<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: ordinarily</title><link>https://news.ycombinator.com/user?id=ordinarily</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 07 Apr 2026 01:56:39 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=ordinarily" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by ordinarily in "Show HN: I built a tiny LLM to demystify how language models work"]]></title><description><![CDATA[
<p>It's genuinely a great introduction to LLMs. I built my own awhile ago based off Milton's Paradise Lost: <a href="https://www.wvrk.org/works/milton" rel="nofollow">https://www.wvrk.org/works/milton</a></p>
]]></description><pubDate>Mon, 06 Apr 2026 02:57:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=47656472</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=47656472</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47656472</guid></item><item><title><![CDATA[New comment by ordinarily in "[dead]"]]></title><description><![CDATA[
<p>Nope, they're surprisingly hard to get ahold of. So I've resorted to being extremely noisy online.<p>Temporal accumulation: imagine you're observing a signal through a narrow window and you can only see a partial, noisy snapshot each time. "Temporal accumulation" is what happens when you let the observer remember previous snapshots and use them to improve its prediction of the next one. The persistence advantage P measures how much that memory helps, the difference in prediction error between an observer that accumulates across episodes and one that only uses the most recent snapshot.<p>For a black hole shadow (EHT), P ≈ 0: each snapshot already contains the full picture, memory adds nothing. For gravitational wave strain (LIGO), P is large and positive, the chirp evolves across snapshots, so memory is essential. The question the papers ask is: what determines how much memory helps? The answer turns out to be spectral entropy of the waveform, not mass.</p>
]]></description><pubDate>Tue, 31 Mar 2026 01:28:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=47581723</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=47581723</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47581723</guid></item><item><title><![CDATA[A 500K-parameter system that recovers invariant physics from observation alone]]></title><description><![CDATA[
<p>Article URL: <a href="https://erebus.org/">https://erebus.org/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47519296">https://news.ycombinator.com/item?id=47519296</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 25 Mar 2026 16:06:38 +0000</pubDate><link>https://erebus.org/</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=47519296</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47519296</guid></item><item><title><![CDATA[New comment by ordinarily in "Scaling Karpathy's Autoresearch: What Happens When the Agent Gets a GPU Cluster"]]></title><description><![CDATA[
<p>I've been doing this for about a month. I also have wildly complicated ML pipelines working similarly in parallel. When Karpathy's 'autoresearch' came out I was surprised by how novel it was treated.</p>
]]></description><pubDate>Fri, 20 Mar 2026 14:06:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=47454729</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=47454729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47454729</guid></item><item><title><![CDATA[New comment by ordinarily in "Hackers (1995) Animated Experience"]]></title><description><![CDATA[
<p>Awhile I made something really dumb: <a href="https://www.warpstream.com/etc/terminal" rel="nofollow">https://www.warpstream.com/etc/terminal</a> you can enter a 'gibson' command.</p>
]]></description><pubDate>Fri, 06 Feb 2026 16:39:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=46915033</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=46915033</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46915033</guid></item><item><title><![CDATA[The Art of Being Lazy(log)]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.warpstream.com/blog/the-art-of-being-lazy-log-lower-latency-and-higher-availability-with-delayed-sequencing">https://www.warpstream.com/blog/the-art-of-being-lazy-log-lower-latency-and-higher-availability-with-delayed-sequencing</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46890941">https://news.ycombinator.com/item?id=46890941</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 04 Feb 2026 20:09:08 +0000</pubDate><link>https://www.warpstream.com/blog/the-art-of-being-lazy-log-lower-latency-and-higher-availability-with-delayed-sequencing</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=46890941</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46890941</guid></item><item><title><![CDATA[The Case for an Iceberg-Native Database]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.warpstream.com/blog/the-case-for-an-iceberg-native-database-why-spark-jobs-and-zero-copy-kafka-wont-cut-it">https://www.warpstream.com/blog/the-case-for-an-iceberg-native-database-why-spark-jobs-and-zero-copy-kafka-wont-cut-it</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=45279876">https://news.ycombinator.com/item?id=45279876</a></p>
<p>Points: 8</p>
<p># Comments: 1</p>
]]></description><pubDate>Wed, 17 Sep 2025 18:53:38 +0000</pubDate><link>https://www.warpstream.com/blog/the-case-for-an-iceberg-native-database-why-spark-jobs-and-zero-copy-kafka-wont-cut-it</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=45279876</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45279876</guid></item><item><title><![CDATA[New comment by ordinarily in "Show HN: Engineering.fyi – Search across tech engineering blogs in one place"]]></title><description><![CDATA[
<p>This is awesome! We got some really technical deep dives here <a href="https://www.warpstream.com/blog-category/engineering" rel="nofollow">https://www.warpstream.com/blog-category/engineering</a></p>
]]></description><pubDate>Mon, 11 Aug 2025 03:26:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=44860544</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=44860544</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44860544</guid></item><item><title><![CDATA[New comment by ordinarily in "Thunderbird: Fluent Windows 11 Design"]]></title><description><![CDATA[
<p>Horizontal space is still a premium regardless of monitor size when designing/building for responsive viewports. Vertical space is almost zero cost in terms of design constraints.<p>Even on large monitors you'd be surprised the number of people at 150% zoom with small windows opened instead of fullscreen.</p>
]]></description><pubDate>Wed, 16 Jul 2025 15:20:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=44583265</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=44583265</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44583265</guid></item><item><title><![CDATA[New comment by ordinarily in "SVGs that feel like GIFs"]]></title><description><![CDATA[
<p>I used SVG animations (and sites like <a href="https://www.svgator.com/" rel="nofollow">https://www.svgator.com/</a>) long before stuff like Rive or Lottie was commonplace. SVG animations are great.</p>
]]></description><pubDate>Tue, 08 Jul 2025 19:40:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=44503313</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=44503313</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44503313</guid></item><item><title><![CDATA[New comment by ordinarily in "The Darwin Gödel Machine: AI that improves itself by rewriting its own code"]]></title><description><![CDATA[
<p>I think the general tone is more of a warning than an endorsement.</p>
]]></description><pubDate>Fri, 30 May 2025 14:37:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=44136660</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=44136660</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44136660</guid></item><item><title><![CDATA[New comment by ordinarily in "The Darwin Gödel Machine: AI that improves itself by rewriting its own code"]]></title><description><![CDATA[
<p>The pieces are coming together quickly <a href="https://ai-2027.com/" rel="nofollow">https://ai-2027.com/</a>.</p>
]]></description><pubDate>Fri, 30 May 2025 14:30:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=44136598</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=44136598</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44136598</guid></item><item><title><![CDATA[Reddit AMA with the WarpSteam co-founders]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.reddit.com/r/apachekafka/s/sKsTyk90n0">https://www.reddit.com/r/apachekafka/s/sKsTyk90n0</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=43985429">https://news.ycombinator.com/item?id=43985429</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 14 May 2025 15:06:45 +0000</pubDate><link>https://www.reddit.com/r/apachekafka/s/sKsTyk90n0</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43985429</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43985429</guid></item><item><title><![CDATA[New comment by ordinarily in "Fleurs du Mal"]]></title><description><![CDATA[
<p>I have a variety of early printings of this. My favorite being a 1931 edition illustrated by Major Felten, its beautiful.</p>
]]></description><pubDate>Fri, 09 May 2025 23:36:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=43941917</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43941917</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43941917</guid></item><item><title><![CDATA[New comment by ordinarily in "Bento gets a makeover"]]></title><description><![CDATA[
<p>Was hoping you'd appreciate our efforts to retain your original quirky vision. We named the rice ball Geoff as a homage to you (intentionally spelled the silly way). (<a href="https://warpstreamlabs.github.io/bento/docs/about" rel="nofollow">https://warpstreamlabs.github.io/bento/docs/about</a>)/</p>
]]></description><pubDate>Fri, 09 May 2025 14:11:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=43936998</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43936998</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43936998</guid></item><item><title><![CDATA[New comment by ordinarily in "Bento gets a makeover"]]></title><description><![CDATA[
<p>Lol I was actually looking at the Wiki earlier <a href="https://en.wikipedia.org/wiki/Bento_(database)" rel="nofollow">https://en.wikipedia.org/wiki/Bento_(database)</a></p>
]]></description><pubDate>Fri, 09 May 2025 04:02:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=43933694</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43933694</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43933694</guid></item><item><title><![CDATA[New comment by ordinarily in "Bento Gets a Makeover"]]></title><description><![CDATA[
<p>Pretty good community, Slack has 600+ users.</p>
]]></description><pubDate>Fri, 09 May 2025 01:36:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=43933055</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43933055</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43933055</guid></item><item><title><![CDATA[Bento gets a makeover]]></title><description><![CDATA[
<p>Article URL: <a href="https://warpstreamlabs.github.io/bento/">https://warpstreamlabs.github.io/bento/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=43931566">https://news.ycombinator.com/item?id=43931566</a></p>
<p>Points: 14</p>
<p># Comments: 11</p>
]]></description><pubDate>Thu, 08 May 2025 21:30:41 +0000</pubDate><link>https://warpstreamlabs.github.io/bento/</link><dc:creator>ordinarily</dc:creator><comments>https://news.ycombinator.com/item?id=43931566</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43931566</guid></item></channel></rss>