<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: inciampati</title><link>https://news.ycombinator.com/user?id=inciampati</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sun, 24 May 2026 23:22:44 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=inciampati" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by inciampati in "Memory has grown to nearly two-thirds of AI chip component costs"]]></title><description><![CDATA[
<p>Memory makes computation universal.</p>
]]></description><pubDate>Sun, 24 May 2026 21:18:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=48261079</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48261079</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48261079</guid></item><item><title><![CDATA[New comment by inciampati in "A Wayland Compositor in Minecraft"]]></title><description><![CDATA[
<p>Finally, I can escape to paradise and work remote.</p>
]]></description><pubDate>Fri, 22 May 2026 21:57:44 +0000</pubDate><link>https://news.ycombinator.com/item?id=48242193</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48242193</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48242193</guid></item><item><title><![CDATA[New comment by inciampati in "An OpenAI model has disproved a central conjecture in discrete geometry"]]></title><description><![CDATA[
<p>I am also using these models to accelerate scientific discovery. Yes, they are making all the difference at the frontier. At least, they feel they are. The messy thing is that we still need to communicate with each other and that's not getting dramatically faster or better. As you note the models need to be built so they do more work to participate in our communication economy. Or we will do so much, alone, to get nowhere fast because so much of our behavior is still bound up in old (good, tested, but clunky) ways of building shared knowledge.</p>
]]></description><pubDate>Thu, 21 May 2026 03:45:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=48217629</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48217629</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48217629</guid></item><item><title><![CDATA[New comment by inciampati in "I’ve joined Anthropic"]]></title><description><![CDATA[
<p>Just remember that the will do clever but useless things to improve. Like changing the random seed as per autoresearch's hero image. lol! imo, out of the box thinking is needed.</p>
]]></description><pubDate>Thu, 21 May 2026 00:01:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=48216034</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48216034</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48216034</guid></item><item><title><![CDATA[New comment by inciampati in "Zerostack – A Unix-inspired coding agent written in pure Rust"]]></title><description><![CDATA[
<p>> Integrated Ralph Wiggum loops: looping capabilities for long-horizon tasks<p>Imo, this shouldn't be embedded in the executor layer. Orchestration should handle this.</p>
]]></description><pubDate>Sun, 17 May 2026 00:30:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=48165044</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48165044</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48165044</guid></item><item><title><![CDATA[New comment by inciampati in "DeepClaude – Claude Code agent loop with DeepSeek V4 Pro"]]></title><description><![CDATA[
<p>poorly vibe coded. machines can check details easily, use them.</p>
]]></description><pubDate>Mon, 04 May 2026 00:35:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=48003194</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=48003194</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48003194</guid></item><item><title><![CDATA[New comment by inciampati in "Anthropic takes $5B from Amazon and pledges $100B in cloud spending in return"]]></title><description><![CDATA[
<p>Didn't OAI just try that 18 months ago?</p>
]]></description><pubDate>Tue, 21 Apr 2026 15:04:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=47849850</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47849850</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47849850</guid></item><item><title><![CDATA[New comment by inciampati in "Anthropic takes $5B from Amazon and pledges $100B in cloud spending in return"]]></title><description><![CDATA[
<p>They do approximate any function... within the range they're trained on. And that range is human limited, at least today.</p>
]]></description><pubDate>Tue, 21 Apr 2026 14:52:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=47849662</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47849662</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47849662</guid></item><item><title><![CDATA[New comment by inciampati in "Anthropic takes $5B from Amazon and pledges $100B in cloud spending in return"]]></title><description><![CDATA[
<p>I'm already living in this future. In a decent execution framework, with context management, memory via unix, and mechanisms for web search and access, local models are effectively on par with frontier ones. And they can often be much faster. I'll keep paying fees for the AI companies until they stop truly subsidizing and leading. They are getting close to the edge of utility, but we can use their services now to bootstrap their own demise. Long live running your own software on your own computer.</p>
]]></description><pubDate>Tue, 21 Apr 2026 14:50:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=47849623</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47849623</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47849623</guid></item><item><title><![CDATA[New comment by inciampati in "Global warming has accelerated significantly"]]></title><description><![CDATA[
<p>I prefer to think of it as a reference <i>in</i> the Torment Nexus.</p>
]]></description><pubDate>Fri, 06 Mar 2026 15:25:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=47276088</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47276088</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47276088</guid></item><item><title><![CDATA[New comment by inciampati in "What AI coding costs you"]]></title><description><![CDATA[
<p>get the system to build a clean architecture and explain it to you. it will help it to build a better system. a huge part of working with these models for engineering is getting them to create reports. for themselves and of course for us to read and understand. the bottleneck is actually our verification capacity.</p>
]]></description><pubDate>Sat, 28 Feb 2026 14:44:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=47195938</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47195938</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47195938</guid></item><item><title><![CDATA[New comment by inciampati in "Don't trust AI agents"]]></title><description><![CDATA[
<p>Lines of code are nothing. It's verification that creates value.</p>
]]></description><pubDate>Sat, 28 Feb 2026 14:43:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=47195922</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=47195922</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47195922</guid></item><item><title><![CDATA[New comment by inciampati in "Canada's deal with China signals it is serious about shift from US"]]></title><description><![CDATA[
<p>If you don't service the debt, your assets will be repossessed and sold off.</p>
]]></description><pubDate>Sat, 17 Jan 2026 20:20:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=46661655</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46661655</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46661655</guid></item><item><title><![CDATA[New comment by inciampati in "Always bet on text (2014)"]]></title><description><![CDATA[
<p>But a graph, which provides a view at a certain level of resolution, can often be described in a few consise statements. That's why we make them, to get a view we can condense.</p>
]]></description><pubDate>Sat, 27 Dec 2025 11:40:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=46401095</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46401095</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46401095</guid></item><item><title><![CDATA[New comment by inciampati in "Always bet on text (2014)"]]></title><description><![CDATA[
<p>> CAD. Sure, you can do that in text... but it takes much more room to do and magnitudes more time to interpret.<p>Fascinating example for me. I do CAD... using text! My only experience with it is programmatic in openscad. We check the visualization, but only on output of the final product. For me it's dramatically easier to work with. That may be a personal defect but it's also consistent. Underneath the rendering is always data, which is text, markup, but strings of fundamental data.<p>And in science it's not a stretch at all that numbers come first. I'll argue you're reaching. Today no one is drawing their numbers from experiments directly on a graph. They record them digitally. In textual form typically, and then render them visually to obtain generic understanding. But also there, in the end, your conclusions (per tradition) need to be point estimates with error bounds expressible in concise textual terms. You may obtain them from looking at images but the hard truth is numerical, digital, textual.</p>
]]></description><pubDate>Sat, 27 Dec 2025 11:35:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=46401071</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46401071</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46401071</guid></item><item><title><![CDATA[New comment by inciampati in "Always bet on text (2014)"]]></title><description><![CDATA[
<p>It is true that graphs communicate very well. But they do come from text... And in the end we need to be able to describe what we see in them in text.</p>
]]></description><pubDate>Sat, 27 Dec 2025 07:47:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=46400017</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46400017</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46400017</guid></item><item><title><![CDATA[New comment by inciampati in "NOAA deploys new generation of AI-driven global weather models"]]></title><description><![CDATA[
<p>To have a competitive code writer with ngrams you need more than to "scale up the ngrams" you need to have a corpus that includes all possible codes that someone would want to write. And at that point you'd be better off with a lossless full text index like an r-index. But, the lack of any generalizability in this approach, coupled with its markovian features, will make this kind of model extremely brittle. Although, it would be efficient. You just need to somehow compute all possible language before hand. tldr; language models really are reasoning and generalizing over the domain they're trained on.</p>
]]></description><pubDate>Sat, 20 Dec 2025 16:58:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=46337549</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46337549</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46337549</guid></item><item><title><![CDATA[New comment by inciampati in "Size of Life"]]></title><description><![CDATA[
<p>Very beautiful. Love this.<p>If it helps, AFAIK (I do atomic force microscopy of DNA), DNA's height is closer to 2nm than 4.</p>
]]></description><pubDate>Wed, 10 Dec 2025 17:25:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=46220554</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46220554</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46220554</guid></item><item><title><![CDATA[New comment by inciampati in "The universal weight subspace hypothesis"]]></title><description><![CDATA[
<p>The authors study a bunch of wild low rank fine tunes and discover that they share a common... low rank! ... substructure which is itself base model dependent. Humans are (genetically) the same. You need only a handful of PCs to represent the cast majority of variation. But that's because of our shared ancestry. And maybe the same thing is going on here.</p>
]]></description><pubDate>Tue, 09 Dec 2025 04:28:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=46201226</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46201226</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46201226</guid></item><item><title><![CDATA[New comment by inciampati in "We're learning more about what Vitamin D does"]]></title><description><![CDATA[
<p>> The entire reason for skin color variations is a genetic optimization for UV absorption at specific latitudes vs sunburn risks.<p>This seems obvious but was not confirmed by genetic evidence. The rate of adaptation turns out to be much higher than can be explained by skin cancer.<p>The real cause appears to be fertility. UV radiation breaks down folate (vitamin B9) in the bloodstream, and folate is critical for DNA synthesis and repair. Folate deficiency causes serious problems in pregnancy, neural tube defects like spina bifida, and may impair sperm production. So darker skin in high-UV equatorial regions likely evolved partly to protect reproductive capacity.<p>In the other direction, lower melanin production helps with vitamin D synthesis in lower sunlight environments. Vitamin D requires UV-B radiation to be synthesized in the skin, and melanin inhibits this. Vitamin D is also linked to fertility. It's involved in sex hormone production and has been associated with successful implantation and pregnancy outcomes.<p>If you're curious, check out Nina Jablonski and George Chaplin's work. Their hypothesis is that skin color evolution as fundamentally about reproductive fitness: dark enough to protect folate, light enough to synthesize vitamin D. Both nutrients affect fertility, fetal development, and offspring survival. They have an immediate primary impact on fertility and success, while skin cancer even in the most extreme environment/phenotype mismatch, has an onset after reproductive age.</p>
]]></description><pubDate>Sat, 29 Nov 2025 19:52:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=46090240</link><dc:creator>inciampati</dc:creator><comments>https://news.ycombinator.com/item?id=46090240</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46090240</guid></item></channel></rss>