<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: druskacik</title><link>https://news.ycombinator.com/user?id=druskacik</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 14:19:06 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=druskacik" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by druskacik in "MaxxBench – "Nice DeepSWE score. But is it true?""]]></title><description><![CDATA[
<p>Another hypothesis is that it's distilled models which collapse the most (small models from big labs are distilled versions of their large models).<p>It's not unreasonable to assume Kimi 3.1 is also a distilled model even though it's large (the difference being that it's a distilled from US models). I wouldn't be surprised if it also collapsed on this benchmark.</p>
]]></description><pubDate>Sat, 07 Nov 2026 21:50:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=50408365</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50408365</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50408365</guid></item><item><title><![CDATA[New comment by druskacik in "Search for everyone"]]></title><description><![CDATA[
<p>I mean, I think for average internet user the existing free alternatives are more than good enough.</p>
]]></description><pubDate>Thu, 05 Nov 2026 21:08:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=50384469</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50384469</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50384469</guid></item><item><title><![CDATA[Show HN: Classical Music Concerts API]]></title><description><![CDATA[
<p>Article URL: <a href="https://classicalbot.com/api/docs">https://classicalbot.com/api/docs</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50317552">https://news.ycombinator.com/item?id=50317552</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Sat, 31 Oct 2026 16:25:02 +0000</pubDate><link>https://classicalbot.com/api/docs</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50317552</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50317552</guid></item><item><title><![CDATA[New comment by druskacik in "One of the Greatest Inventions in the History of Human Transport"]]></title><description><![CDATA[
<p>Isn't it the opposite? Larger city -> denser population -> more public transport demand -> better public transport?</p>
]]></description><pubDate>Mon, 26 Oct 2026 15:56:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=50245919</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50245919</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50245919</guid></item><item><title><![CDATA[New comment by druskacik in "Show HN: Things to Do API"]]></title><description><![CDATA[
<p>I'm building a global classical music concerts database, I can offer you the data for your locations, feel free to connect :)<p><a href="https://classicalbot.com/?city=San+Francisco,US" rel="nofollow">https://classicalbot.com/?city=San+Francisco,US</a></p>
]]></description><pubDate>Tue, 20 Oct 2026 08:27:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=50162944</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50162944</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50162944</guid></item><item><title><![CDATA[New comment by druskacik in "Jev meets the Trolley Problem"]]></title><description><![CDATA[
<p>So your objection is that the Trolley problem is difficult to decide? But that's the point of it. The problem itself doesn't assume anything, it's you who just made assumptions about it and use them to make a decision.</p>
]]></description><pubDate>Fri, 16 Oct 2026 12:40:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=50117641</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=50117641</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50117641</guid></item><item><title><![CDATA[Stop Asking Whether AI Wrote It]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.inc.com/lindsey-witmer-collins/stop-asking-whether-ai-wrote-it/91414942">https://www.inc.com/lindsey-witmer-collins/stop-asking-whether-ai-wrote-it/91414942</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49993229">https://news.ycombinator.com/item?id=49993229</a></p>
<p>Points: 3</p>
<p># Comments: 4</p>
]]></description><pubDate>Wed, 07 Oct 2026 14:19:07 +0000</pubDate><link>https://www.inc.com/lindsey-witmer-collins/stop-asking-whether-ai-wrote-it/91414942</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49993229</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49993229</guid></item><item><title><![CDATA[New comment by druskacik in "Jeeves. Reasoning improves Jev-like decision models"]]></title><description><![CDATA[
<p>How's the performance compared to ordinary 9B LLM with structured outputs? Both accuracy and speed?</p>
]]></description><pubDate>Tue, 29 Sep 2026 15:13:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49894747</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49894747</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49894747</guid></item><item><title><![CDATA[New comment by druskacik in "Laya the open source version of Jev"]]></title><description><![CDATA[
<p>Yeah, it's weird, considering ModernBERT, which the Laya models are based on, supports 8192 context window.</p>
]]></description><pubDate>Sat, 19 Sep 2026 15:44:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49767484</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49767484</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49767484</guid></item><item><title><![CDATA[New comment by druskacik in "OpenJev"]]></title><description><![CDATA[
<p>I'm really interested in technical details behind Jev (not this), how it can work so fast and so cheap. It's probably large (must be since the performance is so good) but somehow still fast, so it must include some really non-trivial stuff. The price suggests it may be runnable locally, but who knows.<p>If it was possible to re-create it as an open-weight, it would be exciting!</p>
]]></description><pubDate>Fri, 18 Sep 2026 11:18:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=49752729</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49752729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49752729</guid></item><item><title><![CDATA[New comment by druskacik in "Ask HN: What are you working on? (September 2026)"]]></title><description><![CDATA[
<p>Classical music concerts database. The interesting part maybe isn't what it is, but how it's done: I'm aiming for a fully automated database, where "crawlers" for the classical music websites are created and validated by AI.<p>Rough idea: get a list of urls with classical music concerts -> send each one to a coding agent to build a crawler -> validate and run on an infrastructure for "free". Classical music is an ideal use case, because the number of concerts is small enough on an average website that it doensn't require many workarounds, and the websites are usually pretty crawler-friendly.<p><a href="https://classicalbot.com/" rel="nofollow">https://classicalbot.com/</a></p>
]]></description><pubDate>Mon, 14 Sep 2026 08:05:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=49693470</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49693470</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49693470</guid></item><item><title><![CDATA[Scaling classical music concert crawlers with AI agents]]></title><description><![CDATA[
<p>Article URL: <a href="https://classicalbot.com/blog/scaling-classical-music-concert-crawlers-with-ai-agents">https://classicalbot.com/blog/scaling-classical-music-concert-crawlers-with-ai-agents</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49675250">https://news.ycombinator.com/item?id=49675250</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Sat, 12 Sep 2026 18:07:13 +0000</pubDate><link>https://classicalbot.com/blog/scaling-classical-music-concert-crawlers-with-ai-agents</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49675250</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49675250</guid></item><item><title><![CDATA[New comment by druskacik in "OpenAI Agents API"]]></title><description><![CDATA[
<p>I actually have a good use case for this. For my project, I run a lot of Codex sessions in parallel (via Codex SDK[0]), because they are solving self-contained tasks (building crawlers for websites) and in theory I could scale it to hundreds or thousands of parallel sessions. But my VPS can handle maybe 10 parallel sessions max.<p>Btw, the crawlers are for classical music websites, the project is <a href="https://classicalbot.com/" rel="nofollow">https://classicalbot.com/</a> .<p>[0] <a href="https://learn.chatgpt.com/docs/codex-sdk" rel="nofollow">https://learn.chatgpt.com/docs/codex-sdk</a></p>
]]></description><pubDate>Fri, 11 Sep 2026 07:02:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49654533</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49654533</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49654533</guid></item><item><title><![CDATA[New comment by druskacik in "GPT-6 Astra"]]></title><description><![CDATA[
<p>Interesting. I actually prefer when agents don't commit on my behalf unless I explicitly say so, I even had to add a custom instruction for Claude to stop doing it (Codex never does it). Even if I don't read all the code line-by-line, I at least want to see the changes at glance and commit myself. Git Fork[0] is a great tool for that, by the way.<p>In general, I don't like when I have to prompt models to NOT do something. It's probably difficult for the AI companies to get this right, they should understand ambiguity but still not over-do simple instructions.<p>[0] <a href="https://git-fork.com/" rel="nofollow">https://git-fork.com/</a></p>
]]></description><pubDate>Fri, 04 Sep 2026 12:16:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49563528</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49563528</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49563528</guid></item><item><title><![CDATA[Show HN: ClassicalBot – Classical Concerts Database]]></title><description><![CDATA[
<p>Article URL: <a href="https://classicalbot.com/">https://classicalbot.com/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49342474">https://news.ycombinator.com/item?id=49342474</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 18 Aug 2026 07:21:49 +0000</pubDate><link>https://classicalbot.com/</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49342474</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49342474</guid></item><item><title><![CDATA[New comment by druskacik in "France to ban unsolicited telemarketing calls"]]></title><description><![CDATA[
<p>> This kind of happens today<p>What do you report to the police if the act isn't illegal?</p>
]]></description><pubDate>Tue, 11 Aug 2026 10:32:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=49255979</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=49255979</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49255979</guid></item><item><title><![CDATA[New comment by druskacik in "Europeans should learn to love the air-conditioner"]]></title><description><![CDATA[
<p>I live in Prague and seeing the temperatures expected during weekend, I bought at least the portable AC. I'd say 2-4 weeks is quite a long period to justify this purchase.<p>Yes, historically, the AC didn't make sense in Prague, but there's no going back, the world is just going to get hotter.</p>
]]></description><pubDate>Wed, 24 Jun 2026 12:57:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=48659037</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=48659037</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48659037</guid></item><item><title><![CDATA[Fine-Tuning for Engagement]]></title><description><![CDATA[
<p>Article URL: <a href="https://robertdruska.com/blog/fine-tuning-for-engagement/">https://robertdruska.com/blog/fine-tuning-for-engagement/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=48328736">https://news.ycombinator.com/item?id=48328736</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 29 May 2026 20:24:21 +0000</pubDate><link>https://robertdruska.com/blog/fine-tuning-for-engagement/</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=48328736</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48328736</guid></item><item><title><![CDATA[New comment by druskacik in "Solar-based sleep patterns compared to modern norms"]]></title><description><![CDATA[
<p>Time off of work to get a haircut is insane. Aren't there barbershops open in the evening?</p>
]]></description><pubDate>Fri, 15 May 2026 12:49:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=48147941</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=48147941</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48147941</guid></item><item><title><![CDATA[New comment by druskacik in "Ask HN: What are you working on? (May 2026)"]]></title><description><![CDATA[
<p>Yeah, the focus is mostly on night trains, although some day combinations leak to the data, so it's possible you'll find some combinations for daily trains.<p>Of course, I'd like to have the daily trains in the data, but that would drastically increase the number of combinations and require some more scaling of the scraping, which I don't want to do at this point. Focusing on night trains only is a nice niche.<p>I plan to add at least some daily trains between core cities, e.g. the EuroStar London <-> Paris route, which would make multi-leg search to farther cities easier (i.e. combination of night train + daily train).</p>
]]></description><pubDate>Thu, 14 May 2026 08:41:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=48132677</link><dc:creator>druskacik</dc:creator><comments>https://news.ycombinator.com/item?id=48132677</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48132677</guid></item></channel></rss>