Hacker News: thrill

New comment by thrill in "Dallas Fed: 30% of housing cost increase driven by unauthorized immigration [pdf]"

thrill — Sun, 21 Jun 2026 00:01:22 +0000

If only builders could figure out some way to consolidate multiple living units under single roof, with shared building systems. Maybe something in the vertical dimension?

New comment by thrill in "Now You Don't: When Espionage Meets Magic"

thrill — Sat, 20 Jun 2026 20:03:47 +0000

What the hell is that popover?

New comment by thrill in "David Sacks on Anthropic export control"

thrill — Mon, 15 Jun 2026 15:48:01 +0000

"even a small jailbreak should cause them to pull back and fix it first, right"

You do realize that LLMs are summarizations of vast numbers of weights, don't you? You don't "fix" a weight and suddenly everything is alright. You can only probe constantly in a vast space and see if the results you can command matter or not.

New comment by thrill in "Apple Foundation Models"

thrill — Mon, 15 Jun 2026 15:40:50 +0000

Yes. Fable is much more organized and consistent at taking small bites of the (sorry) apple when solving a problem. Specifically I'm talking about a machine learning problem I'd been working on for awhile with Opus and it was (and is, again) constantly stating that all the signal is exploited, everything is now overfit, etc, etc, etc. The first day I pointed Fable at the situation I got a 10% improvement by paying attention to the little details that Opus instead took slightly negative results and extrapolated to "fully exploited". I've had to drop back, again, to forcing Opus to explain what it's looked at and the detail it has quietly assumed away.

It's like the difference to talking to two smartest kids in a class, but one really belongs a grade higher - and the other hasn't learned yet to ask the questions that encourage it to dig in that little bit more for the additional multi-order effects.

New comment by thrill in "Amazon CEO's talks with U.S. officials triggered crackdown on Anthropic models"

thrill — Sat, 13 Jun 2026 20:47:13 +0000

People designated that school as a valid target - using fancy calculators does not remove that the pass/fail rests with people. AI models have no agency. Even if they are given autonomy - it is given.

New comment by thrill in "US Government directive to suspend access to Fable 5 and Mythos 5"

thrill — Sat, 13 Jun 2026 01:49:42 +0000

Typical admin move here - give our foreign competitors as much time to catch up as possible.

New comment by thrill in "Cybersecurity researchers aren't happy about the guardrails on Anthropic's Fable"

thrill — Wed, 10 Jun 2026 23:06:09 +0000

The thing triggered on a generic white paper I'd stored in a virtual cell competion from last year when I asked it to refer to the paper while working on a rather vanilla data science problem in a different domain . A little frustrating, and in my opinion more than a little pointless in total.

New comment by thrill in "Why isn't the U.S. better at soccer?"

thrill — Sun, 07 Jun 2026 20:32:37 +0000

He asks if it's an MLS problem but doesn't dig into the poor management decisions that persist at the MLS level, and though he does touch on the lack of relegation, and he doesn't touch at all on the US Soccer Federation's consistently poor choices for USMNT head coach. We're presently stuck with a guy who praises the players getting into constant on-field fights and each manager constantly makes nepotistic selections for team slots who are under-performers yet constantly praised, or even worse this year, who seem to be chosen for their unfocused non-game-enhancing aggression. That sort of mis-focus might work if playing in leagues where that is the norm, but world-class players and teams play with technique, and the rough play BS lasts as long as the officials allow it and doesn't win in the long term, which is where you have to aim to win championships. It's arguable if we'd even be in the Cup this time if we weren't hosting, and I'll be surprised if we get out of the group stage this time.

New comment by thrill in "NSA using Anthropic's Mythos for cyber attacks"

thrill — Thu, 04 Jun 2026 21:30:08 +0000

That headline is doing a lot of heavy making-it-up summarization.

New comment by thrill in "Congress Wants You to Pay $130 a Year Just to Drive an Electric Car"

thrill — Tue, 19 May 2026 19:36:03 +0000

18 wheeler type trucks do over 80% of the damage to roads. They could pay for it all and we'd all simply share the cost in the price of goods, and collection would be vastly simpler and cheaper.

But, there wouldn't be the opportunity for asking for political favors, so don't expect anyone who likes you having to beg to champion such a process.

New comment by thrill in "Pope Leo XIV’s first encyclical Magnifica humanitas to be published May 25"

thrill — Tue, 19 May 2026 00:24:45 +0000

I bet it will be 100 AI written, with guidance, natch, just because...

New comment by thrill in "We let AIs run radio stations"

thrill — Mon, 18 May 2026 22:39:51 +0000

Tossing turkeys out of a helicopter?

New comment by thrill in "I tried to make Claude make me money on open-source bounties"

thrill — Sun, 17 May 2026 00:44:42 +0000

That's kind of like saying no one should participate in Kaggle contests because we all have access to the same datasets.

New comment by thrill in "The AI Backlash Could Get Ugly"

thrill — Wed, 13 May 2026 15:48:48 +0000

Plenty of people certainly seem to have desires to make it ugly.

New comment by thrill in "Setting up a free *.city.state.us locality domain (2025)"

thrill — Wed, 13 May 2026 15:46:01 +0000

Aren’t there several states that have the same city name repeated within the state? I think there’d need to be a county delineator here too.

New comment by thrill in "US Government releases first batch of UAP documents and videos"

thrill — Fri, 08 May 2026 15:39:32 +0000

The Files That Must Not Be Released have not been released - oh look a party balloon floating by!

New comment by thrill in "245TB Micron 6600 ION Data Center SSD Now Shipping"

thrill — Wed, 06 May 2026 19:05:51 +0000

Yes, this why drives come in the mail pre-wiped - it saves on shipping.

New comment by thrill in "White House Considers Vetting A.I. Models Before They Are Released"

thrill — Mon, 04 May 2026 19:59:09 +0000

Sure, let’s kill what little lead the US AI industry has while the rest of the world kicks ass - it’s working so well in all our other endeavors.

New comment by thrill in "Let's Buy Spirit Air"

thrill — Mon, 04 May 2026 00:32:06 +0000

Let’s not.

New comment by thrill in "AI chatbots could be making you stupider"

thrill — Mon, 20 Apr 2026 17:34:28 +0000

The constant whining about them certainly is.