Hacker News: lou1306

New comment by lou1306 in "The European Social Stack"

lou1306 — Sat, 20 Jun 2026 17:02:43 +0000

Perhaps relevant context: The EU commission just ignored the "Tech Sovereignty Package" it launched ~3 weeks ago, and explicitly referred open-source as a core element of their strategy, and endorsed W, another ATproto-based social that recently a) closed their code and b) ...had its CEO attend Davos. Make of that what you will.

New comment by lou1306 in "There are no instances in ATProto"

lou1306 — Sat, 20 Jun 2026 09:56:28 +0000

Not just all BSky traffic: all traffic from every ATProto service

New comment by lou1306 in "Can Europe train a frontier AI model on the compute it owns?"

lou1306 — Mon, 15 Jun 2026 21:00:46 +0000

Are we really discussing a plan that has never been attempted before (training a frontier model on federated hardware) and that would require coordination on a continental scale, sketched on a five-page PDF on GitHub with no discernible author or affiliation? a PDF that, I am sorry to point out, reeks of AI prose in more than one passage?

What are we even doing here.

New comment by lou1306 in "What happened to nerds?"

lou1306 — Mon, 15 Jun 2026 12:06:04 +0000

The "good lawyer" is/was a major archetype in modern Western culture. See "The Devil and Daniel Webster", "To Kill a Mockingbird", "Paths of Glory", just to name a few examples.

New comment by lou1306 in "UK Brings in Full Social Media Ban for Under-16s"

lou1306 — Mon, 15 Jun 2026 09:38:36 +0000

Good. This will force kids to be creative, hack around, and question authority.

New comment by lou1306 in "Anti-social: It's fads, not friends, which now dominate social media feeds"

lou1306 — Tue, 09 Jun 2026 09:21:10 +0000

Well the size of IJ is itself a statement: 1. there's an awful lot to unpack once you start exploring the intersection of capitalism, addiction, and entertainment; 2. can you even _manage_ to go through it all without switching to mindless consumption every 5 minutes?

New comment by lou1306 in "Meta confirms 1000s of Instagram accounts were hacked by abusing its AI chatbot"

lou1306 — Sat, 06 Jun 2026 22:21:55 +0000

If the backend function was so poorly coded to allow such a gargantuan security hole, then it is an even worse problem. Basically Meta is throwing its own engineers under the bus so that its AI chatbot can save face. Scary stuff.

Unless the backend was _also_ vibe-coded, in which case it is still an AI problem.

New comment by lou1306 in "Meta confirms 1000s of Instagram accounts were hacked by abusing its AI chatbot"

lou1306 — Sat, 06 Jun 2026 22:18:39 +0000

You joke but this is almost literally what Chain-of-Thought does, at least in the early days. They basically just added "Wait," to the model's output and fed it back to the model iirc

New comment by lou1306 in "Meta launches Instagram, Facebook, and WhatsApp subscriptions"

lou1306 — Mon, 01 Jun 2026 12:33:48 +0000

Well people _used_ to pay for it, in fact. Except it was 3 EUR/year rather than 3 EUR/month...

New comment by lou1306 in "Why are large language models so terrible at video games?"

lou1306 — Mon, 01 Jun 2026 12:08:46 +0000

To be honest, Zork at times makes precious little sense: you are supposed to die over and over before you figure stuff out. For instance, you have to grab the endless-light-source treasure very early on, or you mathematically cannot win. And the game does not spell anything out for you, you just have to "get it" by watching closely at how/why you die.

This is a tall order for an LLM: it needs a lot of context but most of the context will be just noise.

New comment by lou1306 in "Two Ways to Draw Infinite Jest's Sierpinski Gasket"

lou1306 — Mon, 01 Jun 2026 09:42:35 +0000

I think the order in which the different elements of the book are introduced is crucial, as it leads to a lot of "aha!" moments.

> What put me off it is it just kind of reads like a rambling stoner conversation.

Yeah well, that book may not be your cup of tea then. The book _is_ rambling, plus a lot of the characters _are_ actual stoners/addicts/recovering addicts. But keep in mind that most of the book is in the third person, not in the first (as the first few pages would make you assume).

New comment by lou1306 in "I am retiring from tech to live offline"

lou1306 — Fri, 29 May 2026 15:19:04 +0000

If things get as much as 10% as chaotic as you predict, there will be massive turmoil and the "Destroy all Clankers" party will actually be the one in charge.

New comment by lou1306 in "Microsoft starts canceling Claude Code licenses"

lou1306 — Sat, 23 May 2026 13:41:44 +0000

This has really little to do with embrace and extend. They are not taking over an open standard or anything like that.

If anything, it's forced dogfooding, i.e., forcing their own workforce to beta-test their product.

New comment by lou1306 in "New arXiv policy: 1-year ban for hallucinated references"

lou1306 — Fri, 15 May 2026 14:17:33 +0000

I don't think they mention it in the CoC, but anyone can upload an ArXiv preprint listing literally anybody else as a coauthor, so it is only logical that only "confirmed" coauthors should be affected.

New comment by lou1306 in "New arXiv policy: 1-year ban for hallucinated references"

lou1306 — Fri, 15 May 2026 08:53:45 +0000

As I mentioned in another thread:

To be a coauthor on a preprint that you have not submitted, you have to actively "claim" it (using a password given to the author who submitted). It's on you to double-check before claiming.

I surely hope that only "confirmed" coauthors will get the ban, it's only logical.

New comment by lou1306 in "New arXiv policy: 1-year ban for hallucinated references"

lou1306 — Fri, 15 May 2026 08:51:34 +0000

To be a coauthor on a preprint that you have not submitted, you have to actively "claim" it (using a password given to the author who submitted). It's on you to double-check before claiming.

New comment by lou1306 in "New arXiv policy: 1-year ban for hallucinated references"

lou1306 — Fri, 15 May 2026 08:49:16 +0000

> some people don't like LLMs

It's not even that they "don't like LLMs". They just don't like academic fraud! If references were fabricated with a Markov chain it would be just as bad!

New comment by lou1306 in "Flickr: The first and last great photo platform"

lou1306 — Sun, 26 Apr 2026 09:19:47 +0000

EXIF data enabled some pretty cool things, like PhotoSynth [1]. This stuff looks futuristic even now, ~20 years later, basically what the Metaverse could/should have been.

[1] https://www.ted.com/talks/blaise_aguera_y_arcas_how_photosyn...

New comment by lou1306 in "There Will Be a Scientific Theory of Deep Learning"

lou1306 — Sat, 25 Apr 2026 11:33:30 +0000

But most "normal" neural networks are feed-forward, so they are guaranteed to terminate in a bounded amount of time. This rules Turing completeness right out. And even recurrent NNs can be "unfolded" into feed-forward equivalents, so they are not TC either.

You need a memory element the network can interact with, just like an ALU by itself is not TC, but a barebones stateful CPU (ALU + registers) is.

New comment by lou1306 in "Why Not Venus?"

lou1306 — Fri, 24 Apr 2026 10:37:53 +0000

> Now imagine trying to land on that airship from orbit or get back into orbit (and beyond) from that airship. None of this is easy here on Earth.

Also, if anything goes wrong on Earth and you're in the atmosphere, there's still a chance you might soar to lower altitudes, eject, parachute, and get rescued. On Venus it's a death sentence.