<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: SomewhatLikely</title><link>https://news.ycombinator.com/user?id=SomewhatLikely</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 15 Apr 2026 21:03:13 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=SomewhatLikely" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by SomewhatLikely in "How people woke up before alarm clocks"]]></title><description><![CDATA[
<p><a href="http://astronaut.io/" rel="nofollow">http://astronaut.io/</a></p>
]]></description><pubDate>Fri, 13 Mar 2026 06:27:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=47361305</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=47361305</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47361305</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Three Years from GPT-3 to Gemini 3"]]></title><description><![CDATA[
<p>I've seen it so this too.  I had it keeping a running tally over many turns and occasionally it would say something like: "... bringing the total to 304.. 306, no 303.  Haha, just kidding I know it's really 310." With the last number being the right one.  I'm curious if it's an organic behavior or a taught one.  It could be self learned through reinforcement learning, a way to correct itself since it doesn't have access to a backspace key.</p>
]]></description><pubDate>Tue, 25 Nov 2025 06:23:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=46042880</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=46042880</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46042880</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "The surprise deprecation of GPT-4o for ChatGPT consumers"]]></title><description><![CDATA[
<p>The default outputs are considerably shorter even in thinking mode. Something that helped me get the thinking mode back to an acceptable state was to switch to the Nerd personality and in the traits customization setting tell it to be complete and add extra relevant details.  With those additions it compares favorably to o3 on my recent chat history and even improved some cases.  I prefer to scan a longer output than have the LLM guess what to omit. But I know many people have complained about verbosity so I can understand why they may have moved to less verbiage.</p>
]]></description><pubDate>Sat, 09 Aug 2025 05:48:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=44844323</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44844323</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44844323</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "GPT-5: Key characteristics, pricing and system card"]]></title><description><![CDATA[
<p>Nobody's preventing them from rendering it and refining. That's certainly what we'd expect an AGI to do.</p>
]]></description><pubDate>Fri, 08 Aug 2025 06:17:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=44834044</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44834044</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44834044</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "The Bluesky Dictionary"]]></title><description><![CDATA[
<p>It's likely that the commenter has read less than 5 million posts worth of text though. So perhaps this still points to a lack of diversity in content.</p>
]]></description><pubDate>Thu, 07 Aug 2025 05:18:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=44820866</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44820866</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44820866</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "At a Loss for Words: A flawed idea is teaching kids to be poor readers (2019)"]]></title><description><![CDATA[
<p>What was the alternative you went with?</p>
]]></description><pubDate>Mon, 04 Aug 2025 06:51:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=44782765</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44782765</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44782765</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "At a Loss for Words: A flawed idea is teaching kids to be poor readers (2019)"]]></title><description><![CDATA[
<p>I saw a very similar timely appeal here on Hacker News a few years ago and taught my son with this book at the age of 4.  It has become my go-to comparison when prompting chat bots on what I want in a teaching material for other subjects.  I listened to the entire article posted here and it makes me wonder if schools are getting something as foundational as reading wrong how can we trust the attention to research on anything else they're teaching?  Don't get me wrong, I'm not going to pull my kid out of school but I'll dig a little deeper into how well he's learning.  For math, we've been doing the Beast Academy books. It has gone... Okay.  I like that they approach problems from many different ways which simulate the many different ways math is hidden in our interactions with the world.  For my younger son I've recently started Teaching Your Child... because of how well it went for his brother but for math I may try something else to have a new data point.  Something that occurred to me listening to the article is I wonder if certain skills are learned much faster with one on one instruction like the book has you do.  Our schools pretty much never teach that way out of efficiency, though home schools often do.  It may not be true for most subjects though or home school students would be so far ahead by college and that's not the impression I have.</p>
]]></description><pubDate>Sun, 03 Aug 2025 07:44:14 +0000</pubDate><link>https://news.ycombinator.com/item?id=44774880</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44774880</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44774880</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Spending Too Much Money on a Coding Agent"]]></title><description><![CDATA[
<p>It's pretty damn capital intensive to be a productive farmer today.  That said, AI will likely, hopefully, get cheaper over time.</p>
]]></description><pubDate>Fri, 04 Jul 2025 01:11:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=44460370</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=44460370</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44460370</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "How we made our AI code review bot stop leaving nitpicky comments"]]></title><description><![CDATA[
<p>You could probably modify the metric to addressed comments per 1000 lines of code.</p>
]]></description><pubDate>Sun, 22 Dec 2024 21:36:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=42489336</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=42489336</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42489336</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "ARIA: An Open Multimodal Native Mixture-of-Experts Model"]]></title><description><![CDATA[
<p><i>"Here, we provide a quantifiable definition: A multimodal native model refers to a single model with strong understanding capabilities across multiple input modalities (e.g. text, code, image, video), that matches or exceeds the modality specialized models of similar capacities."</i></p>
]]></description><pubDate>Fri, 11 Oct 2024 07:33:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=41807183</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41807183</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41807183</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Practices of Reliable Software Design"]]></title><description><![CDATA[
<p>My first thought upon seeing the prompt:<p><pre><code>    If you would build an in-memory cache, how would you do it?

    It should have good performance and be able to hold many entries. 
    Reads are more common than writes. I know how I would do it already, 
    but I’m curious about your approach.
</code></pre>
Was to add this requirement since it comes up so often:<p><pre><code>    Let's assume that keys accessed follow a power law, so some keys get 
    accessed very frequently and we would like them to have the fastest 
    retrieval of all.
</code></pre>
I'm not sure if there are any efficient tweaks to hash tables or b-trees that might help with this additional requirement.  Obviously we could make a hash table take way more space than needed to reduce collisions, but with a decent load factor is the answer to just swap frequently accessed keys to the beginning of their probe chain?  How do we know it's frequently accessed? Count-Min sketch?<p>Even with that tweak, the hottest keys will still be scattered around memory.  Wouldn't it be best if their entries could fit into fewer pages?  So, maybe a much smaller "hot" table containing say the 1,000 most accessed keys.  We still want a high load factor to maximize the use of cache pages so perhaps perfect hashing?</p>
]]></description><pubDate>Wed, 09 Oct 2024 06:08:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=41784984</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41784984</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41784984</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Bitten by Unicode"]]></title><description><![CDATA[
<p>Where I thought this might be going from the first paragraph:<p>Negative numbers are sometimes represented with parentheses: (234.58)<p>Tables sometimes tell you in the description that all numbers in are in 1000's or millions.<p>The dollar sign is used by many currencies, including in Australia and Canada.<p>I'd probably look around for some other gotchas.  Here's one page on prices in general: <a href="https://gist.github.com/rgs/6509585" rel="nofollow">https://gist.github.com/rgs/6509585</a> but interestingly doesn't quite cover the OP's problem or the ones I brought up, though the use cases are slightly different.</p>
]]></description><pubDate>Mon, 09 Sep 2024 05:47:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=41485772</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41485772</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41485772</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "How does cosine similarity work?"]]></title><description><![CDATA[
<p>Something worth mentioning is that if your vectors all have the same length then cosine similarity and Euclidean distance will order most (all?) neighbors in the same order. Think of your query vector as a point on a unit sphere.  The Euclidean distance to a neighbor will be a chord from the query point to the neighbor.  Just as with the angle between the query-to-origin and the neighbor-to-origin vectors, the farther you move the neighbor from the query point on the surface of the sphere, the longer the chord between those points gets too.<p>EDIT: Here's a better treatment, and it is the case that they give the exact same orderings: <a href="https://ajayp.app/posts/2020/05/relationship-between-cosine-similarity-and-euclidean-distance/" rel="nofollow">https://ajayp.app/posts/2020/05/relationship-between-cosine-...</a></p>
]]></description><pubDate>Sat, 07 Sep 2024 06:06:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=41471809</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41471809</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41471809</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Manipulating large language models to increase product visibility"]]></title><description><![CDATA[
<p>This feels similar to those adversarial examples that first came out that were very tuned for a specific image recognizer.  I haven't followed the research but I know they had some very limited success to getting it to work in the real world.  I'm not sure if they ever worked across different models though.<p>The paper claims there is literature with more success for LLMs:<p><pre><code>   Large language models have been shown to be vulnerable to adversarial
   attacks, in which attackers introduce maliciously crafted token sequences
   into the input prompt to circumvent the model’s safety mechanisms and 
   generate a harmful response [1, 14].</code></pre></p>
]]></description><pubDate>Sat, 07 Sep 2024 05:20:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=41471624</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41471624</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41471624</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Why Don't Tech Companies Pay Their Engineers to Stay?"]]></title><description><![CDATA[
<p><a href="https://jonpauluritis.com/articles/why-arent-developers-paid-more/" rel="nofollow">https://jonpauluritis.com/articles/why-arent-developers-paid...</a></p>
]]></description><pubDate>Fri, 06 Sep 2024 06:08:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=41463334</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41463334</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41463334</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Show HN: I'm making an AI scraper called FetchFox"]]></title><description><![CDATA[
<p>Can I recommend you provide some cost estimates next to the examples for using your own key?  I tried a few custom extractions and then checked my usage dashboard and it was already over $2.</p>
]]></description><pubDate>Wed, 04 Sep 2024 07:39:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=41442907</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41442907</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41442907</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Architectural Effects on Maximum Dependency Lengths of Recurrent Neural Networks"]]></title><description><![CDATA[
<p><a href="https://arxiv.org/abs/2006.16236" rel="nofollow">https://arxiv.org/abs/2006.16236</a> Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention</p>
]]></description><pubDate>Sat, 31 Aug 2024 06:31:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=41406936</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41406936</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41406936</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Diffusion models are real-time game engines"]]></title><description><![CDATA[
<p>Video games are gonna be wild in the near future.  You could have one person talking to a model producing something that's on par with a AAA title from today.  Imagine the 2d sidescroller boom on Steam but with immersive photorealistic 3d games with hyper-realistic physics (water flow, fire that spreads, tornados) and full deformability and buildability because the model is pretrained with real world videos.  Your game is just a "style" that tweaks some priors on look, settings, and story.</p>
]]></description><pubDate>Wed, 28 Aug 2024 07:21:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=41376879</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41376879</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41376879</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Shell.how: Explain Shell Commands"]]></title><description><![CDATA[
<p>I've switched to LLM's for explaining commands.  Here's ChatGPT4o:<p>The command tar xvzf 1.zip appears to be trying to extract a compressed archive file, but there's a mismatch between the file extension (1.zip) and the command options used for tar. Let me break it down:<p>tar: This is the command-line utility for working with tar archives (tape archives). It can be used to create, extract, and manipulate tar files.<p>x: This option tells tar to extract files from an archive.<p>v: This stands for "verbose", which means that tar will list the files being extracted to the terminal.<p>z: This option tells tar to handle a compressed archive using gzip. It is typically used with .tar.gz or .tgz files.<p>f: This tells tar that the next argument is the name of the archive file.<p>However, the file 1.zip is a ZIP file, not a tar.gz file. ZIP files are handled by the unzip command, not tar. Using tar xvzf on a .zip file would result in an error, as tar expects a .tar.gz file when using the z option.<p>To properly extract a .zip file, you should use:<p>unzip 1.zip<p>This command will extract the contents of the ZIP file.</p>
]]></description><pubDate>Sun, 25 Aug 2024 03:37:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=41344159</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41344159</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41344159</guid></item><item><title><![CDATA[New comment by SomewhatLikely in "Maps Mania: How the World Powers Itself"]]></title><description><![CDATA[
<p>I suspect there may be a flaw in the visualization where smaller dots are drawn with disproportionately more pixels than bigger dots.  Just looking at how much green is on the map is way more than numbers I can find support for.  For example, spain visually looks like more than 90% solar/wind but in reality it appears to only be about 60% [ <a href="https://www.reuters.com/business/energy/renewables-produce-almost-60-spains-electricity-2024-07-02/" rel="nofollow">https://www.reuters.com/business/energy/renewables-produce-a...</a> ]</p>
]]></description><pubDate>Thu, 22 Aug 2024 07:12:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=41317598</link><dc:creator>SomewhatLikely</dc:creator><comments>https://news.ycombinator.com/item?id=41317598</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=41317598</guid></item></channel></rss>