Hacker News Reader

Comment by goto11 | Hacker News Reader

That would mean you ultimately have to personally reproduce the results under controlled conditions before you can trust anything. This is just not realistically possible for anybody, so in reality you have to trust somebody.

View on HN

I did not say there was a need to personally verify the results of every experiment. Merely pointing out that the scientific process is not based on trust, but evidence. I would not believe a claim made by a single experimenter/group unless independent evidence corroborates it. Neither should you. Look up the Heidelberg-Moscow experiment, or pentaquarks.

You are still putting trust into the fact that the groups/experimenters are actually independent. Unless you have personally verified every experiment, you are still putting trust into the fact that others have done the work as they claim. What you are doing is being _pragmatic_ about your trust. You have a level of trust that is acceptable.

The reality is that we all are trusting "science" to one degree or another. I absolutely agree that all experiments should be repeatable and independently verifiable, but in the real world science (in my opinion) very much depends on trust. We don't all have time to verify every experiment.

Separate experimenters making claims to the same results is in itself a form of evidence. Believing otherwise requires either the existence a common error or a conspiracy.

> Believing otherwise requires either the existence a common error or a conspiracy.

Both of which are common.

Yeah but how do you then know the "independent scientists" actually reproduced the results and didn't just make it up? In the end it comes down to trust in the process and the community.

You don't. Case in point the drift in measured values of the fundamental charge of the electron over time starting with Millikan. However, given enough independent evidence you can conclude whether or not the results of a measurement are more likely than not.

Reply on news.ycombinator.com