Conversation
Notices
-
Embed this notice
How did they prove to the common man that the earth revolves around the Sun?
image.png
-
Embed this notice
@TURBORETARD9000 @Awoo Unfortunately this only works if you trust the person replicating the experiment and the experiment itself.
Here is a pathological example which I have personally encountered: a professor claims that their neural network performs a task, and the performance is improved by both X data and Y data. CMD's coworker replicates the experiment uncritiically and gets the same result. CMD replicates the experiment, makes an observation about the results, and then only feeds the network X while feeding in a constant value for Y. The neural network's performance is improved vastly by only feeding the network X, and the entire improvement over the baseline was caused by increased depth of the network made to feed the network both X and Y.
Even though the experiment could be replicated, because there was a flaw in the experiment we had an incorrect conclusion. Much assmad was caused, and months of R&D money went up in flames because some faggot at a university pulled the wool over the referee's eyes.
-
Embed this notice
@Awoo replication
-
Embed this notice
@Awoo see: replication crisis
Solution: start replicating like a motherfucker
-
Embed this notice
In fact how did they convince anyone of anything that was backed by science?
image.png
-
Embed this notice
@ceo_of_monoeye_dating @TURBORETARD9000
image.png
-
Embed this notice
@Awoo @TURBORETARD9000 ...tl;dr, even when your friend replicates something it doesn't mean they actually did it right.
-
Embed this notice
@ceo_of_monoeye_dating @Awoo @TURBORETARD9000 Just a good analogy frim doing everything your self as well.