Philosophy

If I predict rain tomorrow based on satellite and radar imagery showing a cold front forming, and it does rain, I’m right.

If I predict rain because I saw a black cat walk under a ladder, and it rains, am I right (but for the wrong reasons) or am I just wrong?

Can one ever be right for the wrong reasons?

Aviator, diver, Etherean, nexialist, PhD in aviation ∩ climate science | Host of /diving /ethfinance /eth-staker /geopolitics /science /singapore /thomas

The only problem is externality pricing.

Building https://negationgame.com

Right, and that would require you to accurately record that the predictions were *right*.

But if you practice justificationism you'll happily throw out the predictions because they come from a methodology (a model / explanation) you don't endorse, so they can't be right.

iow, to be wrong you must grant they're right

Like everything else, repeatability under controlled conditions (that and Nestle would have purchased the rights to all black cats)

What information would you have to see in order to conclude that black cats under ladders is actually informative of weather patterns?

Thank you! Getting problems are indeed what I had in mind, inspired by this old post:

Be specific.    
Building productscore.org, @germanify // leonasskau.co.uk // Hosting /geopolitics, /strategy, /leo
d33m:mun

some fun facts about Gettier:
1. my smoke/fire example comes from 8th century Indian philosopher Dharmottara, who mentioned this problem well before the West even had universities

Thank you! Getting problems are indeed what I had in mind, inspired by this old post: https://www.lesswrong.com/posts/7c7crvbG62KL8kAuW/right-for-the-wrong-reasons

This is almost a Gettier problem

The traditional criteria for knowledge is a Justified True Belief. Here, your belief is not justified, because cats do not relate causally with rain

So you are just wrong
But it gets more interesting…

21. hacking.
previously @spire, intern Nethermind, epns.io

yeah absolutely
ig that's why deutsche gives so much emphasis on explanations

That’s my view as well, and yet it’s difficult in practice to tell people that they were not “right” even though they correctly predicted an outcome, because their reasoning was flawed or provably unrelated to the determining factors of the outcome. We see this all the time among influencooors in crypto for ex

Dad, geek, conversation liquidity provider. vrypan.net (home), @l--o--l (fun). devcoin:$lemon3

… that the 1st coin was rigged & had heads on both sides, & I knew that. So not only was I right in my prediction, I have the reasoning to prove why.

The second coin was fair, and I was just lucky in making a 50/50 guess. I predicted right, but my method was no better than chance. Is that the same “being right”?

My concern is where the prediction maps onto reality but for erroneous (i.e., probably wrong) reasons.

Let me attempt one more example:

I predict that two separate coins, when tossed, will fall on heads. I toss both coins, and sure enough, they fall on heads. I was right in both predictions.

It is then revealed…

You have three "objects":
Predictions, Reality, and a function Outcome(p, for p in Predictions) that maps Predictions to Reality.

So, what's your question?
If Outcome() is a good function? If outcome(p1) for a specific prediction p1 was right?

If there’s one thing I’ve learned from LLMs, it’s about sustained signal over time

Being sometimes right but much more often wrong on a specific topic is a net negative

You can test for causality in your reasoning for being right and that's really what matters https://ftp.cs.ucla.edu/pub/stat_ser/r350.pdf