Hacker News: skybrian

New comment by skybrian in "I Am Not a Reverse Centaur"

skybrian — Fri, 12 Jun 2026 21:38:13 +0000

Because doing enough security reviews to remove all the security bugs gets expensive, and if you use well-reviewed code, it's already been done.

New comment by skybrian in "Why the AI Renaissance Keeps Not Arriving"

skybrian — Fri, 12 Jun 2026 21:35:41 +0000

Doing things the same, standard way each time is often good. Most of the code we write is obvious. We notice it when it's a weird quirk rather than just being the most straightforward way.

I wonder how long it will take to fix the quirks?

New comment by skybrian in "I Am Not a Reverse Centaur"

skybrian — Fri, 12 Jun 2026 19:31:27 +0000

I think this makes sense for apps, but the apps will still need infrastructure and common protocols to interoperate. It still won’t make sense to implement your own cryptography.

New comment by skybrian in "Claude Fable is relentlessly proactive"

skybrian — Fri, 12 Jun 2026 02:23:13 +0000

We do have ways to avoid giving an LLM any secrets, but it needs to be the simple, default solution.

New comment by skybrian in "Claude Fable is relentlessly proactive"

skybrian — Fri, 12 Jun 2026 02:19:27 +0000

There are plenty of good sandboxes out there but somehow no "obvious right answer" that everyone knows to recommend. Seems like a missed opportunity.

(I'm happy with exe.dev, but I'm not sure what I'd use if I were coding on a Mac.)

New comment by skybrian in "Doing nothing at work"

skybrian — Thu, 11 Jun 2026 18:35:09 +0000

This sounds a lot like being on call. If you like that kind of work, why not be an SRE instead? Maybe there also need to be people “on call” for responding to other kinds of events, though?

It also sounds a lot like getting pulled into meetings. People complain about it, but sometimes that’s the job.

New comment by skybrian in "Waymo Premier"

skybrian — Thu, 11 Jun 2026 18:23:13 +0000

I assume it’s for business travelers.

New comment by skybrian in "The Abundance Illusion"

skybrian — Wed, 10 Jun 2026 23:44:15 +0000

Looking at nationwide statistics is also a bit misleading because nobody is average. Under that scenario, oil consumers are exposed and oil producers will win big.

New comment by skybrian in "Anthropic requires 30 day data retention for Fable and Mythos"

skybrian — Wed, 10 Jun 2026 23:37:26 +0000

Yes, it certainly is an odd situation when some people believe you cannot use Mythos-class models because security while others believe you must do code reviews with Mythos-class models because security.

New comment by skybrian in "The Abundance Illusion"

skybrian — Wed, 10 Jun 2026 20:47:41 +0000

> The hard work Carter asked for — building the physical capacity to never need the buffer — was quietly abandoned.

But since then, US natural gas production doubled [1], and solar power is growing exponentially [2]. Leaving that out of the history seems excessively gloomy.

[1] https://www.eia.gov/dnav/ng/hist/n9050us2a.htm [2] https://www.statista.com/statistics/183447/us-energy-generat...

New comment by skybrian in "I'm Eric Ries, author of "The Lean Startup" and new book "Incorruptible" – AMA"

skybrian — Wed, 10 Jun 2026 18:16:44 +0000

Do you think their dispute with the Department of Defense would have gone the same way? We didn't see that at OpenAI or Google.

(Although it does remind me a bit of Google pulling out of China back in the day.)

New comment by skybrian in "DiffusionGemma: 4x Faster Text Generation"

skybrian — Wed, 10 Jun 2026 18:06:00 +0000

Could you say more about how you use it? What does your workflow look like?

New comment by skybrian in "Building an HTML-first site doubled our users overnight"

skybrian — Wed, 10 Jun 2026 15:25:26 +0000

What do people like for form validation?

In this article he recommends the “validation-enhancer” library:

https://www.npmjs.com/package/validation-enhancer

I’ve also seen one called “formisch” that the author of valibot is working on:

https://github.com/open-circle/formisch

They’re both pretty new. Has anyone tried them?

New comment by skybrian in "Rich Sutton on AI creativity and discovery"

skybrian — Wed, 10 Jun 2026 05:48:17 +0000

He's saying that pre-training an LLM alone can't do it, but if you run an LLM in a loop with tools (like any coding agent) then it can. Also, the technique his group came up with should be used more:

> This is the weakness of deep learning that is alleviated with a new algorithm that my group presented in Nature a couple of years ago. Our “continual backpropagation” made one small change: every so often a less-used neuron would be re-initialized to small random weights. This allows the variation to continue and plasticity to be retained.

Here's the paper: https://www.nature.com/articles/s41586-024-07711-7

It has a fair number of citations, but I haven't looked into how much it's used.

New comment by skybrian in "Albania Is Not for Sale: Kushner's $4B Resort Triggers'Flamingo Revolution'"

skybrian — Wed, 10 Jun 2026 00:11:18 +0000

Many Americans didn't vote for Trump at all, so to the extent this is true, it still can't be accurately summarized as "American voters are fine with being robbed blind." Maybe some still are, but they don't speak for everyone.

New comment by skybrian in "Albania Is Not for Sale: Kushner's $4B Resort Triggers'Flamingo Revolution'"

skybrian — Tue, 09 Jun 2026 20:41:37 +0000

They didn't vote for a war with Iran.

New comment by skybrian in "Test-case reducers are underappreciated debugging tools"

skybrian — Tue, 09 Jun 2026 20:39:43 +0000

Property-based testing frameworks will often do test case reduction as well (called shrinking).

New comment by skybrian in "Albania Is Not for Sale: Kushner's $4B Resort Triggers'Flamingo Revolution'"

skybrian — Tue, 09 Jun 2026 16:36:26 +0000

Americans aren’t “fine with it.” Trump’s approval ratings are very low.

https://www.reuters.com/business/energy/trump-approval-stays...

New comment by skybrian in "Replies to comments on my "LLMs are eroding my career" post"

skybrian — Tue, 09 Jun 2026 04:18:13 +0000

No, that was not "effectively what happened" in the Industrial Revolution. That was an enormous change, but it didn't "completely eliminate the need for a human workforce." That's just hype.

New comment by skybrian in "We Think the SpaceX IPO Is Overvalued"

skybrian — Tue, 09 Jun 2026 04:16:15 +0000

Would it make any sense for it to orbit next to a Starlink satellite?