HNHacker News
TopNewBestAskShowJobs

jstummbillig

6,608 karma · joined February 14, 2015

submissionscomments
jstummbillig··on China's open-weights AI strategy is winning
It's interesting that building models goes one of two ways: Either you do it on your own (with data from debatable sources, maybe) or you do it by using a model that did it with data from debatable sources.

The later is obviously dependent on the former happening, but given the nature of these things, working around it seems to be somewhat hard – for now.

What happens, though, when frontier models become far less public? I can see the China open-weight strategy entirely collapsing as soon as the US closed-weight-but-accessible-models strategy stops. Hard to say how much they lean on it right now.

jstummbillig··on Airport Simulator
Mini Airways - ATC simulator comes close in base mechanics but adds some complexity

https://store.steampowered.com/app/2289650/Mini_Airways__ATC...

jstummbillig··on Claude Code uses Bun written in Rust now
> But the whole point of their product is that it supposedly nullifies such "business" concerns around the use of technology, by making it cheap and fast to build whatever you like automatically.

Eventually? I am sure they would agree. Currently it's you (and lots of people like you) who are doing the supposing, not Anthropic.

jstummbillig··on What I learned selling 2,500 MIDI recorders: Hardware is not so hard
> Have a strong anti-counterfeit strategy. Don’t overlook this.

Would love to hear more on that but I suppose the omission is not an accident.

jstummbillig··on The Kimi K3 Moment
It's more simple: They infringe on the IP by way of violating the ToS. If you violate ToS and the company suffers financial harm, they usually can (usually) sue you in civil court for damages.
jstummbillig··on The Kimi K3 Moment
> Distillation “attacks” are not attacks.

If "distillation attacks" happen, we have to conclude there is some value add in what model labs do. Regardless of how we feel about using existing human knowledge in the way they currently do, it's simply impractical to infer that everything that happens downstream of LLMs can not be an attack on some IP because of it.

So both things can be true: a) People infringe on Anthropics IP and b) what Anthropic did to build their models is legally questionable (or might be ruled illegal, even though I doubt it).

jstummbillig··on Show HN: IKEA Complexity Index
Codex inferred Assembly minutes = round(0.44 * steps + 0.70 * parts) from the data.
jstummbillig··on Show HN: IKEA Complexity Index
Cool! I had codex have a go at calculating "effective price" on a few items, that prices in the assembly labor, and then also calculates the delta between price and effective price. Obviously very depended on how you value your time, but here are some items at $30/hour:

BRIMNES storage bed + headboard, Queen $549 + 320 min labor = $709 (delta: +$160 / +29%)

HEMNES 8-drawer dresser $380 + 236 min labor = $498 (delta: +$118 / +31%)

STORKLINTA 6-drawer dresser $250 + 224 min labor = $362 (delta: +$112 / +45%)

SLÄKT storage bed, Twin $450 + 212 min labor = $556 (delta: +$106 / +24%)

BRIMNES 3-door wardrobe $250 + 189 min labor = $344.50 (delta: +$94.50 / +38%)

ALEX drawer unit: $95 + 96 min labor = $143 (delta: +$48 / +51%)

BRIMNES cabinet with doors: $99 + 75 min labor = $136.50 (delta: +$37.50 / +38%)

KALLAX 2x4 shelf unit: $65 + 39 min labor = $84.50 (delta: +$19.50 / +30%)

Formula: effective price = sticker price + (estimated assembly minutes / 60 * hourly value of time).

I was surprised by how similar the % diff is across the board.

jstummbillig··on Apple targets dozens of OpenAI employees with legal letters
Is this uncommon when it comes to corporations? Sue the people who (allegedly) did the thing?
jstummbillig··on Apple targets dozens of OpenAI employees with legal letters
What do you mean "this far"? How far is this?

Corps lose law suits all the time. They always have to go whatever "this far" is before it happens, surely?

jstummbillig··on $100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol
> A talented creative with a vision could make something more interesting and enjoyable in an afternoon with a $0 budget.

Only if you are really sloppy with your book keeping.

jstummbillig··on OnePlus halts operations in USA and Europe
I mean... the major change is that it changed, no? It's kind of unprecedented, or, at least, highly unusual?
jstummbillig··on We don't use AI in any of our design or production processes
Everywhere, presumably, not vocal, doing other things.
jstummbillig··on Ask HN: Does anyone let AI agents play games just for fun?
There is a world. In that world data centers are not a net problem, but net solution.

Looking at car exhaust and concluding that cars must be bad would be a questionable conclusion. We tolerate them despite the pollution they cause, not for it.

The same can go for data centers, again, if you conclude that they are net solution. If you don't, then it won't for you. The market simply disagrees with you in that case.

jstummbillig··on Are we offloading too much of our thinking to AI?
I have no idea what people are doing. I am thinking more about harder problems than ever before.

It goes like: "Here is this thing I wonder about", and the LLM is like "Yeah sure, consider these things that are super related to what you are doing, that you probably know nothing about yet (but you know... if you are interested...)".

And that goes in any direction, for any depth. Anything that is made trivial now, is just replaced by something more consequential a level or two higher. You can just get much better at things that matter more.

jstummbillig··on Codex starts encrypting sub-agent prompts
> protect what they see as the moat, that a model is "good at spawning sub-agents"

Yes, that is the obvious answer. I was looking for an explanation as to why and why now. Codex is open source after all. They used to not do it. Agent prompts more generally are also not encrypted, and continue to be.

This particular change just looks unintuitive to me.

jstummbillig··on Codex starts encrypting sub-agent prompts
What's the idea here? Why does this seem important to OpenAI?
jstummbillig··on European "age verification" "app" forcing everyone to use Android or iOS
> Of course Roblox have more player, but does your child really need millions of players?

Nobody uses platforms because they are are looking to exercise billions of options. The point is easy commonality. You sit next to a kid, and, what do you know, they are into Roblox too. Cool. Wanna play?

jstummbillig··on European "age verification" "app" forcing everyone to use Android or iOS
Pretending that those options are equal is a false dichotomy. Not participating is an option up to a point, and then it is increasingly limiting all other options.
jstummbillig··on Ask HN: Does anyone let AI agents play games just for fun?
"waste"

All of a sudden we are selectively squeamish with computer resource usage, when we were fine having all that fun with computers and hardware, 3 monitor setups, using graphic cards to play games (dear lord!) and tinkering around with home rigs of every proportion and wattage for no reason at all.

jstummbillig··on Chatto is now open source
The anxious pendulum swings hard between "show me something real that is well done" and "what is interesting about it, that's what everyone does".
jstummbillig··on FCC approves test of space mirror to light night sky
Oh, absolutely. The space mirror idea might be a net terrible idea; more panels and better batteries could simply be it.

I just don't see why, if this turns out to work, the benefits would not diffuse (principally because I don't see a line of previous innovations where this was not the case, electricity being a prime example of an extremely potent and well diffused commodity).

jstummbillig··on FCC approves test of space mirror to light night sky
> Anyway, even if it proves useful for something, it would join the long list of innovations doing one thing at the expense of everyone else (externality == light pollution in this case).

That's a strange reading. What would keep the benefits from cheaper electricity from diffusing to roughly everyone, as they usually do?

jstummbillig··on Companies are scrambling to curtail soaring AI costs
How does that concern individual companies as AI consumers? The hyperscalers have no trouble utilizing their GPUs (and I would suspect they mathed it out to be profitable but, again, that's just a different topic).
jstummbillig··on Apple sues OpenAI, accuses ex-employees of stealing trade secrets
That was a wild read.
jstummbillig··on Companies are scrambling to curtail soaring AI costs
I suspect this will turn out to be a super overblown issue: AI spend is literally the easiest spend to regulate in the entirety of businesses. No machinery is grinding to a halt over it, no asset that had to be bought and is now useless. You don't even have to employ or fire staff to give it a go (of course, you can still do both for other reasons). There are a lot of options that you can try out and substitute for each other, as new stuff comes up, because most things are compatible.

Sure, if you start at this point, where a good chunk of employees, who never had that ability, can now spend a lot of money at their discretion, that's probably going to be costly at first. Then people will learn from that and set direction adn guardrails.

jstummbillig··on GLM 5.2 is nearly as accurate as a human book keeper
1 Just as with traditional non-ai ways of doing things, there is a deterministic layer that can trigger when things are off. This can range from letting the ai know to stopping before a human confirms.

2 Usually, specially for SMBs, nobody goes to prison over accounting errors. That's because SMB owners make mistakes all the time and, because of that, authorities are fairly practiced in understanding how mistakes look, and how fraud looks.

jstummbillig··on GPT-5.6
> maybe ultra is not good value regarding tokens?

Well, yes, as explicitly stated on https://openai.com/index/gpt-5-6/: "ultra goes further by coordinating four agents in parallel by default, trading higher token use for stronger results and faster time-to-result on demanding tasks."

jstummbillig··on GPT-5.6
Why would you need a guide for that now? We long had to pick different models (and thinking levels) by task and feel.
jstummbillig··on GPT-5.6
"GPT‑5.6 is available starting today across ChatGPT, Codex, and the OpenAI API. The rollout is starting globally now and will continue gradually toward full availability over the next 24 hours."
← PreviousPage 8 of 34Next →