Hacker News: mountainriver

New comment by mountainriver in "Unified Memory, Explained: Why Mini PCs Can Run 70B Models a Big GPU Can't"

mountainriver — Fri, 10 Jul 2026 14:50:50 +0000

Yeah this is such a funny thing going around. Try to run or train a small/medium sized model on a MacBook. It doesn’t go very good compared to a dedicated gpu

This is likely the right path in the future but it isn’t there yet today

New comment by mountainriver in "Java 27: What's New?"

mountainriver — Fri, 10 Jul 2026 13:12:02 +0000

The best thing to ever happen to the software world was Oracle buying Sun and consequently Java.

Java was great at its time but then had to go. Oracle gaining control over them sped up this process.

Poor Android got caught up and now half our phones don’t work

New comment by mountainriver in "Java 27: What's New?"

mountainriver — Fri, 10 Jul 2026 13:07:32 +0000

Do they have backwards compatibility guarantees?

Otherwise what are we doing here?

New comment by mountainriver in "Ben Bernanke Joins Anthropic Oversight Trust"

mountainriver — Fri, 10 Jul 2026 04:19:06 +0000

Except Anthropic has delivered a truly world changing product…

New comment by mountainriver in "Interview with Mitchell Hashimoto about Ghostty and Zig"

mountainriver — Thu, 09 Jul 2026 23:25:18 +0000

I feel this way about most Hashi tools, they just seem massively overrated to me.

Ghostty is fine I guess, I find it to be way buggier than iterm with a fraction of the features.

Zig is fine, has some cool stuff, the community seems roughly the same as the rust, with again just way less features.

The rest of the hashi tools are fine, I don’t really use any of them anymore. Vault was a big deal at some point I guess

New comment by mountainriver in "GPT-5.6"

mountainriver — Thu, 09 Jul 2026 17:31:43 +0000

There was just a study showing that when presented blindly no one could tell the difference yet users were avid they could

New comment by mountainriver in "GPT-5.6"

mountainriver — Thu, 09 Jul 2026 17:30:26 +0000

Agree, I think there was just a blind study that showed no one could tell the difference even though the users were avid they could

New comment by mountainriver in "Small AI Models Gain Traction In places with unreliable networks"

mountainriver — Tue, 07 Jul 2026 05:29:00 +0000

I looked into this a bit but unfortunately because of starlink most of this won’t be needed

New comment by mountainriver in "Giant trees have no trouble pumping water to top branches: new research"

mountainriver — Sat, 04 Jul 2026 04:39:15 +0000

With a lot of software getting eaten up I’m increasingly interested in biology. Seems like one of the later frontiers that could have massive benefits, and AI is really well suited to help us understand it.

New comment by mountainriver in "Right to Local Intelligence"

mountainriver — Fri, 03 Jul 2026 04:11:24 +0000

The new frontier! I love it

New comment by mountainriver in "Reduce GVisor Cold Starts with GPU Snapshotting"

mountainriver — Wed, 01 Jul 2026 17:03:41 +0000

How does this compare to the CRIU work? Or does it use that under the hood?

New comment by mountainriver in "Trump made more than $1B on crypto deals, part of 2025 windfall"

mountainriver — Wed, 01 Jul 2026 02:09:30 +0000

Corruption is what kills countries.

Our inability to stop this is a very bad sign about our future.

New comment by mountainriver in "AI boom risks global financial crash, warn central bankers"

mountainriver — Tue, 30 Jun 2026 19:02:46 +0000

You are really comparing this to Watson??

If you can fully automate software you are fully missing the point of everything you can do with that and how valuable it is

New comment by mountainriver in "AI boom risks global financial crash, warn central bankers"

mountainriver — Mon, 29 Jun 2026 02:08:49 +0000

Meanwhile AI has gotten so good it can just about one shot a SaaS app.

I’m not worried about it…

New comment by mountainriver in "Ford AI hiccups push carmaker to rehire ‘gray beard’ inspectors"

mountainriver — Sun, 28 Jun 2026 20:49:23 +0000

Fully red pilled, I literally cannot believe the rhetoric I’ve been hearing inside major companies.

New comment by mountainriver in "Turn your site into a place people can bump into each other"

mountainriver — Sun, 28 Jun 2026 02:16:28 +0000

I love the idea but all I’m seeing is people saying really explicative things. Hopefully there are better ways of moderating

New comment by mountainriver in "Don't trust large context windows"

mountainriver — Sun, 14 Jun 2026 14:09:23 +0000

“Memory” systems are a way for developers to feel like they are contributing to AI

New comment by mountainriver in "Shall we play a game? My AI nuclear simulation"

mountainriver — Sat, 13 Jun 2026 22:12:34 +0000

If we want to train a model excessively on SVGs it will obviously be able to do this. We have only just started trying to do that

New comment by mountainriver in "Shall we play a game? – LLMs use tactical nukes in 95% of simulations"

mountainriver — Thu, 11 Jun 2026 23:13:17 +0000

I can’t believe anyone still thinks this given their unbelievable ability to write code.

Self awareness? Probably not. Intelligence? You would have to be high to think that’s not the case.

People are feeling threatened, and rightfully so. LLMs are already insanely intelligent and continue to improve

New comment by mountainriver in "Shall we play a game? My AI nuclear simulation"

mountainriver — Thu, 11 Jun 2026 23:10:57 +0000

Intelligence is the ability to compress information. World models are just one aspect of that