<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: geeksinthewoods</title><link>https://news.ycombinator.com/user?id=geeksinthewoods</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 21 Nov 2026 00:24:36 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=geeksinthewoods" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by geeksinthewoods in "Attention Lottery: DeepSeek, Sparse Attention, and the Future of AI Cognition"]]></title><description><![CDATA[
<p>The attention lottery framing feels especially timely now that DeepSeek's V3.2 tech report is out in the open. Seeing the actual top-k sparse routing and the post-training RL numbers spelled out makes the trade-offs concrete. Huge wins on speed and context, but every pruned token really is a quiet bet against the weird tail stuff that sometimes sparks real leaps...<p>What struck me most is how much DeepSeek's transparency accidentally lights up the closed models too. Long-context traces and million-token windows almost certainly lean on some variant of this under the hood. This article makes those black boxes feel a lot less mysterious. It leaves me both impressed by the engineering and quietly worried about the curiosity cost.<p>Also, the song / music video at the end is absurd in the best way!</p>
]]></description><pubDate>Fri, 05 Dec 2025 15:30:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=46162587</link><dc:creator>geeksinthewoods</dc:creator><comments>https://news.ycombinator.com/item?id=46162587</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46162587</guid></item><item><title><![CDATA[New comment by geeksinthewoods in "Attention Lottery: DeepSeek, Sparse Attention, and the Future of AI Cognition"]]></title><description><![CDATA[
<p>Ya. It seems like evolution itself has been running a sparsity experiment for millions of years. Sparse attention may be the universal price of survival: efficiency over imagination, precision over possibility.<p>The line about "missing insights being rare, deferred, and distributed" is like the hardest to notice in practice: optimization wins are loud (speed, cost, scores). Meanwhile the things we prune are often counterfactual ideas that never form, weird bridges that never get built, questions that never feel worth asking because our router did not surface them.<p>One thing I'm still unsure about (and would love to think about more) is how direct the analogy should be. In models, sparsity is engineered / learned under explicit objectives. In biology and culture it's much more emergent and multi-objective.</p>
]]></description><pubDate>Fri, 05 Dec 2025 15:22:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=46162469</link><dc:creator>geeksinthewoods</dc:creator><comments>https://news.ycombinator.com/item?id=46162469</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46162469</guid></item><item><title><![CDATA[New comment by geeksinthewoods in "Ask HN: What Are You Working On? (Nov 2025)"]]></title><description><![CDATA[
<p>lol. I keep on hitting my usage limits but I haven't upgraded my Claude Code CLI... so I just continue<p><a href="https://www.youtube.com/watch?v=2jmWnPfgtik" rel="nofollow">https://www.youtube.com/watch?v=2jmWnPfgtik</a><p><pre><code>  [End] --- for the visuals, reflect on the lyrics... think about AI learning more about how its thinking interacts with manifolds...
  ⎿  Claude usage limit reached. Your limit will reset at 3pm.
     /upgrade to increase your usage limit.
</code></pre>
> now let's continue</p>
]]></description><pubDate>Sat, 15 Nov 2025 03:04:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=45934706</link><dc:creator>geeksinthewoods</dc:creator><comments>https://news.ycombinator.com/item?id=45934706</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45934706</guid></item></channel></rss>