<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: mindwok</title><link>https://news.ycombinator.com/user?id=mindwok</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 17 Aug 2026 08:51:50 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=mindwok" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by mindwok in "What happens when an LLM never sees material beyond fifth grade?"]]></title><description><![CDATA[
<p>Maybe that'd work, but I think it'd come across too mechanical. If it was going to refuse something it'd need to be congruent with its "personality" I think.</p>
]]></description><pubDate>Sun, 16 Aug 2026 08:31:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=49318079</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49318079</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49318079</guid></item><item><title><![CDATA[New comment by mindwok in "What happens when an LLM never sees material beyond fifth grade?"]]></title><description><![CDATA[
<p>Yep, agree, very succinct way of describing my issue with it.</p>
]]></description><pubDate>Sun, 16 Aug 2026 08:31:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=49318077</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49318077</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49318077</guid></item><item><title><![CDATA[New comment by mindwok in "What happens when an LLM never sees material beyond fifth grade?"]]></title><description><![CDATA[
<p>True, but fancy autocomplete keeps exceeding my expectations in what it can do, so why not this one!</p>
]]></description><pubDate>Sun, 16 Aug 2026 08:30:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=49318072</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49318072</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49318072</guid></item><item><title><![CDATA[New comment by mindwok in "What happens when an LLM never sees material beyond fifth grade?"]]></title><description><![CDATA[
<p>Something related I've been thinking about lately is that one of the biggest problem with LLMs is their seeming inability to say no. Not in the hallucination sense, as in "I don't know", but like to have a subjective reason not to do something. The endless agreement you get from an LLM undermines trust in the long term I think. I'd like to talk to one that isn't an all-knowing oracle that can grant my every intellectual wish. (Or maybe what I'm asking for is just... a human, lol).</p>
]]></description><pubDate>Sun, 16 Aug 2026 08:06:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49317926</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49317926</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49317926</guid></item><item><title><![CDATA[New comment by mindwok in "Understanding is the new bottleneck"]]></title><description><![CDATA[
<p>Conversely I'm a solo-dev and I ship a lot of slop I don't even look at. Granted my work is just basic CRUD apps, and I focus my efforts on validating important consequences (like does this break accounting invariants or something)</p>
]]></description><pubDate>Fri, 14 Aug 2026 06:22:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=49295306</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49295306</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49295306</guid></item><item><title><![CDATA[New comment by mindwok in "Understanding is the new bottleneck"]]></title><description><![CDATA[
<p>Aren't you just taking the same dismissive attitude towards management and doing it to engineers?</p>
]]></description><pubDate>Fri, 14 Aug 2026 06:20:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=49295296</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49295296</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49295296</guid></item><item><title><![CDATA[New comment by mindwok in "GLM-5.3: Frontier coding with emergent cyber capabilities"]]></title><description><![CDATA[
<p>At least OpenAI seems to want to do that, but the US is now forcing them to go through approvals. Anthropic seems much more hesitant.</p>
]]></description><pubDate>Fri, 14 Aug 2026 06:17:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=49295283</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49295283</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49295283</guid></item><item><title><![CDATA[New comment by mindwok in "Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD"]]></title><description><![CDATA[
<p>It’s not just speed, it’s also the rich feature set. Trigram search, JSONB, transactions, GIN indexes, full text search - not many databases combine them all into a Swiss Army knife like Postgres does.</p>
]]></description><pubDate>Sat, 08 Aug 2026 11:45:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=49220885</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49220885</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49220885</guid></item><item><title><![CDATA[New comment by mindwok in "Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD"]]></title><description><![CDATA[
<p>I don’t think this is true. There’s plenty of signals for trust - if pgrust keep pouring time, money, and effort into this and get people actually using it even in non critical workloads, it’ll snowball from there. Imagine 2 years from now it’s being used in a couple dozen large companies - people wouldn’t think twice about using it.</p>
]]></description><pubDate>Sat, 08 Aug 2026 11:42:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49220873</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49220873</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49220873</guid></item><item><title><![CDATA[New comment by mindwok in "Qwen3.8 Max now ranked as the best overall model by agentic index"]]></title><description><![CDATA[
<p>For me now it’s simply cost and speed. With GPT5.6 and Fable (and respective open models since then) we passed a threshold where intelligence is sufficient. Now I just need speed of iteration and good prices.</p>
]]></description><pubDate>Thu, 06 Aug 2026 22:28:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=49203505</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49203505</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49203505</guid></item><item><title><![CDATA[New comment by mindwok in "LLMs won't break symmetric crypto"]]></title><description><![CDATA[
<p>> mathematical impossibility of reversing some integer/mod calculation
> There is a loss of information in mods and integer computations making them one-way<p>That's not correct. Trapdoor functions aren't one way because they destroy information, and if they were they wouldn't be very useful because you wouldn't be able to go back the other way (i.e. decrypt the text). You'd end up with many possible inputs for a given output, like a hash.</p>
]]></description><pubDate>Thu, 06 Aug 2026 04:22:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49192396</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49192396</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49192396</guid></item><item><title><![CDATA[New comment by mindwok in "I am retiring from fulltime writing (& pseudonymity) to launch Guardian Angel"]]></title><description><![CDATA[
<p>Yeah but... Both things are true. LLMs are doing things we would have considered absolute AGI god tier magic 5 years ago. They also still can't count to 100 yet.</p>
]]></description><pubDate>Wed, 05 Aug 2026 02:47:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=49178012</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49178012</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49178012</guid></item><item><title><![CDATA[New comment by mindwok in "libexpat now funded by the City of Munich for up to 6 months"]]></title><description><![CDATA[
<p>Whether they should or shouldn't is a worthy question that they have no influence over. Taking the initiative and supporting the software they rely on is a good step in the meantime while we argue on HN about whether it's optimal or not.</p>
]]></description><pubDate>Wed, 05 Aug 2026 02:42:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49177966</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49177966</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49177966</guid></item><item><title><![CDATA[New comment by mindwok in "How to Exist"]]></title><description><![CDATA[
<p>I think this makes as dominate other species even more. Society, and especially market capitalism, is like a giant optimization function for productivity. If you step off the treadmill, you will get dominated by those who are on it and have the latest ICBMs, AI models, infrastructure out the wazoo, can produce 10,000 drones a week, etc.</p>
]]></description><pubDate>Sat, 01 Aug 2026 06:09:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=49131555</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49131555</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49131555</guid></item><item><title><![CDATA[New comment by mindwok in "2x, not 10x: coding with LLMs in 2026"]]></title><description><![CDATA[
<p>I’ve been vibe coding apps professionally for the last year and 100% agree with your first take. I really underestimated how much work there is in what I just call the “coordination” of developing software.<p>Users pointing out things that are missing, or me forgetting to tell them I’ve added something, or both of us having a different idea in our minds of what something should do. It’s still extremely time consuming.</p>
]]></description><pubDate>Thu, 30 Jul 2026 23:48:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=49117317</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49117317</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49117317</guid></item><item><title><![CDATA[New comment by mindwok in "Starling: The first real desktop written by AI"]]></title><description><![CDATA[
<p>This post makes me feel grief. I mean no offence to you OP, this looks like a beautiful desktop and I am sure there was a lot of time put into this. But just sharing my grief.<p>3 years ago if I saw a post on the HN front page about a desktop environment written in Flutter (or a Swift port of Flutter), I'd be absolutely blown away.<p>Now, I see this, and I feel nothing. I'm not impressed. I don't want to know how it works. I'm not going to read 300k lines of AI slop code. I don't care about architectural decisions made because they're not made under any true constraints. I'm not even excited about the AI part of it, because I know AI can do this. We could all build this if we just prompt hard enough.<p>I'm tired, boss.</p>
]]></description><pubDate>Wed, 29 Jul 2026 12:27:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49096582</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49096582</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49096582</guid></item><item><title><![CDATA[New comment by mindwok in "Running Kimi K3 on a M1 Max"]]></title><description><![CDATA[
<p>Ok in terms of running a 2.8T parameter model, that's true.<p>Looking more broadly though, a model I can run on my laptop (Gemma 4) is ~4 points away from GPT-5.3 codex or Sonnet 4.5 on arena.ai LLM leaderboard. Those models were SOTA less than a year ago.</p>
]]></description><pubDate>Wed, 29 Jul 2026 03:30:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49093097</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49093097</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49093097</guid></item><item><title><![CDATA[New comment by mindwok in "Running Kimi K3 on a M1 Max"]]></title><description><![CDATA[
<p>Does anyone else feel like the writing is on the wall for a future of local models? Spamming data centres everywhere, powering them, having to commit insane capital to hardware, all the effort to serve inference over a network reliably - when here we are with a frontier model nearly running on a laptop.<p>Local AI on your device seems like a much more likely future to me than datacenters in space. For inference at least, training is another story.</p>
]]></description><pubDate>Wed, 29 Jul 2026 00:13:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=49091748</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49091748</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49091748</guid></item><item><title><![CDATA[New comment by mindwok in "Ask HN: Why is every company encorporating AI everywhere?"]]></title><description><![CDATA[
<p>It boils down to this: Nobody is willing to bet <i>against</i> AI right now.<p>If you're an investor, you're probably managing other people's money. Would you want to explain to your customers if & when you were wrong, why you didn't go all in on AI? This is the most transformative technology humanity has ever conceived and we've been making sci fi movies about for decades, and <i>didn't</i> put all our money into it?<p>So now that investor money is too scared to go anywhere else, that means if you're a CEO who cares about your share price, and who has to front up to the board every few weeks and defend your strategy lest they fire you, you must care about AI too. We aren't just a stiff old database anymore, noooo sir, we are the <i>context layer for agents</i>.<p>So now, if you're the CTO, are you going to tell the CEO that AI has no use case here? Not if you care about your job. Repeat, all the way down to the bottom.<p>There will be an emperor has no clothes moment. It might not be as dramatic as everyone is hoping, and it doesn't mean AI isn't world changing and is not here to stay, but the survival-induced psychosis at every level of the corporate world is absolutely unsustainable.</p>
]]></description><pubDate>Tue, 28 Jul 2026 10:00:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49081652</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49081652</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49081652</guid></item><item><title><![CDATA[New comment by mindwok in "Open-weight AI is having its Kubernetes moment"]]></title><description><![CDATA[
<p>Pretty close, they're very close to Heroku in spirit except the new wave of hosted backends (Vercel, and to a lesser extent Supabase and Cloudflare Workers) are much more tightly integrated with the app layer. You can almost think of them as libraries that run in the cloud which you can just use inside your app.</p>
]]></description><pubDate>Sun, 26 Jul 2026 04:14:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49054677</link><dc:creator>mindwok</dc:creator><comments>https://news.ycombinator.com/item?id=49054677</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49054677</guid></item></channel></rss>