still recuperating from the CTO of Bluesky saying that statistics and logic are the same thing
*https://bsky.app/profile/amyhoy.bsky.social/post/3muzbdnwckc2h
Discussion
still recuperating from the CTO of Bluesky saying that statistics and logic are the same thing
*https://bsky.app/profile/amyhoy.bsky.social/post/3muzbdnwckc2h
Logic is… what the hell.
@thomasfuchs dear god they even double down
> i am quite proud of my reply, which really gets to the nut of the problem with claiming that statistics=logic
don't let them near bayes theorem
@thomasfuchs assuming this is a about LLM.
I don't know who this person is but perhaps they are a good reason to never join blusky.
There is heuristical decision making but it's not identical to logic and LLM are not decision making. In fact a computer cant make a decision. Even swiping away this fact of our reality an LLM does not do heuristical decision making.
I understand why the argument "feels right" because the real world is messy and often we need heuristics. It's still wrong here
@saxnot I’d say statistics is deep down basically the opposite of reason, it’s guessing (more or less backed by data) instead of deduction.
For a lot of things the guesses can indeed be very good and totally fine!
But yeah as you say that’s irrelevant here.
@thomasfuchs
:blinkblink:
What do you even say, in response to something like that?
@thomasfuchs
In that case Corollary 1 says
Correlation is causation 😬🤯
@thomasfuchs there is a Venn diagrams of problems that can be solved logically and/or problems with a statistical solution. The confusion arises inside the intersection, where some people seem to be proud of using probabilistic tools even if more efficient exact methods exist.
@pautasso @thomasfuchs a probabalistic method might be more efficent.
A few years ago I worked in logistics and very often it makes no sense to solve for optimality becuse many of logistic ploblems are variants of TSP and thus can't be solved polynomically.
Instead there are heuristics which might find the optimal solution on accident but usually just produce a "good enough" result.
e. g. savings algorithm for TSP
Why is a raven like a writing-desk?
@thomasfuchs Re-enforces my rule:
If a tech company has more than 250 people, the C-Suite are all idiots and should be fired immediately.
@thomasfuchs Still not sure if I should laugh or cry...
@thomasfuchs Someone at the bluesky hq are smoking a lot
https://atproto.at/uri/at://did:plc:oky5czdrnfjpqslsw2a5iclo/app.bsky.feed.post/3mv2iclsgxk2r#preview
at://did:plc:oky5czdrnfjpqslsw2a5iclo/app.bsky.feed.post/3mv2iclsgxk2r
Post by @jay.bsky.team