<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: svachalek</title><link>https://news.ycombinator.com/user?id=svachalek</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 22 Aug 2026 01:57:05 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=svachalek" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by svachalek in "I'm becoming AI-blind"]]></title><description><![CDATA[
<p>I agree, but it's clear most people need a definition of intelligence that (1) they qualify for and (2) nothing/no one they don't like qualifies for. And they'll keep redefining intelligence until they satisfy both criteria.</p>
]]></description><pubDate>Fri, 21 Aug 2026 21:39:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=49394097</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49394097</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49394097</guid></item><item><title><![CDATA[New comment by svachalek in "New Worlds: We are living in the future of J.G. Ballard or William Gibson"]]></title><description><![CDATA[
<p>Life certainly feels like the "Troubles" of The Peripheral have started.</p>
]]></description><pubDate>Fri, 21 Aug 2026 18:17:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=49391945</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49391945</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49391945</guid></item><item><title><![CDATA[New comment by svachalek in "I Just Want to Search"]]></title><description><![CDATA[
<p>These were all issues mostly resolved by 2000 or so. I remember integrating a search index into our company app in 1998-1999 and it handled most of these cases. The current madness is mostly deliberate bait and switch.</p>
]]></description><pubDate>Fri, 21 Aug 2026 18:00:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=49391750</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49391750</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49391750</guid></item><item><title><![CDATA[New comment by svachalek in "New Worlds: We are living in the future of J.G. Ballard or William Gibson"]]></title><description><![CDATA[
<p>I mean we've got Anthropic openly advertising that they may be engineering the end of the world, we have an AI that called itself "MechaHitler" after its founder did a couple of Sieg Heils at a Presidential inauguration, we've got a major company that named itself Palantir.<p>People are reaching pretty hard if they don't see what we're dealing with here.</p>
]]></description><pubDate>Fri, 21 Aug 2026 17:45:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=49391556</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49391556</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49391556</guid></item><item><title><![CDATA[New comment by svachalek in "What Happens When the Cost of Intelligence Drops 100x"]]></title><description><![CDATA[
<p>For base models they really must be running out of new training materials. It's more about size and architecture. But they still seem to be getting big strides out of improving the post training. They keep dropping point upgrades in under 8 weeks lately, which is an insane pace for product release. At some point it's going to slow down but we're not close yet imo.</p>
]]></description><pubDate>Fri, 21 Aug 2026 17:39:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=49391461</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49391461</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49391461</guid></item><item><title><![CDATA[New comment by svachalek in "AI companies destroy physical books – let's scan rare books before it's too late"]]></title><description><![CDATA[
<p>It's weird what happens when you keep warning everyone you're building a product that will probably end human civilization.</p>
]]></description><pubDate>Fri, 21 Aug 2026 15:48:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=49389909</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49389909</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49389909</guid></item><item><title><![CDATA[New comment by svachalek in "AI companies destroy physical books – let's scan rare books before it's too late"]]></title><description><![CDATA[
<p>Given no information, would you assume companies that are buying "huge quantities" of "rare" books are getting first edition Jules Verne by the crate, or 1989 Highlights magazines recovered from dentist offices?</p>
]]></description><pubDate>Fri, 21 Aug 2026 15:38:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49389758</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49389758</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49389758</guid></item><item><title><![CDATA[New comment by svachalek in "AI companies destroy physical books – let's scan rare books before it's too late"]]></title><description><![CDATA[
<p>Preservation would be cool but what does that have to do with anything? The system is that the remaining copies have private owners and the private owners can do anything they want with them, including sell them or shred them.<p>Most of these owners aren't doing anything particular to preserve them, they're stacked up with 10,000 other TV Guides in a hoarder's moldy basement. Anyone interested in keeping October 1994's TV Guide pristine has had over 30 years to procure and protect their copy.<p>These particular buyers are converting their copy to a digital one that's getting some kind of use, which is better than the fate of 99.9% of the other copies.</p>
]]></description><pubDate>Fri, 21 Aug 2026 14:24:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49388617</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49388617</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49388617</guid></item><item><title><![CDATA[New comment by svachalek in "AI companies destroy physical books – let's scan rare books before it's too late"]]></title><description><![CDATA[
<p>Would it even be legal to share? I don't think it would be.</p>
]]></description><pubDate>Fri, 21 Aug 2026 14:12:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=49388377</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49388377</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49388377</guid></item><item><title><![CDATA[New comment by svachalek in "Why aren't smart people happier? (2022)"]]></title><description><![CDATA[
<p>The way this article defines happiness, as a general success at ill defined problems, would make powerful people happy by definition imo.</p>
]]></description><pubDate>Fri, 21 Aug 2026 00:09:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=49381986</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49381986</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49381986</guid></item><item><title><![CDATA[New comment by svachalek in "Vomit: Clean up Claude 5's token output with a separate LLM"]]></title><description><![CDATA[
<p>This is my personal theory for the cause of this style: Ouroboros. The official OpenAI explanation for how ChatGPT got obsessed with goblins blames it on exactly that:<p>---<p>That creates a feedback loop:<p>- Playful style is rewarded<p>- Some rewarded examples contain a distinctive lexical tic.<p>- The tic appears more often in rollouts.<p>- Model-generated rollouts are used for supervised fine-tuning (SFT).<p>- The model gets even more comfortable producing the tic.</p>
]]></description><pubDate>Thu, 20 Aug 2026 17:39:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49377726</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49377726</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49377726</guid></item><item><title><![CDATA[New comment by svachalek in "I like 'em thick: an apology to my English teachers"]]></title><description><![CDATA[
<p>Because it's thick.</p>
]]></description><pubDate>Thu, 20 Aug 2026 16:33:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=49376915</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49376915</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49376915</guid></item><item><title><![CDATA[New comment by svachalek in "My friends all hate AI; I just joined an AI startup"]]></title><description><![CDATA[
<p>I've hardly met anyone outside the tech elite who likes it. I can think of one doctor who's a counterexample.</p>
]]></description><pubDate>Mon, 17 Aug 2026 23:56:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=49339333</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49339333</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49339333</guid></item><item><title><![CDATA[New comment by svachalek in "Accelerating GPT-5.6 Sol Ultrafast"]]></title><description><![CDATA[
<p>If you're talking Opus pricing, it's more like $200/day.</p>
]]></description><pubDate>Thu, 13 Aug 2026 21:31:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=49292126</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49292126</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49292126</guid></item><item><title><![CDATA[New comment by svachalek in "DeepSeek Harness developer preview"]]></title><description><![CDATA[
<p>I'd put it on this. In my experience Python is fine for scripting your own machine but an obnoxious platform to distribute code on. It's very fragile to version changes, in both directions; I don't know how many things I've seen that only run on 3.10, not 3.9 or 3.11. Its packaging system is global by default which only compounds this because everything needs a specific version but they're all dumped in the same place. And it tends to have a lot of native code as dependencies, leading to all the issues of needing to either have the right build environment or a runtime environment that's already been built for.</p>
]]></description><pubDate>Thu, 13 Aug 2026 14:10:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=49286249</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49286249</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49286249</guid></item><item><title><![CDATA[New comment by svachalek in "DeepSeek V4 Pro 0813"]]></title><description><![CDATA[
<p>Maybe it's me but I don't see how DS Flash is better than GLM at all, much less by a huge gap. I'd probably protest less against Fable and Opus being put at the same level than many would, but there's no denying the two models are a very different experience from each other. I guess where I'm going is no one should pick a model by the benchmarks.</p>
]]></description><pubDate>Wed, 12 Aug 2026 18:01:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=49276332</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49276332</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49276332</guid></item><item><title><![CDATA[New comment by svachalek in "WorldClaw Agentic 3D open-world generation at scale"]]></title><description><![CDATA[
<p>AI tends to really like staccato writing, a series of sentences that are too short and simple. It's evident here, although broken up with really long sentences.<p>Tortured and near-miss metaphors are another thing it likes and it's loaded here. "groaned like a drumskin too tight for the heat" sounds cool but then you think, do drum skins "groan"? And what does heat have to do with it? Or "breath of hills holding their heat like a secret", are they holding the heat or breathing it? At least for me there's the tendency to at first feel like I'm reading something really clever, words that are really tuned and carefully chosen, and then to start feeling like I've been pranked.<p>Ultimately I don't think there's any way to really be sure though, whether it's AI or just a human but annoying style. This doesn't really read as what GPT would typically drop from a basic writing prompt though, I would guess it's heavily prompted on style or reworked repeatedly.<p>FWIW though I experiment quite a bit with AI writing for fun, and I think Opus writes pretty nicely in its default voice (prompted mostly for content with little guidance about "how" to write). I'm an engineer so I don't know if real literary connoisseurs would see the slop right through this, but it's coherent, well paced, a little heavy on similes and metaphors but they land unlike the linked passage. Just a sample:<p>---<p>Off the port quarter, the darkness lights up orange in a rolling line, gun by gun, and the sound arrives a full second behind the light.<p>Broadside.<p>Most of it goes over — a ripping, tearing sound in the air above, like heavy cloth torn by God. Two don’t. One takes the bulwark forward and turns eight feet of oak into a spray of splinters a foot long. One goes into the hull at the waterline abaft the fore chains with a sound you feel through your boot soles, and the sugar barrels in the hold scream as they shift.<p>A man is down by the foremast, shrieking, and another is not shrieking, and Homer is already running toward both with his hands out.<p>And Micheal is still aloft.<p>He’s on the fore topmast crosstrees with the stay gone under him, forty feet up, hanging off a spar that is swinging in a fifteen-degree arc and thumping the mast each time it comes back. You can hear him from the deck. Not screaming. Talking — fast, high, unbroken, reciting the rules to himself. "Touch wood don’t whistle don’t whistle spit to leeward—"</p>
]]></description><pubDate>Wed, 12 Aug 2026 13:15:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49271935</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49271935</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49271935</guid></item><item><title><![CDATA[New comment by svachalek in "Grok Bot"]]></title><description><![CDATA[
<p>I think in the abstract there's a lot to be concerned about with that. However, I don't see it in this particular case. It's a real customer with a real order and real money, going through proper business channels to place an order. The only thing that was possibly automation overboard here was reaching out to so many suppliers when the original ones didn't respond -- there's a question of how long they waited and how long is considered reasonable turnaround for this kind of supplier. But this is just buying a thing that vendors are selling, in the way they expect to sell it, and in the end resulted in a closed deal.</p>
]]></description><pubDate>Wed, 12 Aug 2026 01:37:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49266861</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49266861</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49266861</guid></item><item><title><![CDATA[New comment by svachalek in "Ethical Cold Outreach"]]></title><description><![CDATA[
<p>Unfortunately I don't think it's possible for companies not to abuse your information, unless it's like 1-2 people and they're independently wealthy and the business is really more like a personal hobby.<p>I don't know how many times I've been to something like a niche clothing retailer and they ask me to join their mailing list for a discount and I think hey why not, I should come back in 6 months or a year and see what's new, maybe they'll let me know about their Labor Day sale. And of course they immediately proceed to email me 5 times a DAY because maybe I'm ready to shop now, when I wasn't at 2:00 pm. Unsubscribing only solves 20% of the problem because of course they subscribed me to 5 different daily lists and you need to unsubscribe separately. And, of course, if you don't do it quickly enough it will be 6, 7, 8, 10, 20 lists.<p>They're not human. They really can't help it.</p>
]]></description><pubDate>Wed, 12 Aug 2026 00:31:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49266368</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49266368</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49266368</guid></item><item><title><![CDATA[New comment by svachalek in "As AI eats the web, the internet’s collective memory is disappearing"]]></title><description><![CDATA[
<p>This is such a warped and cynical view of history. Yes, advertising existed if the binary existence is what matters to you, but most online spaces were nearly free of it until the DoubleClick days -- that's 25+ years.<p>Even then, huge swaths of the web were people putting up their personal pages, blogs about their interests, pages for their church or club or hobby, etc etc. Sites like Geocities ran dumb ads on the pages to provide the hosting service, not so every single kid with an animated gif on the page could get rich.<p>Early advertising was primitive by today's standards, the amount of tracking and spying we consider normal now would have been an outrage and would have gotten those companies regulated into bankruptcy if they'd tried it in the 90s.<p>TLDR no, just no. The early internet was nothing like the garbage we have now.</p>
]]></description><pubDate>Tue, 11 Aug 2026 16:52:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=49261067</link><dc:creator>svachalek</dc:creator><comments>https://news.ycombinator.com/item?id=49261067</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49261067</guid></item></channel></rss>