Ask HN: What ever happened to probabilistic thinking? | Hacker News Reader

Ask HN: What ever happened to probabilistic thinking?

4 pointsarisAlexis4 comments

I thought probabilistic thinking was the way programmers and rational people in the tech & science world used.

Lately I see b&w messaging like:

"We will never cure aging". "We will never reach AGI in our lifetimes". "AI has 0% of catastrophic risk - Lecunn". Etc.

How can people in tech think in absolutes? Why not think "hey, maybe there is 10% we cure aging or 10% chance we all die and these are super important"? After all isn't it obvious to everybody that probability * magnitute = impact?

Said more clearly so we don't get into a discussion about correctness of percentages:

Possible and Impossible are hugely different.

I've always felt like stating percentages only provide an illusion of rigor. They're still based on vague intuition. I also think there's nothing wrong with sometimes thinking in absolutes. In some cases it's okay to think that you have enough information to tell that something definitely will or will not happen. Hedging your bets in every single case is a lack of optimization. You need to restrict the search space to save time and energy, like how we use the cut operator in Prolog.

> Why not think "hey, maybe there is 10% we cure aging or 10% chance we all die and these are super important"?

Because "maybe there is 10%" of anything is an absolutely useless metric. It's even more useless as a projection of a future event contingent on unknowable future breakthroughs. A percentage doesn't offer any useful modelling in those scenarios.

Unless you believe in determinism or prophetic eschatology, knowing the percent chance that we "cure aging" is always a lie. The most powerful science and most accurate statistics can't read the future.

the 10% maybe is useless but isn't there a huge difference between impossible and possible?

I think there is a 75% chance this Ask HN will get popular.

Reply on news.ycombinator.com