<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: contravariant</title><link>https://news.ycombinator.com/user?id=contravariant</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 21 Nov 2026 10:17:34 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=contravariant" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by contravariant in "What is the best harness for each model?"]]></title><description><![CDATA[
<p>I think the law is more powerful if you view it as a force of nature. You don't <i>have</i> to make your organisation conform to the shape of the software, you can often do it the other way around.<p>The trouble starts when the requirements for your software are at odds with the requirements of your organisational structure.</p>
]]></description><pubDate>Tue, 17 Nov 2026 21:05:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=50538173</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50538173</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50538173</guid></item><item><title><![CDATA[New comment by contravariant in "What is the best harness for each model?"]]></title><description><![CDATA[
<p>I think Conway's law applies. In an ideal situation your code mimics your organisational structure.<p>The places where you let an agent do its thing is bound by the (implicit) interface your prompt creates. In an <i>ideal</i> situation this interface is a simple isolated thing in your application.<p>If you just let it do its thing, expect a monolith. You shouldn't rely on it coming up with interfaces that mesh well with what you had in mind, and each of those interfaces represents a decision that it took, restricting future iterations of the code. So a monolith is not only what you can expect, but also the best you can hope for.<p>If you want to let it cook, create more places for code that does a thing and you don't care how.</p>
]]></description><pubDate>Tue, 17 Nov 2026 16:31:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=50533876</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50533876</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50533876</guid></item><item><title><![CDATA[New comment by contravariant in "New Evidence Shows School Phone Bans Are Gradually Improving Test Scores"]]></title><description><![CDATA[
<p>True, but probably not enough to delay implementing a policy that seems to be giving small but noticeable improvements across the board.</p>
]]></description><pubDate>Tue, 17 Nov 2026 14:02:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=50531511</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50531511</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50531511</guid></item><item><title><![CDATA[New comment by contravariant in "New Evidence Shows School Phone Bans Are Gradually Improving Test Scores"]]></title><description><![CDATA[
<p>Individual test students are going to be hard to predict anyway. Though I think this is compared to the sample standard deviation, and for most students their improvement over time is going to be smaller than the difference between two random students.</p>
]]></description><pubDate>Tue, 17 Nov 2026 13:57:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=50531447</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50531447</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50531447</guid></item><item><title><![CDATA[New comment by contravariant in "New Evidence Shows School Phone Bans Are Gradually Improving Test Scores"]]></title><description><![CDATA[
<p>I'm both confused and not terribly surprised that people don't seem to recall that's an option.</p>
]]></description><pubDate>Tue, 17 Nov 2026 13:47:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=50531323</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50531323</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50531323</guid></item><item><title><![CDATA[New comment by contravariant in "New Evidence Shows School Phone Bans Are Gradually Improving Test Scores"]]></title><description><![CDATA[
<p>Does it matter? Let's keep at it for a while and see if it continues.<p>Though I imagine they're pretty certain whether it's one or the other. Teachers at least seem to be quite certain, though 'common sense' doesn't always pan out.</p>
]]></description><pubDate>Tue, 17 Nov 2026 13:45:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=50531299</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50531299</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50531299</guid></item><item><title><![CDATA[New comment by contravariant in "Rows vs. Columns: How different are they, really?"]]></title><description><![CDATA[
<p>I think there's never going to be a point where you can have the kind of performance you can get by optimising a datastructure for a 'fixed' dataset. Nor can you get the maximal performance updating data if you need to maintain a useful datastructure.<p>A simple case is a sorted list, you can sort it after every update, you can ignore the ordering and sort it as needed, or you can use a datastructure that is typically O(log(N)) at both updates and retrieval. Better datastructurs help, but they cannot remove the tradeoff entirely (in fact O(log(N)) inserts are kind of theoretically minimal, though a trie can do it in O(1) for fixed length datatypes).</p>
]]></description><pubDate>Tue, 17 Nov 2026 13:42:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=50531246</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50531246</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50531246</guid></item><item><title><![CDATA[New comment by contravariant in "Formal Verification Is Not Verifiable"]]></title><description><![CDATA[
<p>What do you think of Hoare / Dijkstra's approach of having pre and post conditions for every piece of code? It seems like the kind of structure that allows to talk about properties of code in a intelligible way (the classical example being a sort function with pre-condition 'a list' and post-condition 'the same list but sorted').<p>If you have dependent types then that approach is essentially the strongly typed version, and I think you're proposing a more gradually typed alternative?</p>
]]></description><pubDate>Tue, 17 Nov 2026 12:44:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=50530649</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50530649</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50530649</guid></item><item><title><![CDATA[New comment by contravariant in "Universal XSS in uBlock Origin from website-driven element picker"]]></title><description><![CDATA[
<p>The other way round it looks like a website's javascript being allowed to interfere with extensions is also a bit of a problem. I can't really think of a reason you'd want to let it interecept requests to a moz-extension: URL or allow it to make an iframe for one.<p>Neither could be considered a security bug, but it's a bit odd the entire interface is wide open.</p>
]]></description><pubDate>Mon, 16 Nov 2026 09:26:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=50515656</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50515656</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50515656</guid></item><item><title><![CDATA[New comment by contravariant in "Claude Is Down"]]></title><description><![CDATA[
<p>It's one way to describe the way some people write code I suppose.</p>
]]></description><pubDate>Fri, 13 Nov 2026 11:15:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=50481882</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50481882</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50481882</guid></item><item><title><![CDATA[New comment by contravariant in "Service degradation on Microsoft 365"]]></title><description><![CDATA[
<p>Sounds like it running correctly was the meta-stable state. Failure is actually more stable since it can reach a failed state from the ground state of simply not running.</p>
]]></description><pubDate>Fri, 13 Nov 2026 11:04:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=50481770</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50481770</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50481770</guid></item><item><title><![CDATA[New comment by contravariant in "AlphaProof Code"]]></title><description><![CDATA[
<p>Dijkstra's approach would be to prove there is a strictly decreasing natural number associated to the loop.<p>This limits programs to the ones that can be formally shown to terminate. Which is good enough for most purposes.<p>My approach is to accept programs that run forever as long as I'm fairly sure each loop makes progress of some sort (instead of simply doing "while not solved: continue").<p>Edit: There's also a fun algorithm that, as far as I know, only provably terminates by being bounded by a sequence of strictly decreasing ordinals. Which breaks all kinds of models for proving it terminates.</p>
]]></description><pubDate>Tue, 10 Nov 2026 12:37:44 +0000</pubDate><link>https://news.ycombinator.com/item?id=50439472</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50439472</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50439472</guid></item><item><title><![CDATA[New comment by contravariant in "Fold Epic Paper Airplanes"]]></title><description><![CDATA[
<p>Figured something like that was the case, glad you could fix it! :)</p>
]]></description><pubDate>Mon, 09 Nov 2026 21:40:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=50432631</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50432631</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50432631</guid></item><item><title><![CDATA[New comment by contravariant in "Fold Epic Paper Airplanes"]]></title><description><![CDATA[
<p>Intermediate shows up twice for some reason.</p>
]]></description><pubDate>Mon, 09 Nov 2026 12:40:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=50424786</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50424786</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50424786</guid></item><item><title><![CDATA[New comment by contravariant in "AI is forcing us to pay off 40 years of technical debt"]]></title><description><![CDATA[
<p>If you don't particularly care how the code works or specifically what it does it seems pointless to worry about abstraction. Since, as you say, it solved the what you wanted solved and did so quickly.<p>But when you <i>do</i> care, it becomes important.<p>And, I suppose, computer code in general is <i>terrible</i> about writing things at the right level of abstraction, or at least human generated code is and I think LLMs do worse, somehow. So there is something to be said for prose as a specification language in most cases.<p>But I still think we're going to end up with some kind of code to reason about programs. Same way mathematicians have been using mathematical notation to reason about maths. But maybe it will be different code. Who knows, literary programming might catch on.</p>
]]></description><pubDate>Thu, 05 Nov 2026 00:20:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=50373321</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50373321</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50373321</guid></item><item><title><![CDATA[New comment by contravariant in "AI is forcing us to pay off 40 years of technical debt"]]></title><description><![CDATA[
<p>The most pessimistic interpretation is that AI is superhumanly good at exposing technical debt and not that good at avoiding it.<p>We'd be replacing decades of technical debt with debt we'd have no hope of paying.</p>
]]></description><pubDate>Wed, 04 Nov 2026 12:57:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=50364917</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50364917</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50364917</guid></item><item><title><![CDATA[New comment by contravariant in "AI is forcing us to pay off 40 years of technical debt"]]></title><description><![CDATA[
<p>I'm still unsure if we won't figure out that the right abstraction level to describe what code means is, well, code.</p>
]]></description><pubDate>Wed, 04 Nov 2026 12:55:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=50364887</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50364887</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50364887</guid></item><item><title><![CDATA[New comment by contravariant in "Efficient algorithms, the Exponential Time Hypothesis, and losing the plot"]]></title><description><![CDATA[
<p>Maybe I'm reading a bit too much into it but knowing it is NP (i.e. can be verified in polynomial time by a Turing machine) means it can be solved in polynomial time by a nondeterministic turing machine and there's at worst 2^N branches in it. Sort of. Giving a naive expectation of 2^N runtime times some polynomial. Which is exponential.</p>
]]></description><pubDate>Tue, 03 Nov 2026 13:23:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=50351681</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50351681</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50351681</guid></item><item><title><![CDATA[New comment by contravariant in "Supermarket freezers across Europe hit by second suspected cyber-attack"]]></title><description><![CDATA[
<p>General unrest, as well as stress testing critical infrastructure, and perhaps most importantly shifting the window for what counts as acts of aggression that don't trigger retaliation.</p>
]]></description><pubDate>Mon, 02 Nov 2026 14:57:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=50337675</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50337675</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50337675</guid></item><item><title><![CDATA[New comment by contravariant in "Mistral Large 4 (Le Chonk) is now Open Weight"]]></title><description><![CDATA[
<p>It's always a bit unclear to me to what extent you're bound by a copyright when you're not distributing a work.<p>If someone sends me a copy while fully adhering to the license, what right does someone have to tell me what I can do with it? If I buy a book do I not have a right to read it? Lend it to people?<p>It's probably a good thing libraries already existed pre modern copyright.</p>
]]></description><pubDate>Thu, 29 Oct 2026 15:02:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=50290242</link><dc:creator>contravariant</dc:creator><comments>https://news.ycombinator.com/item?id=50290242</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50290242</guid></item></channel></rss>