4 ms·
> I assure you with large enough sample size you will find some correlations though. I can correlations the drop in pirates with a rise in global temperature,
by gnaritas 10y ago
> I assure you with large enough sample size you will find some correlations though.
I can correlations the drop in pirates with a rise in global temperature, so what... correlations don't mean a damn thing by themselves, that's just data-mining bias.
- nonbel 10y agoNo, it is not data-mining bias. There really is a correlation. It is real, but realizing that has no usefulness at all.
- gnaritas 10y agoThe correlation with pirates is real too, that's my point, correlations don't mean anything. Only causation is meaningful. The only use of a correlation is as a starting point for finding causation.
- nonbel 10y agoActually... I think causation may largely be a red herring as well. Our most successful models really do not incorporate it at all. They are in the form of F=ma or PV=nRT. There is no causality there.
- gnaritas 10y agoModels are final solutions; causation is how you end up finding those final solutions: causation is understanding, it's vitally important.