Hacker News: geraneum

New comment by geraneum in "Show HN: Specification First Agentic Development Plugin and Prove"

geraneum — Thu, 29 Oct 2026 07:30:24 +0000

> Don't take it too seriously.

I didn’t, hence the question.

New comment by geraneum in "OpenAI's agents talked to each other through the CPU Cache"

geraneum — Thu, 29 Oct 2026 06:32:24 +0000

> How would they have both agreed to use the cache though

I think it’s safe to assume that extracting information using timing in cache is already known in training set, since we have the likes of Specter.

New comment by geraneum in "StackOverflow now gets fewer than 1000 questions a month"

geraneum — Thu, 29 Oct 2026 06:23:15 +0000

Maybe they visited the website and the existing questions were enough? These stats seems to be questions asked. Maybe that’s the reason.

New comment by geraneum in "Evidence of a spying operation by ICE against left-wing social movements"

geraneum — Wed, 28 Oct 2026 14:03:23 +0000

It’s so cute that some people imagine the armed forces and police were on the other side or are on their side. If you’re average Joe, the police in a tyrannical state are never on your “side”.

New comment by geraneum in "The Exponential Time Hypothesis Is False [pdf]"

geraneum — Wed, 28 Oct 2026 10:00:56 +0000

Do you foresee that AI will be able to count Rs in strawberry by then?

Jokes aside, by then we're gonna have StarTrek Replicators. Exciting!

New comment by geraneum in "The Exponential Time Hypothesis Is False [pdf]"

geraneum — Wed, 28 Oct 2026 06:13:16 +0000

So when they release those model, we all will be solving millennium prize problems?

New comment by geraneum in "Show HN: Specification First Agentic Development Plugin and Prove"

geraneum — Tue, 27 Oct 2026 21:48:04 +0000

But people do unpaid painting, horseback riding, calligraphy, etc. which are outdated because they love it. Why is it so unfortunate for you?

New comment by geraneum in "Show HN: Specification First Agentic Development Plugin and Prove"

geraneum — Tue, 27 Oct 2026 21:11:11 +0000

> And unfortunately, that works pretty well.

Why unfortunately?

New comment by geraneum in "Three new discoveries in mathematics, computer science, and physics"

geraneum — Tue, 27 Oct 2026 20:24:54 +0000

There’s no such thing as a read/write head in transformers. You can use it as a metaphor but that’s a leaky metaphor, hence all these misunderstandings.

> In fact it has N read / write heads equal to the size of the context

And what is the asymptotic notation when computation scales linearly with the size of context?

> Either way, the simple truth is that you can train a transformer system

Which still doesn’t mean they have a read/write head or have O(1) complexity for token computation with “infinite context”.

New comment by geraneum in "AI promised less work. I've never worked more"

geraneum — Tue, 27 Oct 2026 15:55:46 +0000

> If the volume of work goes up, its cost goes down. Supply and demand.

What?! Good to know. When contractor comes to do the kitchen, I will make them do the garden as well and will reduce the money I’d pay them, because supply and demand!

> Unless you think you should pay more now for more content

We do already! You realize you’re paying by the token? More token -> bigger bill!

New comment by geraneum in "Three new discoveries in mathematics, computer science, and physics"

geraneum — Tue, 27 Oct 2026 15:01:38 +0000

> Same idea for that transformer

It’s not the same for transformers, especially not with the theoretical infinite context. It simply doesn’t process tokens in O(1) in your given setup.

In practice, with frontier LLMs, it’s of course more complicated, and also they don’t reveal how they do lookups, but still the theoretical limit applies regardless.

Do you mind elaborating or perhaps providing some reference as to why you think transformers should be similar to a Turing machine on this specific criteria?

New comment by geraneum in "AI promised less work. I've never worked more"

geraneum — Tue, 27 Oct 2026 13:56:29 +0000

So are you saying they should be paying more for more work but they are not?

New comment by geraneum in "AI promised less work. I've never worked more"

geraneum — Tue, 27 Oct 2026 13:23:57 +0000

> employers would ever pay you the same amount of money to do less work?

Shouldn’t they now pay more for more work?

New comment by geraneum in "The Exponential Time Hypothesis Is False [pdf]"

geraneum — Tue, 27 Oct 2026 09:22:06 +0000

Then anyone with access to their api’s could’ve come up with these already.

New comment by geraneum in "Three new discoveries in mathematics, computer science, and physics"

geraneum — Tue, 27 Oct 2026 09:08:47 +0000

Any basic lookup table is turing complete if you bolt an external infinite tape and read/write head onto it. But then that’s moving the goal post since the context wouldn’t be the “limitless tape” and it’s not a transformer “in the normal way” because turing completeness would be coming from external io and not from auto regressive self attention. Different topic.

New comment by geraneum in "Microsoft offers mass buyouts for roughly a quarter of its workforce"

geraneum — Tue, 27 Oct 2026 07:21:26 +0000

> but have stopped hiring for pretty much 4 consecutive years since.

What’s your theory?

New comment by geraneum in "Microsoft offers mass buyouts for roughly a quarter of its workforce"

geraneum — Tue, 27 Oct 2026 07:10:29 +0000

Right now, a hiccup in Chinese production of a certain IC may make buying toilet paper harder in Germany! It’s naive to think that such drastic changes (mass unemployment) will not be extremely turbulent (to put it mildly).

New comment by geraneum in "Three new discoveries in mathematics, computer science, and physics"

geraneum — Tue, 27 Oct 2026 07:05:30 +0000

Of course it is. You’re conflating a turing machine with a transformer. Transformers don’t have a local read/write head. The [self] attention is over the whole context, which makes each step O(N) at best (with caching) and O(N^2) w/o.

New comment by geraneum in "Three new discoveries in mathematics, computer science, and physics"

geraneum — Tue, 27 Oct 2026 06:22:00 +0000

The context acting as a "theoretically limitless tape" while simultaneously the step being O(1) are contradictory. You cannot have it both.

New comment by geraneum in "Microsoft offers mass buyouts for roughly a quarter of its workforce"

geraneum — Tue, 27 Oct 2026 05:55:54 +0000

> History shows us that the more free market an economy is, the more prosperous it is.

Ad infinitum? No it doesn’t.

> Prices for software have gone to zero.

On earth?

> Software engineers routinely become millionaires, and some billionaires.

They don’t but if they did, this could mean the opposite of what you think it means.