<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: dTal</title><link>https://news.ycombinator.com/user?id=dTal</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 16 Nov 2026 08:10:31 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=dTal" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by dTal in "Weight Space Misfolding"]]></title><description><![CDATA[
<p>In other words - LLMs are developing their own culture.</p>
]]></description><pubDate>Sun, 15 Nov 2026 12:14:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=50506378</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50506378</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50506378</guid></item><item><title><![CDATA[New comment by dTal in "Laser-armed AI submarine robots patrol Norwegian salmon farms, killing sea lice"]]></title><description><![CDATA[
<p>See "Itself" by AE Van Voght:<p><a href="https://archive.org/details/Gamma_01_1963-07/page/n39/mode/1up" rel="nofollow">https://archive.org/details/Gamma_01_1963-07/page/n39/mode/1...</a></p>
]]></description><pubDate>Sat, 14 Nov 2026 12:19:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=50496453</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50496453</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50496453</guid></item><item><title><![CDATA[New comment by dTal in "Physically Controllable Relighting of Photographs"]]></title><description><![CDATA[
<p>Only if you think that progress comes from divine inspiration in the ineffable depths of the minds of great men, as opposed to smart people applying themselves to problems using the tools they have available. A lot of new ideas aren't so much someone figuring out how to do something, as someone deciding that something is <i>worth doing</i>. Amazon's infamous patent on "1-click purchasing" wasn't some kind of flying car that the world would have been forever robbed of, had Jeff Bezos not been incentivized to disclose his monumental trade secret.<p>tl;dr NO</p>
]]></description><pubDate>Thu, 12 Nov 2026 13:13:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=50466628</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50466628</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50466628</guid></item><item><title><![CDATA[New comment by dTal in "Hacking the Meta Ray-Ban smartglasses"]]></title><description><![CDATA[
<p>Do I really "have root" if you've put an "unavoidable warning" in the bootloader that I can't remove? It sounds like you're still asserting primary control over the device and giving me a limited playground. Your suggestion is "we're going to run code you don't control on your device, and it's going to run before any of your code, and there's nothing you can do about it, but don't worry we pinky promise to only make it print a warning scolding you for wanting root."<p>Nah. No thanks. I want actual freedom, not a longer chain.</p>
]]></description><pubDate>Thu, 12 Nov 2026 10:21:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=50465303</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50465303</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50465303</guid></item><item><title><![CDATA[New comment by dTal in "Hacking the Meta Ray-Ban smartglasses"]]></title><description><![CDATA[
<p>The point at which we are performing unnecessary surgery to remove perfectly functional assistive technology from people because of some legal fantasy is the point where we need to seriously ask ourselves whether the legal fantasy is really doing anything useful, or whether it's merely some self sustaining cancer on society.</p>
]]></description><pubDate>Thu, 12 Nov 2026 10:14:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=50465267</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50465267</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50465267</guid></item><item><title><![CDATA[New comment by dTal in "Hacking the Meta Ray-Ban smartglasses"]]></title><description><![CDATA[
<p>You're not likely to physically damage a computer by running your own software on it. There's no excuse for not having an easy "reset to factory settings" and a bootloader that's hard to brick. Anything with an SD card slot or a USB port can boot from it; anything with wifi can download a fresh factory image; anything at all can have a "recovery mode" that you don't need to touch to tinker with it. Tinkering is only dangerous when the manufacturer makes it so.</p>
]]></description><pubDate>Thu, 12 Nov 2026 10:11:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=50465235</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50465235</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50465235</guid></item><item><title><![CDATA[New comment by dTal in "Drones built to look like birds evaded detection at Air Force tech exercise"]]></title><description><![CDATA[
<p>A popular approach right now is to have a quadcopter arrangement on a large fixed wing, which can either fly normally or even spin to make a large rotor.</p>
]]></description><pubDate>Wed, 11 Nov 2026 18:06:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=50456463</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50456463</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50456463</guid></item><item><title><![CDATA[New comment by dTal in "The Web Is More Than Enough"]]></title><description><![CDATA[
<p>(Free software) native apps are preferable to the web. I won't install most (proprietary) native apps.<p>Hope this helps.</p>
]]></description><pubDate>Wed, 11 Nov 2026 14:00:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=50453525</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50453525</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50453525</guid></item><item><title><![CDATA[New comment by dTal in "Ways You Can Help Linux Grow (Even If You Don't Know Coding)"]]></title><description><![CDATA[
<p>Oh I'm sure there will be plenty of people who don't know coding helping Linux grow from now on, thanks to LLMs.</p>
]]></description><pubDate>Wed, 11 Nov 2026 13:50:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=50453404</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50453404</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50453404</guid></item><item><title><![CDATA[New comment by dTal in "*sshole, or when are we going to get rid of C?"]]></title><description><![CDATA[
<p>>It is perfectly possible for systematically poor hiring standards or systematically poor ability to fire incompetents, to be the determined root cause of a root cause analysis<p>Yes, exactly. You're agreeing violently. "Systematically poor hiring standards" is not operator error, it is an upstream cause of operator error.</p>
]]></description><pubDate>Sun, 08 Nov 2026 11:53:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=50412479</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50412479</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50412479</guid></item><item><title><![CDATA[New comment by dTal in "AI is dramatically accelerating AI research"]]></title><description><![CDATA[
<p>I think that is a fascinating question and I would like to advance the provocative notion that N is much smaller than you think, maybe <256. Consider that 128 bits is a cryptographically secure key, and English text has an entropy of around  half a bit to a bit per character. 256 characters is sufficient to encode sentences that have never been seen or imagined before, totally unguessable sentences. Who knows what's in that probability space?</p>
]]></description><pubDate>Sat, 07 Nov 2026 19:09:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=50407103</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50407103</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50407103</guid></item><item><title><![CDATA[New comment by dTal in "AI is dramatically accelerating AI research"]]></title><description><![CDATA[
<p>Uh this seems exactly backwards - reinforcement learning with human feedback teaches <i>only</i> the power of persuasion. The only way to get rewarded is to persuade a human to reward you. Why do you think the models are <i>all</i> so damn sycophantic?</p>
]]></description><pubDate>Sat, 07 Nov 2026 19:00:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=50407034</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50407034</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50407034</guid></item><item><title><![CDATA[New comment by dTal in "So my ISP was hijacking my DNS"]]></title><description><![CDATA[
<p>Insanity that people used to pay more for goofy little midi files than they do now for Netflix.</p>
]]></description><pubDate>Thu, 05 Nov 2026 14:49:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=50379777</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50379777</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50379777</guid></item><item><title><![CDATA[New comment by dTal in "Tell HN: Gemini trains on chat history, even for subscribers (personal accounts)"]]></title><description><![CDATA[
<p>Funny, it works just fine for cruise missiles.<p>And on phones. In my experience, a hot start fix (with up to date ephemeris) shouldn't take more than a couple seconds. Maybe you have a bad antenna.</p>
]]></description><pubDate>Thu, 05 Nov 2026 14:41:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=50379665</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50379665</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50379665</guid></item><item><title><![CDATA[New comment by dTal in "Tell HN: Gemini trains on chat history, even for subscribers (personal accounts)"]]></title><description><![CDATA[
<p>You know what makes discovery difficult? A blank page</p>
]]></description><pubDate>Thu, 05 Nov 2026 14:37:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=50379614</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50379614</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50379614</guid></item><item><title><![CDATA[New comment by dTal in "A thousand fewer paths to root"]]></title><description><![CDATA[
<p>We keep paying ever higher architectural prices for system isolation: user accounts, memory protection, unix permissions, capabilities... and it's all for nothing, when a determined program can almost always escape its sandbox. Should we just give up and assume that all software on a machine is cooperative and non-malicious, and just do our best to make sure we don't run malware in the first place? "Defense in depth" soon slips into "don't worry too much about this layer, the next layer will catch it.<p>tl;dr Just run as root bro, focus on the <i>exterior</i> attack surface, and don't download random crap.</p>
]]></description><pubDate>Thu, 05 Nov 2026 13:27:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=50378816</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50378816</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50378816</guid></item><item><title><![CDATA[New comment by dTal in "Vathographs: Photographic depth maps as creative expression"]]></title><description><![CDATA[
<p>I feel like you are saying "deterministic", when what you really mean is "predictable". They're <i>all</i> algorithms, just a bunch of multiplies and adds on the image data really. But the algorithms you don't mind are the ones where you feel you have a good intuitive grasp of what they'll do. The ones that seem eerie and break your sense of indexicality are the ones where 1) you're not sure  what it will do, and relatedly 2) seem to introduce a substantial amount of new information, perhaps drawn from some database of how images look generally.<p>To sum up: Marigold V2 is just as deterministic as Bayer demosaicing. But Marigold has about 20 billion magic numbers in it, thoroughly dwarfing the source image. It feels less like transforming the image with Marigold, and more like steering Marigold with the image. And who <i>knows</i> what's in that thing or how it works...<p>(note: I see the same confusion often in conversations about LLMs. People call them "nondeterministic", as if that were some new category of software. But, with temp 0, they're no less deterministic than any other pile of floating point math - and nobody complains that fluid simulations aren't deterministic. No, what they are is unpredictable.)</p>
]]></description><pubDate>Wed, 04 Nov 2026 20:20:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=50370935</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50370935</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50370935</guid></item><item><title><![CDATA[New comment by dTal in "Synaptics Appears to Use Legal Threats to Block Open-Source DisplayLink Driver"]]></title><description><![CDATA[
<p>How can they simultaneously argue that 1) this is a patent infringement, the entire legal basis of the claim, and 2) it is a unlicensed disclosure of proprietary technology? The whole point of patents is to disclose technology.</p>
]]></description><pubDate>Wed, 04 Nov 2026 19:55:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=50370660</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50370660</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50370660</guid></item><item><title><![CDATA[New comment by dTal in "Vathographs: Photographic depth maps as creative expression"]]></title><description><![CDATA[
<p>In that case, most of the information in every single color digital photo is "hallucinated", because each sensor pixel only captures red, green, or blue, in a checkerboard pattern called a Bayer array; we get uniform RGB photos by applying heuristic interpolation algorithms to (shall we say?) "infer" the missing data.</p>
]]></description><pubDate>Tue, 03 Nov 2026 13:36:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=50351822</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50351822</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50351822</guid></item><item><title><![CDATA[New comment by dTal in "Vathographs: Photographic depth maps as creative expression"]]></title><description><![CDATA[
<p>Is "deterministic" really the right line to draw? On the one hand, this <i>is</i> "deterministic", in the sense that (given a fixed model) the same photo will produce the same depth estimation. On the other hand, even <i>film</i> isn't deterministic, with its random grain. But at least random grain <i>looks</i> random. Perhaps the issue at play is really "indexicality", where perceptible signal should be <i>caused by</i> the scene. But even that is fuzzy. What, <i>qualitatively</i>, is the difference between "touching up" and AI enhancement? It's all just algorithms that combine a starting photo with some knowledge of how humans like photos to look. Again, not even film is free of this - when chemically developing, the choice of how strong a developer to use and how long to leave it in has as much effect on the image as the exposure settings chosen by the photographer. There's nothing "deterministic" about a lab technician eyeballing your photos and tweaking the development process.</p>
]]></description><pubDate>Tue, 03 Nov 2026 12:58:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=50351458</link><dc:creator>dTal</dc:creator><comments>https://news.ycombinator.com/item?id=50351458</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50351458</guid></item></channel></rss>