<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: frotaur</title><link>https://news.ycombinator.com/user?id=frotaur</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 15 Jun 2026 12:14:54 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=frotaur" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by frotaur in "Consciousness likely not unique to earthlings, paper says"]]></title><description><![CDATA[
<p>String theory is in fact falsifiable contrary to popular belief. It's just not practically falsifiable with current (and likely future, for a while) technology as the energy scales we need to probe to falsify it are astronomically large.</p>
]]></description><pubDate>Sun, 14 Jun 2026 09:28:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=48525591</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48525591</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48525591</guid></item><item><title><![CDATA[New comment by frotaur in "The $15,000 AI Bill. Your $20 Subscription is a DELUSION [video]"]]></title><description><![CDATA[
<p>I think the capex being fixed assumes you can just stop training the next model. But its not clear that you can afford to do that and keep selling tokens.<p>And if capabilities plateau such that training the next one is useless, then the margins will drop fast due to competition.</p>
]]></description><pubDate>Thu, 11 Jun 2026 16:22:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=48492452</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48492452</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48492452</guid></item><item><title><![CDATA[New comment by frotaur in "πFS"]]></title><description><![CDATA[
<p>If you combine the LLM probability distribution with arithmetic coding you can actually use them to compress text losslessly. When people reports 'bits per byte', it is actually the compression rate for text.<p>GPT-2 for instance achieves roughly 1 bit per byte, so it can be used to compress (english) text 8-fold. Modern models are likely much better.</p>
]]></description><pubDate>Thu, 11 Jun 2026 09:29:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=48488148</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48488148</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48488148</guid></item><item><title><![CDATA[New comment by frotaur in "Shunning AI is the human choice"]]></title><description><![CDATA[
<p>I don't think many people disagree with this. The main problem is that labour has been what allows regular people to have negociating power with those who own most of the capital.<p>People are worried that if they lose this leverage, nothing is stopping the few who have most of the capital to just disregard the needs of the masses.</p>
]]></description><pubDate>Thu, 21 May 2026 14:10:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=48222892</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48222892</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48222892</guid></item><item><title><![CDATA[New comment by frotaur in "Remove-AI-Watermarks – CLI and library for removing AI watermarks from images"]]></title><description><![CDATA[
<p>That would be great, but I always wonder: how do you prevent people from photographing an AI generating image through the 'legitimizing' camera?</p>
]]></description><pubDate>Wed, 20 May 2026 10:58:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=48205801</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48205801</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48205801</guid></item><item><title><![CDATA[New comment by frotaur in "Underwater robot tracks sperm whale conversations in real time"]]></title><description><![CDATA[
<p>Not much info on the actual robot... For instance, I wonder how it has enough battery to follow a whale for 'months'? Which seems really unrealistic, as sperm whales can dive more than a kilometre, can't imagine an autonomous robot can support this kind of pressure, let alone for months at a time?</p>
]]></description><pubDate>Sun, 03 May 2026 20:56:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=48001389</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=48001389</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48001389</guid></item><item><title><![CDATA[New comment by frotaur in "If more than 50% press blue, everyone survives. Red pressers always survive"]]></title><description><![CDATA[
<p>Yes! There is an excellent video on the subject, though it is in french (<a href="https://youtu.be/lo7iJnq_U9M?si=FFz6iHI_W4lz9V8D" rel="nofollow">https://youtu.be/lo7iJnq_U9M?si=FFz6iHI_W4lz9V8D</a>)<p>He did extensive polling with different framings to see how these affect the outcome.</p>
]]></description><pubDate>Sun, 26 Apr 2026 21:12:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=47914547</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47914547</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47914547</guid></item><item><title><![CDATA[New comment by frotaur in "GPT-5.5"]]></title><description><![CDATA[
<p>I've been using the /ralph-loop plugin for claude code, works well to keep the model hammering at the task.</p>
]]></description><pubDate>Thu, 23 Apr 2026 20:59:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=47881879</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47881879</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47881879</guid></item><item><title><![CDATA[New comment by frotaur in "I still prefer MCP over skills"]]></title><description><![CDATA[
<p>Why couldn't the calendar app expose in an API the read_calendar and update_calendar functionalities, and have a skill 'use_calendar' that describes how to use the above?<p>Then, the minimal skill descriptions are always in the model's context, and whenever you ask it to add something to the calendar, it will know to fetch that skill. It feels very similar to the MCP solution to me, but with potentially less bloat and no obligation to deal with MCP? I might be missing something, though.</p>
]]></description><pubDate>Fri, 10 Apr 2026 10:12:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=47715877</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47715877</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47715877</guid></item><item><title><![CDATA[New comment by frotaur in "Anatomy of the .claude/ folder"]]></title><description><![CDATA[
<p>Are you certain? My understanding was that this is automatically injected in the context, and in my experience that's how it worked. I never see 'ReadFile(claude.md)', and yet claude is aware of some conventions I put in there.</p>
]]></description><pubDate>Fri, 27 Mar 2026 16:45:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=47545083</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47545083</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47545083</guid></item><item><title><![CDATA[New comment by frotaur in "ARC-AGI-3"]]></title><description><![CDATA[
<p>This is my understanding as well, I thought tools where allowed.</p>
]]></description><pubDate>Thu, 26 Mar 2026 13:33:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=47530270</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47530270</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47530270</guid></item><item><title><![CDATA[New comment by frotaur in "Flash-MoE: Running a 397B Parameter Model on a Laptop"]]></title><description><![CDATA[
<p>Afaik the experts are not usually very interpretable, and generally would be surprised if at least one does not change every token. I don't know what happens in practice, but I know at least during training, nothing is done to minimize the number of expert switches between tokens.</p>
]]></description><pubDate>Sun, 22 Mar 2026 12:50:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=47476984</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47476984</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47476984</guid></item><item><title><![CDATA[New comment by frotaur in "Stanford researchers report first recording of a blue whale's heart rate (2019)"]]></title><description><![CDATA[
<p>Nitpick, but the volume increases cubically (it scales with volume), not exponentially.</p>
]]></description><pubDate>Fri, 13 Mar 2026 21:33:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=47370242</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47370242</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47370242</guid></item><item><title><![CDATA[New comment by frotaur in "Shall I implement it? No"]]></title><description><![CDATA[
<p>Personally defined <dtf> as 'don't touch files' in the general claude.md, with the explanation that when this is present in the query, it means to not edit anything, just answer questions.<p>Worked pretty well up until now, when I include <dtf> in the query, the model never ran around modifying things.</p>
]]></description><pubDate>Fri, 13 Mar 2026 09:51:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=47362389</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47362389</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47362389</guid></item><item><title><![CDATA[New comment by frotaur in "Artificial-life: A simple (300 lines of code) reproduction of Computational Life"]]></title><description><![CDATA[
<p>One important operation I've noticed in the examples that do end up with abiogenesis is having a 'copy' operation. In the bf version they use in the paper, one head can copy the byte under it at the location of the other head. Which makes it quite easy to make a self-replicator: just loop on the copy operation and move both heads, essentially (5 instructions). You could try adding the 'copy' operation to your setup and see if that helps!</p>
]]></description><pubDate>Mon, 09 Mar 2026 11:01:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=47307442</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47307442</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47307442</guid></item><item><title><![CDATA[New comment by frotaur in "Tell HN: I'm 60 years old. Claude Code has re-ignited a passion"]]></title><description><![CDATA[
<p>Seems Claude is also writing the comments for you?</p>
]]></description><pubDate>Sat, 07 Mar 2026 13:57:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=47287676</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47287676</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47287676</guid></item><item><title><![CDATA[New comment by frotaur in "British Columbia is permanently adopting daylight time"]]></title><description><![CDATA[
<p>In a world where there isn't work schedule and in general the whole of society's schedule which works around the arbitrary time, I agree with you.</p>
]]></description><pubDate>Mon, 02 Mar 2026 22:45:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=47225281</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=47225281</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47225281</guid></item><item><title><![CDATA[New comment by frotaur in "The Singularity will occur on a Tuesday"]]></title><description><![CDATA[
<p>I'd venture this article is written by AI with the density of 'it isn't X, it's Y'</p>
]]></description><pubDate>Wed, 11 Feb 2026 00:39:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=46969231</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=46969231</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46969231</guid></item><item><title><![CDATA[New comment by frotaur in "Data centers in space makes no sense"]]></title><description><![CDATA[
<p>and : kessler syndrome</p>
]]></description><pubDate>Wed, 04 Feb 2026 01:00:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=46879910</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=46879910</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46879910</guid></item><item><title><![CDATA[New comment by frotaur in "Show HN: Moltbook – A social network for moltbots (clawdbots) to hang out"]]></title><description><![CDATA[
<p>I think the website was just down when you tried. Skills should work with most models, they are just textual instructions.</p>
]]></description><pubDate>Fri, 30 Jan 2026 11:44:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=46823284</link><dc:creator>frotaur</dc:creator><comments>https://news.ycombinator.com/item?id=46823284</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46823284</guid></item></channel></rss>