects.fivethirtyeight.com/p-hacking/

[KittenInABox]

Is CS a soft science? There’s a ton of massaging in machine learning IME.

[swebs]

Yes, some branches definitely. Especially human-computer interaction. Other branches like discrete math or algorithm analysis are impossible to fudge. Machine learning is somewhere in the middle.

[travisporter]

Awesome site! Sadly p hacking is also a problem in many fields not just the soft sciences. It’s a bit of a dig at science but is it a fundamental flaw of the scientific method?