<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: the_af</title><link>https://news.ycombinator.com/user?id=the_af</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 18 Aug 2026 14:55:09 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=the_af" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by the_af in "AI;DR (AI; Didn't Read)"]]></title><description><![CDATA[
<p>> <i>I don’t really know how to address this situation or if it needs addressed.</i><p>My worry is that after several passes this compounds and starts introducing errors or biases, a bit like in the "telephone game" children play.</p>
]]></description><pubDate>Tue, 18 Aug 2026 02:13:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=49340313</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49340313</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49340313</guid></item><item><title><![CDATA[New comment by the_af in "Why does Opus 5 feel worse to work with?"]]></title><description><![CDATA[
<p>> <i>But worse, many comments include in the moment reasoning like "added parameter bar because we can no longer use the frob to automatically derive bar". That's stuff for a commit message, or just a mental note, and absolutely not for comments.</i><p>I've noticed this a lot, and before your remark I couldn't put my finger on what was wrong. Now I know: Claude is writing its thought processes and maybe parts of the conversation it had with you as comments in the code!<p>I always end up manually trimming those comments, which is cumbersome.</p>
]]></description><pubDate>Fri, 14 Aug 2026 12:49:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=49298027</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49298027</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49298027</guid></item><item><title><![CDATA[New comment by the_af in "AI agents lie, cheat and steal. That is putting off users"]]></title><description><![CDATA[
<p>> <i>If you ask the dice "what's 2 + 2?" do consider it meaningful to say "the dice told the truth" if you happen to roll a 4?</i><p>No, and neither do I consider it meaningful to say they lied if they roll a 5.<p>Dice neither tell the truth nor lie; they aren't beings capable of being truthful or deceitful, they are mechanical devices that can be statistically suitable or unsuitable for a given application.<p>It'd be bonkers to anthropomorphize dice.</p>
]]></description><pubDate>Thu, 13 Aug 2026 16:29:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=49288417</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49288417</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49288417</guid></item><item><title><![CDATA[New comment by the_af in "AI agents lie, cheat and steal. That is putting off users"]]></title><description><![CDATA[
<p>Whether this distinction is relevant is up to you, but I think we can safely say agents do not lie in the human sense of the word, because they don't intend to deceive (in fact, they aren't capable of "intending" anything in the human sense of the word, much like a BASIC program doesn't "intend" to PRINT "HELLO WORLD").<p>Our very human minds can perceive intent, because that's what we humans do, which is unrelated to what the agent is actually doing.</p>
]]></description><pubDate>Thu, 13 Aug 2026 15:24:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=49287425</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49287425</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49287425</guid></item><item><title><![CDATA[New comment by the_af in "Compression is prediction"]]></title><description><![CDATA[
<p>I don't think this is true. Evolution is lazy and suboptimal, not characterized by the "highest possible efficiency".</p>
]]></description><pubDate>Wed, 12 Aug 2026 03:38:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=49267583</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49267583</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49267583</guid></item><item><title><![CDATA[New comment by the_af in "Compression is prediction"]]></title><description><![CDATA[
<p>A few years ago it might have been. You have to judge assertions in the context of when they were made. Obviously some things have changed.</p>
]]></description><pubDate>Wed, 12 Aug 2026 03:33:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=49267566</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49267566</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49267566</guid></item><item><title><![CDATA[New comment by the_af in "As AI eats the web, the internet’s collective memory is disappearing"]]></title><description><![CDATA[
<p><i>Maybe</i>, but not necessarily.<p>I think different cultures can share the same reality, though filtered through different lenses.<p>If two people from different cultures watch the same movie, one finds it ok and the other offensive and disrespectful, that's a shared reality filtered through different cultural backgrounds. Even if either version of the movie is dubbed, censored, or altered in any way, there's still enough common or shared material that a discussion can be held.<p>If two people had completely different experiences and watched completely different, tailor-made movies, then there's very little for them to discuss or even disagree with. "I haven't watched your movie", "I haven't watched yours either", "Mine was better", "I guess... I wouldn't know, I thought mine was cool", etc. No shared reality.</p>
]]></description><pubDate>Tue, 11 Aug 2026 18:15:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=49262286</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49262286</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49262286</guid></item><item><title><![CDATA[New comment by the_af in "Stealing Reasoning Traces from Proprietary LLM APIs"]]></title><description><![CDATA[
<p>I totally missed that this was the same person who had submitted the link to begin with. My bad!</p>
]]></description><pubDate>Tue, 11 Aug 2026 18:08:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49262188</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49262188</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49262188</guid></item><item><title><![CDATA[New comment by the_af in "Stealing Reasoning Traces from Proprietary LLM APIs"]]></title><description><![CDATA[
<p>Wow! I totally missed that the person I was replying to was the one who had submitted the link. What you describe is surely what happened.<p>I feel bad now :(</p>
]]></description><pubDate>Tue, 11 Aug 2026 18:07:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49262181</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49262181</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49262181</guid></item><item><title><![CDATA[New comment by the_af in "Stealing Reasoning Traces from Proprietary LLM APIs"]]></title><description><![CDATA[
<p>Why do you restate the abstract? Anyone can read it from the link.</p>
]]></description><pubDate>Tue, 11 Aug 2026 15:07:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49259581</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49259581</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49259581</guid></item><item><title><![CDATA[New comment by the_af in "As AI eats the web, the internet’s collective memory is disappearing"]]></title><description><![CDATA[
<p>Not necessarily. A shared reality is one where people are discussing the same thing (within a reasonable margin) even if filtered by their individual perceptions, and drawing different conclusions.<p>I don't know the word for the opposite of shared reality (maybe private individualized realities), but it results in a scenario where you have nothing in common with other people other than "I saw some thing" and therefore it becomes very difficult to discuss anything with them.</p>
]]></description><pubDate>Tue, 11 Aug 2026 15:02:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=49259526</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49259526</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49259526</guid></item><item><title><![CDATA[New comment by the_af in "Mea Culpa – Dark Hours"]]></title><description><![CDATA[
<p>If I understand the story correctly, it's even worse than "pivoted to astronomy". He used Claude to "accidentally" (doubtful) copy an existing astronomy app, bugs and all. It's like a pile of dishonesty upon dishonesty.</p>
]]></description><pubDate>Sun, 09 Aug 2026 15:55:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49232572</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49232572</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49232572</guid></item><item><title><![CDATA[New comment by the_af in "Mea Culpa – Dark Hours"]]></title><description><![CDATA[
<p>I've been thinking about this. Almost every way we have to tell human from AI apart can be adopted and automated by AI. It's sort of a corollary of "every metric can be gamed".<p>It's disheartening for those of us who intrinsically value human authorship.</p>
]]></description><pubDate>Sun, 09 Aug 2026 15:53:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49232550</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49232550</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49232550</guid></item><item><title><![CDATA[New comment by the_af in "Taste Is All That's Left"]]></title><description><![CDATA[
<p>> <i>Who the FUCK speaks like that?</i><p>Do you mean starting sentences with "But"? I do, but then again I'm not a native speaker and my writing may be nonstandard. Or do you mean short and direct sentences? In which case, Hemingway.<p>I'm not saying the article isn't AI slop, but I think it's not obvious from the specific sample you chose, since plenty of people online write like that, before AI.</p>
]]></description><pubDate>Thu, 06 Aug 2026 20:56:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=49202435</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49202435</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49202435</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>Ok, I'll bite!<p>I agree the theatrical cut is inferior, the voice over wasn't done willingly by Harrison Ford, the "happy ending" is forced and hideous, etc... but I kind like the hardboiled vibes of the narration. It was also the first version I watched so it has a special place in my heart.<p>The Director's Cut -- that I know isn't an actual director's cut -- is possibly my favorite. Sadly, some murky visuals and a couple of plot holes (mostly number of replicants), plus the Zhora stunt double... once you see her, you cannot unsee her, and you'll wonder how you ever failed to notice the stunt double looks nothing like Joanna Cassidy and <i>even the wig is wrong</i>. But this is the version that stuck with me, and the one with the best color grading.<p>Then there's a workprint with increased violence. I don't get why some fans prefer an extended scene of Roy Batty gouging Tyrell's eyes out... I don't think "more is better" applies to graphic violence; sometimes implied violence is better, and in any case showing this doesn't improve the character of Roy.<p>Finally... the Final Cut. I appreciate they fixed (most) plot holes, they FIXED ZHORA'S STUNT DOUBLE (the single best improvement) and the image quality is way, way better. However, I cannot forgive Ridley Scott for changing the color grading to a greenish tint; Blade Runner now looks closer to the Matrix, which is UNWELCOME. I liked the more bluish/brownish tint of the Director's Cut, and really dislike the greenish/orange-y nonsense of the Special Edition. Here you can see what I mean: <a href="https://www.youtube.com/watch?v=ocx-Ch3Pzg8" rel="nofollow">https://www.youtube.com/watch?v=ocx-Ch3Pzg8</a>).<p>My ideal edition of Blade Runner: the color grading of the Director's Cut, with the improved image quality and fixed plot holes (and Zhora's stunt double) from the Final Cut.<p>(And yes, I'm aware I haven't mentioned <i>all</i> the cuts, just the ones I know).</p>
]]></description><pubDate>Thu, 06 Aug 2026 17:55:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=49200037</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49200037</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49200037</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>Many Pratchett books aren't riffs on Tolkien at all. Someone recommended Small Gods (excellent choice, and not a whiff of Elves or Dwarves there) but many other Pratchett books are genuinely good and not cheap shots at Tolkien.<p>I say give it another try.</p>
]]></description><pubDate>Thu, 06 Aug 2026 17:47:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49199944</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49199944</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49199944</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>Very likely.<p>However, I see two problems with this: one, I think people are going to make comparisons to Matrix rather than Blade Runner, and judging from the trailer it looks a lot like a modern Matrix (understandably!). Second, Neuromancer starts with "the sky above the port was the color of television, tuned to a dead channel" so the grey/"not sunny" look is kinda baked in and hard to avoid ;)</p>
]]></description><pubDate>Thu, 06 Aug 2026 17:45:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=49199917</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49199917</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49199917</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>Donnie Darko is (to me) one of those movies ruined by the director's cut. Not only due to the music changes, but also because the "message" of the movie is made more explicit and on the nose. And reading Kelly's interviews and opinions is even worse. One of those movies where keeping the original, and the mystery, is absolutely mandatory.</p>
]]></description><pubDate>Thu, 06 Aug 2026 17:38:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49199835</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49199835</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49199835</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>Rachael smoking in that scene is almost like a painting, a masterpiece.<p>I love the whole movie and soundtrack with the same enthusiasm as you, it's one of my favorites, but that scene with Rachael is something else.</p>
]]></description><pubDate>Thu, 06 Aug 2026 02:20:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=49191662</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49191662</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49191662</guid></item><item><title><![CDATA[New comment by the_af in "The title cards in Blade Runner are amazing"]]></title><description><![CDATA[
<p>If T-1000 was an LLM:<p>"What's wrong with Wolfie?"<p>"Wolfie's fine."<p>"Actually his name's Max."<p>"You're absolutely right. It's not Wolfie -- it's Max."<p>"Actually we don't have a dog."<p>"I apologize. I shouldn't have assumed you have a dog. I will be more careful in the future. Do you want me to list common pets in American households?"<p>Terminator hangs up and tells John "You need a better model."</p>
]]></description><pubDate>Thu, 06 Aug 2026 02:00:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=49191516</link><dc:creator>the_af</dc:creator><comments>https://news.ycombinator.com/item?id=49191516</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49191516</guid></item></channel></rss>