Hacker News: lblume

New comment by lblume in "FBI is investigating Minnesota Signal chats tracking ICE"

lblume — Wed, 28 Jan 2026 10:57:30 +0000

Well, Signal would have to disclose the salt of course.

New comment by lblume in "Find a pub that needs you"

lblume — Thu, 15 Jan 2026 12:31:22 +0000

Why not?

New comment by lblume in "enclose.horse"

lblume — Wed, 07 Jan 2026 07:42:15 +0000

For me at least, no. Making money by training a model from user data on such a game seems like a perfectly fine thing to do.

New comment by lblume in "Total monthly number of StackOverflow questions over time"

lblume — Sun, 04 Jan 2026 10:42:00 +0000

It has often been claimed, and even shown, that training LLMs on their own outputs will degrade the quality over time. I myself find it likely that on well-measurable domains, RLVR improvements will dominate "slop" decreases in capability when training new models.

New comment by lblume in "Show HN: Stop AI scrapers from hammering your self-hosted blog (using porn)"

lblume — Fri, 19 Dec 2025 11:31:16 +0000

From what I can tell: nothing, it's just that they currently do not.

New comment by lblume in "How elites could shape mass preferences as AI reduces persuasion costs"

lblume — Fri, 05 Dec 2025 10:11:16 +0000

On average, Gen Z uses 5 hours of social media per day in the U.S. (3-4 hours in other Western countries). I would refrain from calling this "alright".

New comment by lblume in "How to create accessible PDFs from the start"

lblume — Wed, 12 Nov 2025 20:15:15 +0000

Sure. But it also helps humans, and I'd guess currently more so.

New comment by lblume in "Valdi – A cross-platform UI framework that delivers native performance"

lblume — Sun, 09 Nov 2025 14:11:36 +0000

It is simply inaccessible to anyone not using the platform. You need to create an account and join the community/"server" to see anything posted there. You cannot find anything by using a search engine and are completely unable to export anything for local use.

New comment by lblume in "Heroin addicts often seem normal"

lblume — Sun, 12 Oct 2025 07:55:00 +0000

The cheapness is due to the prevalence, and the prevalence of sugar caused sweetness receptors to be evolutionarily advantageous. There is no world in which sugar is extremely expensive, markets still function basically in the way they do now and humans experience the sensation of sweetness the way they typically do now. Cocaine and other types of "hard" drugs are qualitatively different in that regard.

Your example also doesn't really hold up because people typically don't process cocaine in the way they do with sugar and other carbohydrates. In your hypothetical scenario, we might see people consuming large amounts of pure sugar (or artificial sweeteners), but they wouldn't go to lengths of baking bread using it.

New comment by lblume in "Meta Superintelligence Labs' first paper is about RAG"

lblume — Sun, 12 Oct 2025 03:26:53 +0000

Another commenter claims the latter: https://news.ycombinator.com/item?id=45554169

New comment by lblume in "Meta Superintelligence Labs' first paper is about RAG"

lblume — Sun, 12 Oct 2025 03:24:28 +0000

Sure, but they weren't hired as managers, right?

New comment by lblume in "Daniel Kahneman opted for assisted suicide in Switzerland"

lblume — Sat, 11 Oct 2025 12:03:47 +0000

I agree with your legal assessment and still think of the case as very interesting. The article explicitly talks about how any such decision could have only been premature, for the slow cognitive decline is typically only noticed when it is too late, and because the change is continuous, there can be no good commitment to "I no longer consider this life worthwhile once condition X is no longer satisfied".

New comment by lblume in "A small number of samples can poison LLMs of any size"

lblume — Thu, 09 Oct 2025 20:22:25 +0000

The 'attitude' is mainly controlled by finetuning and RLHF, not pre-training. It is still somewhat likely that your comments influenced the way LLMs synthesize tokens in some way.

New comment by lblume in "A small number of samples can poison LLMs of any size"

lblume — Thu, 09 Oct 2025 20:18:03 +0000

There will always be some string that doesn't really predictably occur in other documents, is just some current name. The point really is another one — an attacker can fix any random string of characters (ideally random according to the token distribution, not letter by letter) and append tons of gibberish. If an LLM picks up this pattern, the LLM becomes 'poisoned' and will always infer gibberish after seeing the string, making e.g. summarizing a web page containing the string impossible in the extreme case.

New comment by lblume in "The least amount of CSS for a decent looking site (2023)"

lblume — Tue, 07 Oct 2025 06:46:52 +0000

The matching is very permissive, and the example just works: https://codepen.io/leo848blume/pen/RNrppdj

New comment by lblume in "How functional programming shaped and twisted front end development"

lblume — Sat, 04 Oct 2025 21:50:13 +0000

Reduce can be very useful to signal that the state used is inherently limited. My rule of thumb is to use reduce when the state is a primitive or composed of at most two primitives, and a for loop otherwise. What counts as "primitive" depends on the language of choice and abstraction level of the program, of course.

New comment by lblume in "How functional programming shaped and twisted front end development"

lblume — Sat, 04 Oct 2025 21:47:15 +0000

list(map(f, ...)) should almost always be replaced with [f(x) for x in ...] though.

New comment by lblume in "PEP 810 – Explicit lazy imports"

lblume — Fri, 03 Oct 2025 20:27:22 +0000

uv is good.

New comment by lblume in "Talent Is Alignment"

lblume — Fri, 03 Oct 2025 10:07:09 +0000

> Forcing kids to spend time on something is an effective way to spoil it for them.

I generally agree. However most school systems force students to learn basic math, which is generally used for testing whether the student is able to think structurally and follow simple rules. Although I agree intrinsic motivation to be extremely valuable, not doing math with children the way school intends may just lead to worse grades and outcomes.

New comment by lblume in "Kagi News"

lblume — Tue, 30 Sep 2025 20:38:55 +0000

Can verify. Settings > Content Filter > Custom Keywords. You can choose whether to hide the stories completely or blur them instead.