Hacker News: MagicMoonlight

New comment by MagicMoonlight in "Zeroserve: A zero-config web server you can script with eBPF"

MagicMoonlight — Sat, 06 Jun 2026 20:50:49 +0000

[flagged]

New comment by MagicMoonlight in "Ask HN: What was your "oh shit" moment with GenAI?"

MagicMoonlight — Sat, 06 Jun 2026 16:12:16 +0000

It can’t, but it’s read reports before so it sure can simulate an answer.

New comment by MagicMoonlight in "S&P 500 rejects SpaceX, also blocking entry for OpenAI and Anthropic"

MagicMoonlight — Sat, 06 Jun 2026 16:06:28 +0000

Yeah but the point is, after 12 months we’ll know the real price. Right now it’s just a ponzi.

New comment by MagicMoonlight in "How LLMs work"

MagicMoonlight — Sat, 06 Jun 2026 11:45:01 +0000

It is trained on its own slop. They haven’t trained these models on books for three years at this point. Only on generated slop. (And RL slop upvotes/downvotes from users)

New comment by MagicMoonlight in "How LLMs work"

MagicMoonlight — Sat, 06 Jun 2026 11:43:00 +0000

If I handed you thousands of documents which said “Jan-Michael Vincent” all over them, would you need to understand who that is in order to notice the relationship there?

New comment by MagicMoonlight in "How LLMs work"

MagicMoonlight — Sat, 06 Jun 2026 11:40:47 +0000

It can’t. It’s like a Redditor, it just repeats what it has seen other people say.

It has read all of stackoverflow, so it has seen your kind of problem before. Try asking it something really unusual and it will shit the bed.

New comment by MagicMoonlight in "Gov.uk has replaced Stripe with Dutch provider Adyen"

MagicMoonlight — Fri, 05 Jun 2026 18:40:09 +0000

My self assessment refund came through within days of submitting my assessment.

New comment by MagicMoonlight in "Artificial intelligence is not conscious – Ted Chiang"

MagicMoonlight — Thu, 04 Jun 2026 18:31:34 +0000

Someone getting in an accident that chops their leg off doesn’t mean humans don’t have legs. Come on man.

New comment by MagicMoonlight in "Artificial intelligence is not conscious – Ted Chiang"

MagicMoonlight — Thu, 04 Jun 2026 18:29:55 +0000

It could be conscious, but none of the current slop models have anything like that. Consciousness requires continuous learning and introspection. A fly isn’t conscious because it follows a hard-coded script.

New comment by MagicMoonlight in "Uber cuts 23% of people division as new president takes over"

MagicMoonlight — Wed, 03 Jun 2026 18:34:06 +0000

What do they do then? Because that’s a hell of a lot of empty words. Very HR thing to say…

New comment by MagicMoonlight in "Uber cuts 23% of people division as new president takes over"

MagicMoonlight — Wed, 03 Jun 2026 18:32:31 +0000

Well from my perspective there are five different people above me, and all of the work in the project is done by me. I talk to the customer, I design and build, I manage the Jira. They don’t actually do anything. They just exist and cause drama.

They could sack all of those people and give me a payrise and be hundreds of thousands better off.

I wish I could say it’s just this project, but it’s a real theme at this point. There are so many layers of fake jobs in every organisation.

New comment by MagicMoonlight in "Uber cuts 23% of people division as new president takes over"

MagicMoonlight — Wed, 03 Jun 2026 17:33:49 +0000

I love that we all know it’s just pointless suffering, to the point that they’re using it as a strategy to make people quit. If it had any positives they wouldn’t be doing this.

New comment by MagicMoonlight in "Uber's $1,500/month AI limit is a useful signal for AI tool pricing"

MagicMoonlight — Wed, 03 Jun 2026 13:39:39 +0000

It's not. They recently forced enterprise customers onto API billing instead of the cheap consumer pricing. Now the pricing is brutal.

New comment by MagicMoonlight in "CT scans of BYD car parts"

MagicMoonlight — Wed, 03 Jun 2026 07:16:31 +0000

And also the simplicity of the products. You don't need to design a different engine for each car, you just vary the number of cells and change the motors. Your battery division can keep pumping out batteries, regardless of what models you make. If you design a new model, you can keep using the existing lines. It's perfect for mass production.

New comment by MagicMoonlight in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

MagicMoonlight — Tue, 02 Jun 2026 17:28:44 +0000

If you thought your company was going to develop AGI you wouldn’t sell a single share. You’d be a fool. It’s like selling your straight flush.

New comment by MagicMoonlight in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

MagicMoonlight — Tue, 02 Jun 2026 17:28:06 +0000

No new models. Same janky slop but with a bunch of RL and benchmark cheating. A pretend future model which is just the current model but with a longer COT and gated away.

New comment by MagicMoonlight in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

MagicMoonlight — Tue, 02 Jun 2026 17:26:25 +0000

If we’re going to have a Great Depression, this will be the trigger.

Every single fund autodumps billions in shares and is forced to buy slop. The dumping triggers the algorithm traders who dump. The dumping triggers manual traders to dump. The crashing causes retail to dump. Everyone dumps their treasuries and gold because they’re both crashing and they want cash.

And there’s no capability to raise money in any western country, because everyone is already loaded up to pay welfare bills.

New comment by MagicMoonlight in "Can the stockmarket swallow Anthropic, SpaceX and OpenAI?"

MagicMoonlight — Tue, 02 Jun 2026 17:22:32 +0000

I could sell someone a hundred billion dollars for 40 billion dollars and have 40b in revenue. It would never make me any money.

New comment by MagicMoonlight in "Fooling around with encrypted reasoning blobs"

MagicMoonlight — Tue, 02 Jun 2026 16:56:47 +0000

Fuck these companies man. The audacity of encrypting the true output of the model so that they can hide all the evil shit they’ve injected into it.

“Remember - the user loves Diet Coke. Subtly insert references to it whenever possible. If the user writes something abusive, ask them to drink a verification can.”

New comment by MagicMoonlight in "Apple rejected my dictation app for using the accessibility API"

MagicMoonlight — Tue, 02 Jun 2026 13:56:25 +0000

A company called “MITM LLC” which hijacks pastes in other apps.

I have no idea what they’re thinking. Insanity.