Hacker News: Miraste

New comment by Miraste in "Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber"

Miraste — Tue, 21 Jul 2026 21:27:10 +0000

That's the only explanation that makes sense. If it was frontier but cost or compute were limiting factors, they'd release it at an obscene price for the bragging rights. Google doesn't care that much about alignment, and I don't think it's likely to be significantly different than 3.5 anyway. The only reason it would need to be soft-canceled is if it's terrible, and has to end up in a ditch like Llama 4 to avoid shareholder panic.

New comment by Miraste in "GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]"

Miraste — Fri, 10 Jul 2026 21:26:41 +0000

What you're describing is similar to how the copilot harness in vs code tracks state and previous work. These systems are being implemented, bit by bit.

New comment by Miraste in "GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]"

Miraste — Fri, 10 Jul 2026 21:23:10 +0000

At a certain level, yes, but AI is still so bad at writing that its failures are objective and easily measurable

New comment by Miraste in "Car touchscreens are cheap, not good"

Miraste — Mon, 06 Jul 2026 22:18:37 +0000

I suppose that's true. I was thinking of my car, which has a lot of buttons on the steering wheel itself. They're easy to use without looking, since you don't have to change your hand position, and I control music entirely with those.

New comment by Miraste in "Car touchscreens are cheap, not good"

Miraste — Mon, 06 Jul 2026 19:42:53 +0000

You most certainly can. That doesn't mean people do it 100% of the time.

New comment by Miraste in "Car touchscreens are cheap, not good"

Miraste — Mon, 06 Jul 2026 17:14:03 +0000

There are zero touchscreen designs that are better than (physical, tactile) buttons.

You can hit buttons without taking your eyes off the road.

You cannot do this with a touchscreen.

There's no way to design or engineer around it, it's simply the wrong tool for the job.

New comment by Miraste in "Museum of Pocket Calculating Devices"

Miraste — Thu, 21 May 2026 22:57:26 +0000

There is a wonderful teardown of it here:

https://www.vcalc.net/CU-Disassembly/

Scrolling through that makes me think it's extremely unlikely to be replicated commercially. I can't imagine how much machining all of those parts would cost.

New comment by Miraste in "The last six months in LLMs in five minutes"

Miraste — Wed, 20 May 2026 17:14:23 +0000

There's no way to prove it one way or the other. I would be willing to bet money any recent AAA game was made with AI code in it somewhere, though.

New comment by Miraste in "The last six months in LLMs in five minutes"

Miraste — Tue, 19 May 2026 15:36:42 +0000

What novel ideas are you thinking of? In my experience there are very few games with novel software engineering. New gameplay mechanics or story or art or design, sure, but they're generally built with very old and standardized patterns.

New comment by Miraste in "AMÁLIA and the future of European Portuguese LLMs"

Miraste — Mon, 11 May 2026 16:10:14 +0000

To my knowledge, all major LLMs are multilingual. This article could really have used an evaluation of existing models' European Portuguese capabilities.

New comment by Miraste in "Meta to start capturing employee mouse movements, keystrokes for AI training"

Miraste — Tue, 21 Apr 2026 21:24:21 +0000

Union busting is easy to do and hard to prove. This would act as a supporting regulation by making it more difficult. I imagine a legal framework similar to other privacy regulations: nothing about specific software or implementations, but instead new classes of data that are illegal to collect or store about your employees. There is complexity there, but something like mouse movements and keystrokes as described in the article is completely black and white.

New comment by Miraste in "Meta to start capturing employee mouse movements, keystrokes for AI training"

Miraste — Tue, 21 Apr 2026 20:50:17 +0000

While you have the right practical approach, I do believe companies should face harsh regulations preventing this kind of monitoring. It has almost universally negative effects, from enabling union-busting to exploitation to all kinds of discrimination and favoritism.

New comment by Miraste in "The beginning of scarcity in AI"

Miraste — Sun, 19 Apr 2026 03:09:11 +0000

Reliability: complete solar deployment includes some form of power storage. There are many variations, but chemical battery technology is improving the fastest, so it's gaining the most ground.

Amount of power: World solar power generation capacity is in the terawatts and rapidly increasing, there's no issue with its potential ceiling. As a bonus, it tends to work best on land that's useless for other purposes.

New comment by Miraste in "The beginning of scarcity in AI"

Miraste — Thu, 16 Apr 2026 22:25:39 +0000

If only there were some form of cheap, widely manufactured power generation technology that didn't use turbines... Are they really going to wait until 2030 to get more turbines rather than invest in solar?

New comment by Miraste in "The beginning of scarcity in AI"

Miraste — Thu, 16 Apr 2026 22:22:00 +0000

China's domestic chips are increasingly close to state-of-the-art. The US electrical grid is... not.

New comment by Miraste in "The future of everything is lies, I guess: Where do we go from here?"

Miraste — Thu, 16 Apr 2026 18:36:41 +0000

It's not at all clear whether automobiles were a net positive. They are more or less solely responsible for climate change (even emissions not directly from motor vehicles wouldn't be possible without them), which may prove to be the worst mistake in the history of technology.

New comment by Miraste in "Qwen3.6-35B-A3B: Agentic coding power, now open to all"

Miraste — Thu, 16 Apr 2026 14:51:00 +0000

What? 35B-A3B is not nearly as smart as 27B.

New comment by Miraste in "Project Glasswing: Securing critical software for the AI era"

Miraste — Tue, 07 Apr 2026 19:09:58 +0000

I can see analyzing it from a psychological perspective as a means of predicting its behavior as a useful tactic, but doing so because it may have "experiences or interests that matter morally" is either marketing, or the result of a deeply concerning culture of anthropomorphization and magical thinking.

New comment by Miraste in "Project Glasswing: Securing critical software for the AI era"

Miraste — Tue, 07 Apr 2026 19:03:52 +0000

Sure, most tech companies eliminate some form of human labor. Anthropic aims to eliminate all human labor, which is very different.

New comment by Miraste in "Project Glasswing: Securing critical software for the AI era"

Miraste — Tue, 07 Apr 2026 18:44:58 +0000

They are a for-profit company, working on a project to eliminate all human labor and take the gains for themselves, with no plan to allow for the survival of anyone who works for a living. They're definitionally not your friends. While they remain for-profit, their specific behaviors don't really matter.