Email or username:

Password:

Forgot your password?
Eniko | Kitsune Tails out now!'s posts Post Back to profile
Top-level
david_chisnall

@eniko You can get more than 50% accuracy predicting the weather by predicting whatever the weather was yesterday. You can get more than 50% accuracy predicting the weather by predicting whatever the weather was this day last year. Getting to be 80% accurate, in contrast, is really hard and requires actually modelling how the atmosphere works.

Probabilistic techniques are great for rapidly getting to a kind-of okay plateau.

4 November at 8:50 | Wall-to-wall | Open on infosec.exchange
6 comments
Sky Leite

@david_chisnall @eniko Interesting, that’s also how rollback netcode for online games works. When you don’t have the other players inputs for a given frame due to network conditions you just assume they’re holding the same inputs as last frame and a lot of the time that turns out to be correct

4 November at 12:35 | Open on social.coop
Locksmith

@sky @david_chisnall @eniko

But when it fails, we have those very weird warps.

4 November at 15:37 | Open on infosec.exchange
Sky Leite

@locksmithprime Yeah. When the prediction fails is when the game has to re-simulate the frames since then, to figure out the correct state of the world, which is the rollback.

4 November at 15:44 | Open on social.coop
Kim Spence-Jones 🇬🇧😷

@david_chisnall @eniko My grandfather was a meteorologist, and he often quoted that fact. In meteorology, it even has a name: “persistence of type”.

4 November at 12:42 | Open on mastodon.social
Varx

@david_chisnall @eniko this tradeoff is why I think the best use in the near future for LLMs will turn out to be voice controls a la "star trek: the next generation".

The captain can ask for simple, commonplace high level commands and be confident the computer will do it mostly correctly. And that's a huge time saver. But for anything that requires any amount of precision, or technical knowledge, you better believe he's asking Geordi La Forge (or maybe a real AGI like Data).

4 November at 15:34 | Open on defcon.social
david_chisnall

@varx @eniko

It often isn't faster. I can't find it, but about 20 years ago I read a paper that did exactly that experiment. They set up voice controls that were really just a person doing the thing for you. They had one group do the tasks using conventional GUIs and the other do the command via the voice command. They subtracted out the time the assistant spent working in the second case. It was still both slower and more error-prone for most tasks. The outliers were very simple things.

Watch any Star Trek episode where someone talks to the computer and listen carefully to what they ask. Aside from Captain 'I don't know how to configure my replicator presets' Picard asking for 'Tea, Earl Grey, Hot', pretty much everything is so ambiguous that it works only because there's a script writer and not a computer deciding what Majel Barrett says.

@varx @eniko

It often isn't faster. I can't find it, but about 20 years ago I read a paper that did exactly that experiment. They set up voice controls that were really just a person doing the thing for you. They had one group do the tasks using conventional GUIs and the other do the command via the voice command. They subtracted out the time the assistant spent working in the second case. It was still both slower and more error-prone for most tasks. The outliers were very simple things.

4 November at 15:42 | Open on infosec.exchange
Go Up