<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: geph2021</title><link>https://news.ycombinator.com/user?id=geph2021</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 08 Sep 2026 16:55:34 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=geph2021" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by geph2021 in "How Deep Can Humans Go?"]]></title><description><![CDATA[
<p>wow, that's terrible.  I was a competitive swimmer in my youth, and we did similar competitions from time to time (very rarely).   I've never seen or even heard of a swimmer passing out underwater.<p>A more common drill we would do in practice were called "hypoxic" sets, where we would do one length of the pool breathing every 3 strokes, then the next every 5 strokes, then 7, 9 etc.. until you were going across the whole length (25 meters) without breathing.  Not everyone could do it towards the longer distances without breathing, and the coaches would look out for "cheaters", but never once did anyone pass out.  Maybe most swimmers, by way of the typical training and exertion in the pool, just don't develop a very good suppression of the "breath signal".   I also never remember seeing anyone purposely hyperventilate so that they can stay under water longer.</p>
]]></description><pubDate>Thu, 22 Aug 2024 21:23:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=41324508</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41324508</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41324508</guid></item><item><title><![CDATA[New comment by geph2021 in "Programmers Don't Read Books – But You Should (2008)"]]></title><description><![CDATA[
<p>That's exactly the author's point:<p><pre><code>  Most programming books suck.
</code></pre>
Yet he has some recommendations:<p><pre><code>   But I do have this call to arms: my top five programming books every working programmer should own – and read.

</code></pre>
They are:<p><pre><code>  Code Complete 2 [1]

  Don't Make Me Think [2]

  Peopleware [3]

  Pragmatic Programmer [4]

  Facts and Fallacies [5]
</code></pre>
1 - <a href="https://www.amazon.com/exec/obidos/ASIN/0735619670/codihorr-20?ref=blog.codinghorror.com" rel="nofollow">https://www.amazon.com/exec/obidos/ASIN/0735619670/codihorr-...</a><p>2 - <a href="http://www.amazon.com/exec/obidos/ASIN/0321965515/codihorr-20?ref=blog.codinghorror.com" rel="nofollow">http://www.amazon.com/exec/obidos/ASIN/0321965515/codihorr-2...</a><p>3 - <a href="http://www.amazon.com/exec/obidos/ASIN/0932633439/codihorr-20?ref=blog.codinghorror.com" rel="nofollow">http://www.amazon.com/exec/obidos/ASIN/0932633439/codihorr-2...</a><p>4 - <a href="http://www.amazon.com/exec/obidos/ASIN/020161622X/codihorr-20?ref=blog.codinghorror.com" rel="nofollow">http://www.amazon.com/exec/obidos/ASIN/020161622X/codihorr-2...</a><p>5 - <a href="http://www.amazon.com/exec/obidos/ASIN/0321117425/codihorr-20?ref=blog.codinghorror.com" rel="nofollow">http://www.amazon.com/exec/obidos/ASIN/0321117425/codihorr-2...</a></p>
]]></description><pubDate>Thu, 15 Aug 2024 20:28:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=41260020</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41260020</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41260020</guid></item><item><title><![CDATA[New comment by geph2021 in "Is running a more efficient way to travel than walking?"]]></title><description><![CDATA[
<p>I think this sort of technique is used in boxing and MMA:  play a good defense, stay out of trouble, and let the aggressor exhaust themselves until later rounds when you feel you may have a fitness advantage.  I think there were good examples of this in the early days of MMA, when there was much less, if any, specialized training, techniques and strategy.  It's probably much less effective now, since a well trained aggressor will know how to pace themselves too.</p>
]]></description><pubDate>Mon, 12 Aug 2024 21:40:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=41229637</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41229637</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41229637</guid></item><item><title><![CDATA[New comment by geph2021 in "Breaking my hand forced me to write all my code with AI for 2 months"]]></title><description><![CDATA[
<p><p><pre><code>   Our motor controller had a 100 page datasheet that was overwhelming and dense - but uploading it to Claude and then asking questions let us quickly resolve one of our issues!
</code></pre>
That's interesting.  I've tried exactly this with chatGPT (enterprise install), and it failed pretty miserably when asking it basic questions about configuration/control.  Although, I do have to add that the datasheet in question was a lot more complicated than a motor controller, and included a Firmware API guide.</p>
]]></description><pubDate>Fri, 09 Aug 2024 19:22:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=41204628</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41204628</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41204628</guid></item><item><title><![CDATA[The future of kdb+?]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.timestored.com/b/the-future-of-kdb/">https://www.timestored.com/b/the-future-of-kdb/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=41143764">https://news.ycombinator.com/item?id=41143764</a></p>
<p>Points: 219</p>
<p># Comments: 66</p>
]]></description><pubDate>Sat, 03 Aug 2024 00:18:13 +0000</pubDate><link>https://www.timestored.com/b/the-future-of-kdb/</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41143764</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41143764</guid></item><item><title><![CDATA[New comment by geph2021 in "Jeff Bezos' management rules are slowly unraveling inside Amazon"]]></title><description><![CDATA[
<p><p><pre><code>   There is no upward mobility at the company, unless you have been in some org 5+ years.
</code></pre>
I think that's true of most large organizations.</p>
]]></description><pubDate>Wed, 31 Jul 2024 17:35:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=41121283</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41121283</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41121283</guid></item><item><title><![CDATA[Ask HN: What are the best resources today for learning AI/LLMs]]></title><description><![CDATA[
<p>I know of a few resources, but I'm sure there are others and would love to gather some pointers from the HN crowd. 
These come to mind:<p>https://course.fast.ai/<p>https://www.deeplearning.ai/courses/<p>Andrej Karpathy youtube and github:<p>https://www.youtube.com/@AndrejKarpathy/videos</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=41105348">https://news.ycombinator.com/item?id=41105348</a></p>
<p>Points: 20</p>
<p># Comments: 9</p>
]]></description><pubDate>Tue, 30 Jul 2024 01:36:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=41105348</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=41105348</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41105348</guid></item><item><title><![CDATA[New comment by geph2021 in "AI's $600B Question"]]></title><description><![CDATA[
<p>People have been comparing and contrasting successive technology innovation cycles for a long time.   OP mentions railroads, but that doesn't mean there's an "effort to hitch" AI to railroads.   It's just a comparison, which may or may not be useful in understanding how the latest innovation cycle may impact the world.</p>
]]></description><pubDate>Fri, 05 Jul 2024 15:40:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=40883685</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40883685</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40883685</guid></item><item><title><![CDATA[New comment by geph2021 in "AI's $600B Question"]]></title><description><![CDATA[
<p>When you think about the promise (or hype) of crypto/bitcoin/blockchain 10 years ago, in some sense it augured equally, if not more, transformative change/disruption than AI.<p>Crypto portended drastic and fundamental changes: programmable money, disintermediation, and the decentralization of the very foundations of our society (i.e. money, banking, commerce).   Suffice to say that nothing close to this has happened, and probably will never happen.<p>So I can see how many people are equally skeptical that AI, as the next hyped transformative technology, will achieve anything near the many lofty predictions.</p>
]]></description><pubDate>Wed, 03 Jul 2024 21:44:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=40870281</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40870281</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40870281</guid></item><item><title><![CDATA[Obtaining source code for training LLMs]]></title><description><![CDATA[
<p>Article URL: <a href="https://shape-of-code.com/2024/06/02/obtaining-source-code-for-training-llms/">https://shape-of-code.com/2024/06/02/obtaining-source-code-for-training-llms/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=40662645">https://news.ycombinator.com/item?id=40662645</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 12 Jun 2024 20:22:16 +0000</pubDate><link>https://shape-of-code.com/2024/06/02/obtaining-source-code-for-training-llms/</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40662645</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40662645</guid></item><item><title><![CDATA[Efficient WaveRNN: Block Sparsity]]></title><description><![CDATA[
<p>Article URL: <a href="https://andrew.gibiansky.com/wavernn-demystified-sparsity/">https://andrew.gibiansky.com/wavernn-demystified-sparsity/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=40652489">https://news.ycombinator.com/item?id=40652489</a></p>
<p>Points: 7</p>
<p># Comments: 2</p>
]]></description><pubDate>Tue, 11 Jun 2024 22:41:38 +0000</pubDate><link>https://andrew.gibiansky.com/wavernn-demystified-sparsity/</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40652489</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40652489</guid></item><item><title><![CDATA[New comment by geph2021 in "Employees who stay in companies longer than two years get paid 50% less (2014)"]]></title><description><![CDATA[
<p>It depends on the counter offer and your situation.  I would have agreed with you, but a couple years ago I told my employer that I had a job offer kind of land in my lap (I wasn't actively looking), that it was good and I was inclined to take it.   I wasn't unhappy at my current job, but the change in title/seniority and compensation of the job offer was definitely worth it.   They didn't want me to leave, so they countered with a very strong offer, which I decided to take and stay with the company.   If you want to leave because you're unhappy with other aspects of your job, then I agree that accepting a counter offer doesn't seem like a great choice.</p>
]]></description><pubDate>Wed, 05 Jun 2024 20:26:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=40590136</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40590136</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40590136</guid></item><item><title><![CDATA[New comment by geph2021 in "Fossil fuels could have been left in the dust 25 years ago"]]></title><description><![CDATA[
<p><p><pre><code>   governments should have been falling over themselves to buy or otherwise subsidise expensive solar PV, because the more we bought, the faster the price would fall
</code></pre>
That's quite a stretch, especially considering that there are plenty of examples where government subsidies and intervention distorts markets and makes them less efficient and more dysfunctional.   I don't disagree with the general premise of the article:  more could have been done to transition to renewable energy sooner.   But it's really simplistic to say there was a clear, easy answer to this, and it simply involved more government spending on solar energy.</p>
]]></description><pubDate>Thu, 23 May 2024 20:43:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=40459667</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40459667</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40459667</guid></item><item><title><![CDATA[New comment by geph2021 in "CADmium: A local-first CAD program built for the browser"]]></title><description><![CDATA[
<p>I don't know a lot about these tools, but I've used QCAD[1] for home improvement projects, drawing schematics and layout diagrams, and been very happy with it.  It seems quite powerful.   It's also open source, although not browser based.<p>I'm just wondering why QCAD doesn't seem to get much if any mention on HN when CAD tools and open source comes up.<p>1 - <a href="https://www.qcad.org/" rel="nofollow">https://www.qcad.org/</a></p>
]]></description><pubDate>Tue, 21 May 2024 19:31:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=40432768</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40432768</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40432768</guid></item><item><title><![CDATA[New comment by geph2021 in "US drug control agency will move to reclassify marijuana"]]></title><description><![CDATA[
<p>same :)</p>
]]></description><pubDate>Wed, 01 May 2024 00:40:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=40218229</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40218229</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40218229</guid></item><item><title><![CDATA[New comment by geph2021 in "World War I dangers in France's red zones"]]></title><description><![CDATA[
<p>Recommendation for anyone wanting to get into the gory, shocking details of WWI:<p>The podcast, Dan Carlin's A blueprint for Armageddon[1].<p>Over 24 hours on the topic, and I was left absolutely devastated by the descriptions of the impact and events of WWI.  I guess I just didn't have a great appreciation or knowledge of WWI.  At least for me, my grade school education was mostly focused on WWII and other more recent conflicts.<p>1 - <a href="https://www.dancarlin.com/product/hardcore-history-50-55-blueprint-for-armageddon-series/" rel="nofollow">https://www.dancarlin.com/product/hardcore-history-50-55-blu...</a></p>
]]></description><pubDate>Fri, 26 Apr 2024 23:23:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=40175421</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=40175421</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40175421</guid></item><item><title><![CDATA[New comment by geph2021 in "Llm.c – LLM training in simple, pure C/CUDA"]]></title><description><![CDATA[
<p>As far as I can tell, its optional dependency is Open MP, not CUDA.  Doesn't seem directly dependent on CUDA.</p>
]]></description><pubDate>Mon, 08 Apr 2024 21:05:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=39973700</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=39973700</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39973700</guid></item><item><title><![CDATA[New comment by geph2021 in "4T transistors, one giant chip (Cerebras WSE-3) [video]"]]></title><description><![CDATA[
<p>Thanks for sharing!   Very interesting.<p>"We call it the engine block because it somewhat resembles a three cylinder motorcycle engine"<p>(referring to the power distribution and cooling)</p>
]]></description><pubDate>Wed, 13 Mar 2024 17:28:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=39694524</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=39694524</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39694524</guid></item><item><title><![CDATA[New comment by geph2021 in "4T transistors, one giant chip (Cerebras WSE-3) [video]"]]></title><description><![CDATA[
<p>and cooling!<p>Imagine the heat sink on that thing.  Would look like a cast-iron Dutch oven :)</p>
]]></description><pubDate>Wed, 13 Mar 2024 16:52:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=39693991</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=39693991</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39693991</guid></item><item><title><![CDATA[New comment by geph2021 in "4T transistors, one giant chip (Cerebras WSE-3) [video]"]]></title><description><![CDATA[
<p>original title is:<p>"4,000,000,000,000 Transistors, One Giant Chip (Cerebras WSE-3)"<p>So I guess they're trying to stay true to it.</p>
]]></description><pubDate>Wed, 13 Mar 2024 16:44:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=39693853</link><dc:creator>geph2021</dc:creator><comments>https://news.ycombinator.com/item?id=39693853</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39693853</guid></item></channel></rss>