Hacker News Reader

Comment by conistonwater | Hacker News Reader

> That said, you'd be surprised how often the fact (pun intended) that it's "impossible to know anything for certain" is lost on people.

I think David Hume was quite good on this subject. If chance of being right is all there is to knowledge, then maximizing your chance of being right is what knowledge is. So all the peer review, the numbers and statistics, for example, contribute to knowledge even if it might be wrong. And if you dismiss, as you do, statistics because they just might be misleading or biased, and are thus not "objective reality fact", that does damage to your ability to know things, because you are discarding data without considering it, and that damage should be avoided. Whereas the proper thing to do is consider it on merits, whatever the actual merits are.

View on HN

Selection bias is a huge problem though. Statistics in politics basically have negative utility because people just omit the ones that contradict their favored narrative, and then you get this:

https://xkcd.com/882/

When all you have is that newspaper headline, your knowledge has gone down rather than up.

Maybe, but consider adding another constraint to the game of maximizing your chance to being right -- you have a finite amount of resources (eg, time) to spend on examining evidence, so you need to strategize even what evidence is worth considering and to what extent -- looking more carefully at one thing means you now have less time to look carefully at something else, nothing is free

Of course, in real life, we do have this constraint

> Of course, in real life, we do have this constraint

Which IMO is what causes confirmation bias. But science should strive to get rid of confirmation bias as much as possible.

Your comment is vacuous if you don't expand on it; it goes without saying we want to try to minimize bias as much as possible. Science is part of a world of limited resources too, and investigators have to decide what experiments to try, what topics to pursue

In the context of confirmation bias being an evolved trait, my comment was meant to imply that we should expect science to spend much more resources than mere intuition, but doing so is still worthwhile.

Any scientist will be subject to cognitive bias at some level (if only by choosing the core principles and axioms on which their reasoning is based), but there are systematic ways to mitigate it. Applying more resources to a particular problem will get rid of at least the superficial levels of cognitive bias.

I love this concept as a defence against the 'well virtually anything can be proven wrong or right' anti-knowledge argument...

Reply on news.ycombinator.com