tauon's recent activity
-
Comment on The coming loop in ~comp
-
Comment on Share your fav environmentally-friendly building tech! in ~enviro
tauon LinkThis is more of an indirectly/adjacent environmentally-friendly one, but you may be curious about Passive houses. Here’s a quite cool video showcasing some techniques which allow these houses to...This is more of an indirectly/adjacent environmentally-friendly one, but you may be curious about Passive houses.
Here’s a quite cool video showcasing some techniques which allow these houses to use little to no external energy while in use (so AFAICT this is mostly about after construction has finished). -
Comment on Zig creator weighs in on the Bun Rust rewrite in ~comp
tauon (edited )Link ParentApropos updates, here’s another nice commentary in the form of a third-party/outsider view: https://raymyers.org/post/zed-creator-calls-spade-a-spade/ (It does say zed in the URL instead of Zig,...Apropos updates, here’s another nice commentary in the form of a third-party/outsider view:
https://raymyers.org/post/zed-creator-calls-spade-a-spade/
(It does say
zedin the URL instead of Zig, not sure whether the author’ll fix that or keep it as it is, but it is about this topic regardless.)Rather than a real tradeoff comparison, we get a “Bun is better in Rust” section covering only upsides. A change like this always has trade-offs, an obvious one would be build time.
[…]
They also seem to be padding the list by mixing in other improvements they’ve made after the rewrite that aren’t really related to it.
And an interesting section I’m not gonna bother copying in its entirety is under the heading:
They didn’t try a style guide?
(Late edit: wasn’t sure whether this was worth a post of its own when I commented, but (see below) that decision was later made for me anyways. :p)
-
Comment on Blog owners, do you write under your real name, or under a pseudonym? in ~hobbies
tauon LinkAs someone who is going to launch a blog for the first time soon-ish (™️) too, I’ve handled this question with a mixed approach for now: The domain and blog’s “name” will be decoupled from me as a...As someone who is going to launch a blog for the first time soon-ish (™️) too, I’ve handled this question with a mixed approach for now: The domain and blog’s “name” will be decoupled from me as a person, however I will have to (and on the empty-but-live website already do) leave a note on my real identity, as that’s actually the law here in Germany (most small sites just choose to ignore it…).
In your situation, if you wouldn’t love talking about this e.g. in front of your friend circle or an extended family group gathering, I’d just decouple it at first.
You can always “re-publish” under your own name in a decade if you don’t care as much anymore – and if you still do, you’d probably be glad to have it not linked to your name in the first place.
As to what I’m planning to write about, I have a couple posts in a very early draft stage, I’m planning to write essentially about whatever bothers me, and hasn’t really been written about in one single (and easily findable/public!) place yet. Two of the three topics I’m more seriously thinking about writing are things that people tend to get wrong conceptually, so with a blog post on those topics I’d just have a ready-made, ideally well-sourced etc. breakdown I could link to.
Not sure whether I’ll link to (or outright post) any of them here, but I guess we’ll see. :D -
Comment on Rewriting Bun in Rust in ~comp
tauon Link ParentIt’s really notable, IMO, that the follow-up took so long to write and/or publish. Another indicator that this is a gigantic Anthropic marketing operation, if you ask me. (Not that that’s the only...It’s really notable, IMO, that the follow-up took so long to write and/or publish.
Another indicator that this is a gigantic Anthropic marketing operation, if you ask me. (Not that that’s the only reason for the port/rewrite or taking a while to write a blog post, but probably a major factor in both nonetheless.) -
Comment on What programming/technical projects have you been working on? in ~comp
tauon Link ParentMost definitely! I’ve believed for some time now that TypeScript has the potential to become a new “default” choice for cross-platform applications (at least for ones where performance doesn’t...Does anybody see value in making static binary from TypeScript?
Most definitely! I’ve believed for some time now that TypeScript has the potential to become a new “default” choice for cross-platform applications (at least for ones where performance doesn’t decide the programming language), mainly for CLI tools (that potentially do things over the network) and maybe some light GUI stuff.
Basically the space that used to be (somewhat? not sure) filled by Python before it became a packaging mess – and would basically require users to have e.g. Astral’suvinstalled.For what it’s worth, I really like your approach with a “custom” IR, this is very interesting stuff.
More proof that demand or at least a baseline of interest is there, by the way, would be Bun’s compilation feature, which also enables static/standalone/non-browser non-Node builds. But if I’m not mistaken, that “just” ships a Bun runtime bundled into the binary, which is probably the simpler approach… but also the less cool one from a computer science perspective :-)
Like, your idea has more steps for sure, but somehow still feels “conceptually” cleaner. And if (I’m guessing) performance gains are an added bonus, even better! -
Comment on What programming/technical projects have you been working on? in ~comp
tauon Link ParentInteresting! Where does Heavy integrate? On each server hosting a web server/serving sites or per-project configuration?Interesting! Where does Heavy integrate? On each server hosting a web server/serving sites or per-project configuration?
-
Comment on I made a satirical AI detector in ~tech
tauon Link ParentIt really is ridiculous. A while ago, I was held for a bot for knowing about the almighty <hr> (horizontal rule/line), which many forum-like platforms seem to implement. … What a shame.It really is ridiculous. A while ago, I was held for a bot for knowing about the almighty
<hr>(horizontal rule/line), which many forum-like platforms seem to implement.
… What a shame.
-
Comment on I'm looking for an adage or "law" (like Conway's law), but for dealing with AI slop in ~tech
tauon Link ParentTangentially related to the idea of the effort put into something, allegedly Epstein made a remark about Trump saying he "is the only person he knows who has 'written' more books than he's read"....Tangentially related to the idea of the effort put into something, allegedly Epstein made a remark about Trump saying he "is the only person he knows who has 'written' more books than he's read".
But I'm not sure anything should be named after either of these people… -
Comment on No, artificial intelligence is not conscious in ~tech
tauon Link ParentOh boy, I get to recommend the Manna story again! I try to bring it up every so often since it’s just so good, and potentially more relevant to humankind’s trajectory now than ever before....I'm having trouble thinking of a sci-fi setting where a post-scarcity economy is portrayed poorly.
Oh boy, I get to recommend the Manna story again! I try to bring it up every so often since it’s just so good, and potentially more relevant to humankind’s trajectory now than ever before.
Specifically, the first half/section of the (free) e-book is about, essentially, a post-scarcity economy that’s not literal utopia.The author in that case was thinking more of “robots” as a driving force behind workforce replacement rather than “AGI”, but the ideas hold up regardless. (Plus, LLMs are of interest to robotics research now, so it might still play out like described in Manna regardless.)
-
Comment on If AI is sentient then so is ‘Age of Empires II’ in ~tech
tauon (edited )Link ParentIt matters…, it matters a whole lot in my opinion! I see where you’re coming from with it, but I don’t think this idea generalizes to a (potential) future time full of technological advances,...That being said, maybe it doesn't really matter? If something is complex enough that it can do useful stuff, then whether it is conscious is really more of a philosophical / ethical question.
It matters…, it matters a whole lot in my opinion!
I see where you’re coming from with it, but I don’t think this idea generalizes to a (potential) future time full of technological advances, under the assumption that we want – at minimum – human beings to be governed under some general principles of nicety, such as the UN's universal human rights declaration.
This was probably a bit abstract. For one example of what I’m talking about: MMAcevedo is a sci-fi-horror “Wikipedia“-style entry, but regardless of that, I’ve turned into a stark opponent of most anything related to brain-scanning and -uploading (whereas before reading it I would’ve considered myself to be neutral towards, maybe even curious about such a future technology). Being sapient or sentient suddenly matters a lot more if you could be made to experience orders of magnitude more things (hours, tasks, emotion, pain signals) than any biological organism was ever meant to live through. That’s dangerous territory!
Now consider this thought experiment: What happens if one day, we actually succeeded in producing working/runnable brain scans, and it hypothetically were to turn out that their internal structure can be transformed to be basically an identical match to how LLMs’ matrices work and look like – would we still feel comfortable sending these machines out to do “useful stuff“ for us, with no limitation on the content or duration of work done?
Could we still treat machines that, while built in a wholly different way, internally function just like a “human [scan]“ as non-sentient? Or, reverse that and conclude that not even humans deserve the rights we fundamentally admitted to sentient/sapient beings? -
Comment on Signs you're a dangerous terrorist: using Signal, moving zines in ~society
tauon LinkReason no. 1358 I can’t and won’t risk traveling to the U.S. – as a tourist or worker – under the current administration, unfortunately, maybe even longer. I’m not a terrorist, yet I’ll continue...Reason no. 1358 I can’t and won’t risk traveling to the U.S. – as a tourist or worker – under the current administration, unfortunately, maybe even longer.
I’m not a terrorist, yet I’ll continue to use Signal to communicate with friends and family. Incomprehensible to the average Trump-loyal prosecutor from Texas (and a bunch of other states just as bad in that regard), it seems.
-
Comment on What about having an LLM teach you to code? in ~comp
tauon Link ParentYou might be interested in Matt Pocock’s teach skill as well. I haven’t personally tried it yet, but it looks quite promising. There is also an accompanying video which introduces the skill a bit...You might be interested in Matt Pocock’s
teachskill as well. I haven’t personally tried it yet, but it looks quite promising.There is also an accompanying video which introduces the skill a bit more.
-
Comment on Does generative AI have a natural limit without a major innovation? in ~comp
tauon (edited )Link ParentWithout delivering concrete proof here, I am fairly certain most major American labs, and for sure (like, confirmed by them) some of the Chinese labs known for “distillation” work (Moonshot,...Without delivering concrete proof here, I am fairly certain most major American labs, and for sure (like, confirmed by them) some of the Chinese labs known for “distillation” work (Moonshot, Zhipu/Z.ai, DeepSeek, Alibaba’s Qwen), are already using synthetic training data, which is to mean data originally produced by an LLM (or a specific/deterministic code-driven process), and then (eventually) fact-checked and/or refined by a human.
It’s worked pretty well so far in the cases that were published, for example Moonshot’s Kimi K-model series:
[Step] 4. Simulate Usage of the Synthetic Agents: The team simulated multi-turn tool-use scenarios in order to generate “trajectories” – a fancy way of saying the detailed set of steps documenting the inputs and steps models take to accomplish their goals. Some of these scenarios simulated “users” – fake people with diverse communication styles – interacting with these agents, while others simulated autonomous usage.
Edit: This is not to say I believe synthetic training data, for LLMs specifically, will get us to “AGI”/further-than-human intelligence. I’m sure there’s an inherent quality ceiling we’ll encounter somewhere.
-
Comment on Does generative AI have a natural limit without a major innovation? in ~comp
tauon (edited )Link ParentTo pile onto that, ErdosBench(mark) is a thing now, comparing multiple models’ performance on (previously not published) adaptations of Erdős problems.To pile onto that, ErdosBench(mark) is a thing now, comparing multiple models’ performance on (previously not published) adaptations of Erdős problems.
-
Comment on Arch User Repository compromised, 1500+ packages affected in ~tech
tauon Link ParentDoes the AUR provide a mechanism to determine package/PKGBUILD (if even applicable?) “orphanaged status” (or if the user providing PKGBUILD recently changed)? That way, in a next step users could...As a user, uninstalling any orphan packages is probably for the best.
Does the AUR provide a mechanism to determine package/
PKGBUILD(if even applicable?) “orphanaged status” (or if the user providingPKGBUILDrecently changed)?That way, in a next step users could set up rules for e.g. warning, then blocked execution for orphans over a given number of months (or the reverse and requiring a cooldown-before-use for the case of submitter-not-maintainer user having changed).
-
Comment on Access to Fable and Mythos 5 cut off after US government order in ~tech
-
Comment on Any fellow software engineers using paid GitHub copilot? in ~comp
tauon Link ParentThis is the part I don’t understand about the current “AI push”: Isn’t awarding employees by most LLM (token) spend just “paid by LOC” all over again? Has management (in the aggregate) really...Coworkers were actively tracked down and applauded for being top AI spenders.
This is the part I don’t understand about the current “AI push”: Isn’t awarding employees by most LLM (token) spend just “paid by LOC” all over again?
Has management (in the aggregate) really learned nothing, or alternatively, forgotten every lesson from that time?I mean, even if you’re not familiar with the SWE side of this (hi)story, maybe you ought to have heard of Goodhart’s law instead if you’re in a position to green-light these massive budgets with absolutely no way of tracking, let alone predicting them?
I know I’d be uncomfortable if my company could just sack my role as the responsible person once things went south. -
Comment on Claude Fable 5 and Claude Mythos 5 in ~tech
tauon Link ParentMea culpa, that’s a great point. If the goal was a very large, general-purpose model, I imagine research that’d come out of potentially resource-constrained environments [in comparison to the “AI”...Mea culpa, that’s a great point.
If the goal was a very large, general-purpose model, I imagine research that’d come out of potentially resource-constrained environments [in comparison to the “AI” big players] would focus on algorithmic and size optimization, regardless. It’s still cutting-edge after all, just potentially at a smaller scale and/or helping achieve smaller hardware requirements.(As an aside, I’m a little embarrassed, but I just now realize the same is actually true for my alma mater: they have both the huge data/computing center as well as a nuclear reactor, too – although I’m not sure how much free computing capacity could be scheduled there towards an intensive task like (very) large language model training.)
-
Comment on Claude Fable 5 and Claude Mythos 5 in ~tech
tauon (edited )Link ParentIt’s an interesting thought, but with what funding? Even excluding the knowledge and skills necessary, the hardware requirements alone probably already halt this idea. Edit: Not even necessarily...It’s an interesting thought, but with what funding? Even excluding the knowledge and skills necessary, the hardware requirements alone probably already halt this idea.
Edit: Not even necessarily the GPUs alone are a bottleneck, just (ideally redundant) storage for the raw training data would already be prohibitively expensive for most entities or orgs entertaining this idea…
Haven’t gotten around to reading the article itself yet, but FYI: While Armin Ronacher’s blog is (I think?) receiving more coverage (and new entries) nowadays, Mario Zechner aka
badlogicis actually the person you’d call “the” developer of Pi.Armin is of course the a long-time power-user, prominent contributor/co-maintainer (and now a co-owner via Earendil), but he is not the original or primary developer, even though I believe he was also actually “there” already when Pi was first written/conceptualized.