Hacker News new | past | comments | ask | show | jobs | submit login

Clearly people can and do make predictions and sometimes those predictions are quite right - and sometimes wrong. Even a really really good predictor like Nate Silver gets things wrong.

Future predictions are not fact checkable. You can argue likelihoods, present contrary evidence or whatnot but predictions are not facts, they are predictions.

A good predictor gives odds to every outcome. That is not something a fact checker can respond to well.




> Even a really really good predictor like Nate Silver gets things wrong.

That's because if he doesn't use mathematical modelling and data he's only ever accidentally correct. There's no such thing as good predictor. If you don't use knowledge (and the only real knowledge comes from scientific process) you can't get your predictions better than chance.

It's a combination of brief luck and "predicting" the obvious.

With interesting personality you can make a good career out of predicting that dice roll will result in >1. You'll by wrong in less than 20% of cases.

Being good predictor is getting popular while you are on the roll. There's nothing else going there.

All predictions that don't have a form of scientific paper with clear mathematical model and ample data are just making stuff up and should be labelled as BS in all media.

World doesn't need prophets.


Knowledge is not a binary. The world is not a mix of "true" and "false", especially not in the realm of analyzing future outcomes.

You need to not only get multiple analysis, but normalize them, give them confidence scores and then interpret them - and you may interpret them differently than their authors. You may entirely discard some.

How do you get a confidence score? Sure sometimes we can analyze the history of an analysis such as with political polling where we have someone like Rasmussen who has done it for a long long time... but you still need to account for changes in their own process. If Rasmussen gets a new head data scientist tomorrow, does that alter our confidence level?

"Just use math!" is about the same as saying "Just don't be wrong!". Math is a tool, not a magic 8 ball.


Knowledge is not binary. Knowledge is a subset of science. If you draw from outside of science you can't have any knowlegde. An can't make any predictions or even evaluations of the present better than chance.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: