rstuart4133

> This indicates you might be in a bubble.

He didn't say he was surprised by the options offered. He said he was reassured by what he considered to be the common sense option winning. As was I.

It didn't just win. Debian uses Condorcet voting, which does the equivalent of running lots of mini elections - pairing the options against each of the others in one on one contests. Close contests need a tie breaker mechanism as you get A beats B, B beats C, C beats A. Not this time - the winning option defeated all others in its one on ones.

I found that surprising. There are a few options close to option 5, the winner - only slightly less liberal. Effectively the most extreme option won, and not by a slim margin.

I, and I suspect the OP, wasn't surprised at the range of options offered. This is par for the course - Debian is a very robust democracy with its fair share of opinionated individuals. There has been a lot of noise about the LLM's. The surprising thing is what I regard as the common sense one was at one extreme, and that "extreme" position won easily.

I guess it's yet another illustration of the level of on online noise being a lousy indicator of what the normies are thinking. Yes, that's obvious, but when the level of noise is high it still can catch you by surprise.

Edit: The two most restrictive options were ranked below "None of the Above". That's the strongest rebuke a Debian GR can deliver to a proposal. Under Debians rules, if an option loses to "None of the Above" it can't win regardless of the outcome of the other mini elections. I don't think Debian could make it's position much plainer: LLM's are just another tool a developer can use at his discretion, and are to be treated no differently to any other tool.

> LLM outputs cannot be copyrighted.

For stuff that isn't Debian supplied Debians only contribution is the packaging. Debian considers "Public Domain" is a perfectly acceptable licence for the packaging work. So LLM produced packaging is also perfectly fine - copyrightable or not.

By the by, even the option most favourable to LLM's in the ballot insists the Debian Developer take responsibility for all his work, regardless of whether he or an LLM produced it.

> That alone starts to invalidate the GPL and other FLOSS licenses.

How?

> And companies (MS, Amazon, etm) will gleefully loot anything marked with "LLM" as a free-for-all.

I'm not fan of their sharp practices either, particularly when I wrote this: https://lwn.net/Articles/1046105/ But what Amazon could do with publically licensed packaging that they can't do now is a mystery to me - it's not terribly useful outside of the Debian ecosystem, and it's not like you are forced to distribute most of it. Many companies just ship Debian binary packages now, no Debian source available. GPL and friends only bite if you distribute it.

I do think LLM's could be a threat to open source licences, but this isn't the mechanism. The real threat is far more insidious: https://lwn.net/Articles/1064113/

So, for Debian packaging LLM's don't create copyright issues, or accoutability issues. If environmental concerns are serious, Debian could always setup it's own LLM server farm running open source models powered with renewable power. That sounds expensive, but whenever Debian needs compute (it already use a lot recompiling all those packages on all supported arches) it just seems to "appear".

I'm not sure what's left, beyond a technophobia of computation done using 4K vectors rather than bits. The underlying silicon is the same after all, lots of packing tools already automate most of the steps, and the output (the packaging) is highly constrained so will be near identical.

It is indeed puzzling. Back in the day, I built myself a music playing device using TTL gates, eagerly studied what CPU's looked like inside and in general wanted to know how everything worked - even though it had no hope of replicating most of it.

Looking back, it was an ideal start to a 40 year career as a software engineer. And of course currently I'm reading "Why Machines Learn: The Elegant Maths Behind Modern AI" because the urge to understand the machines I want to master and control hasn't gone away.

This is just Pournelle’s iron law isn't it? https://executivecoachinglondon.com/career-development/pourn...

You are equating n LLM's output with FORTRAN?

The only time you really need to look at core dumps of a FORTRAN compiler is when you think the compiler has a bug. Apart from that, the abstraction doesn't leak. You don't need to know any details of the CPU to debug what is going on, every compilation you do is deterministic and has a predictable outcome. When you make a change, you should be able to predict the exact outcome. Don't like it? Reverse the change, and you're exactly where you were before. I've coded on a few machines I've never bother to learn what the metal does. It is always like this, and it's fine.

LLMs are nothing like that. Like FORTRAN the LLMs make you feel like they've abstracted you from the code. But you can't get by exclusively feeding instructions to an LLM. The OP made was making that point when he said "When AI can’t solve it, they just keep trying and failing". When your only undo is to say "fix the mess you just created", you can't replay instructions you gave yesterday to give you the same result, and you can't even say "explain what you did" and expect to get a reliable answer, there is no out. You have to look at the code.

When the abstractions leak badly, you have to learn both the abstraction and the thing it's trying to simplify. They are little better than macros that save you typing. An LLM is a very leaky abstraction.

Canada copied Australia - we did it first in 2021. https://journals.sagepub.com/doi/10.1177/14614448241232296

This is just an extension to 2021. The pigs have come back. They want more gravy.

Fairy Wrens are mentioned: https://fatbirder.com/ornithology/maluridae-australasian-wre... Pity they don't show them in their natural habitat: sitting on a car mirror, shitting on it.

you must have stolen it off me.

> Utility scale batteries are now at ~$66m/GWh

That equates to AUD$66/kWh, after shipping, installation and connection to the grid. It seems very low. Have you got citations for that number?

In related news "Big US‑style trucks are killing more pedestrians in Australia":

https://theconversation.com/big-us-style-trucks-are-killing-...

You mentioned in your you were interested in seeing the 18A replacement for the N100 series.

My understanding is the Wildcat lake 304 is that replacement, and it's been "released" for a while. It can be configured down to 10W vs 6W for the N100. Both draw about 7W from the wall when idle. The 304 is about 80% faster for the same power draw, has a better GPU in includes a 15 TOPS NPU.

Despite the 304 being released, it's not easily available yet.

What part of "Amazon's margin is a competitors opportunity" aren't you groking?

I have no idea what a drone factory is, but my mental image is a building with drone parts on one side, people assembling with glue, screwdrivers and soldering irons in the middle, and finished drones in boxes at the far end. It that building is destroyed by a missile you've lost a few drones and a building. No people were hurt, so just move to a new building.

The images out of Russia are a little different. Billions of dollars of plant and equipment blown up, hundreds of ships disabled, entire Wildberry warehouses covering acres going up in smoke. It doesn't seem to be in the same ballpark as a few days outage while you move stock and employees to a new location.

Maybe I'm missing something.

> Why cut revenue to nearly half?

Because, to quote Jeff Bezos, your margin is my opportunity. And did he ever squeeze every drop from that opportunity. Just ask the brick and mortar stores in your street.

> I’m not convinced that basic economics works when output capacity isn’t fully scalable.

Economics works for land, which isn't scalable.

> Same reason no lawn care team tries to deeply undercut competitors

That works right up until Fred down the road loses his job, and then it dawns he could be making more money with his mower because of those sweet fat margins those other lawn care people thought they could keep all to themselves.

> Read that phrase back and then ask "is this the future of computing that we wanted?"

It is what I want.

If you own a mobile phone it's unavoidable. The physical SIM is a guarded area with secrets you can't know or touch - the secrets it holds prove you paid for access to their network.

If you want to carry a zillion others devices you can't access the innards of that prove something about you then good for you. But don't assume the rest of us think that's a good idea. We no more object to carving out a little bit of firewalled memory for the exclusive use of the bank than we object to the electricity meter in our houses. It's also a locked down piece of equipment we can't modify or touch that lives on our property. Most people are happy to grant that intrusion on their personal space in exchange for having the electrity connected.

It's no different on your phone. In return for your phone protecting its details from you, you no longer have to carry a credit card, or driver's licence, or prove you paid for ads to go away in some app, or access your works VPN. I honestly can't see much difference between carrying a credit card the hides some information from me, or putting the same info in the phone and it hiding the info. Except for having one less device to haul around, of course.

In Australia, the cause isn't a mystery. A judge ruled publishers are liable for defamatory comments. Within a few weeks, all the comment sections were gone: https://www.sydney.edu.au/law/news-and-events/news/2021/09/1...

That one ruling took a huge toll on social discourse and civil society in Australia. It was a very sad day. Section 230 was a inspired piece of law making.

Yes, Norway leads. But it's not an apples with apples comparison.

Norway is a smallish country (you can fit 24 Norway's in Australia - so less range anxiety), whose government pushed the transition to EV's hard. Australia has the worlds most competitive new car market and with one of the least restrictive emissions controls on ICE engines.

It's not that surprising that Norway's government engineered transition to EV's was successful. In Australia EV's became around the same price as ICE vehicles in the last 12 months, and it feels like we've hit a market driven tipping point.

If battery developments take their predicted path, with pricing dropping about 10% per years and range per unit weight doubling, it's difficult to see how it isn't all over bar the shouting - for USA car manufacturing.

Trump sealed the USA's fate, when his changes EV policies lead to $51B in direct write downs. I don't see how you can recover from that.

And this pairs nicely with Jane Street's observation that agentic coding changes the formal proof equation: https://news.ycombinator.com/item?id=49064854

Formal proofs of code are almost beyond the capabilities of the best human programmers (3.7 lines per day!), but LLMs can bash out code at an amazing pace. It's often crap, sadly, but the proof they are bashing out is the hard bit. If possible at all, the task is EXPTIME. Verifying the proof is only P, so when it's wrong you tell the LLM to do it again. A stable agentic loop is what makes it possible. The results in the article I linked to speak for themselves.

> invest in high risk shares because a big win is the only way to afford a nice home.

Here I was thinking The Squid Game was a work of fiction.

You are being far too kind. A primate's body shape may be optimal for life in the trees, but it sure isn't optimal for any of the industrial or household tasks we want a robot to do. There are millions of robots around the world doing useful things, from vacuuming houses to assembling iPhones and playing table tennis. None of them look remotely like a human.

If an article features a humanoid robot it's sure sign its a PR fluff piece. This one looks to be a recruitment ad. Tesla's Optimus is being used to pump some sort of investment scam. God knows China's humanoid marathon is about. Whatever it is, it isn't getting from A to B quickly and efficiently.

If your objection is to someone giving an LLM a few dot points asking it to turn them into an article, then that clearly isn't what happened here. The research idea seems novel, the author put a bit of work into exploring the idea, the conclusions are clearly the author's own based on what his exploration found. He then used an LLM to write up what he found because his native language isn't English.

I'm fine with that. Going by the comments here, most people are fine with that. Given the choice between LLM-generated text or not seeing these sorts of novel ideas from different parts of the world on HN, they take the former.

I don't think the HN guideline "Don't post generated text or AI-edited text" is going to age well. My guess is it's already broken more than it's observed. The next bit: "HN is for conversation between humans" will be ageless. The mistake they make is "AI-edited text" is often net positive when humans converse, particularly if they come from very different backgrounds (language or culture).

This isn't a poetry forum. Bad spelling, bad grammar, inappropriate metaphors are all very human - but they get in the way when humans are just trying to communicate rather than create written art. The rule "Don't post generated text or AI-edited text" looks to fall into the category "there is always a well-known solution to every human problem — neat, plausible, and wrong".

+1

This anti LLM crusade on HN is driving me nuts. It means I have to view with showdead on, because otherwise I miss clear, insightful and consequently interesting posts because someone to an dislike to am em dash. Dead posts are so faint I often have to copy and paste them into a comment box to make them readable.

Everyone hates LLM generated "ideas". But LLM phrasing of a humans original ideas is a different thing. Not ideal perhaps, if the alternative is not to hear the ideas and opinions of different cultures and countries, I'll take it every time. For fucks sake, learn to tell the difference between the two - don't just moronically down vote because of LLM tells.

I'm not the OP, but I think he's saying something much simpler than that. Typically with a recursive algorithm, the state you need to push is tiny. Usually a word or two. But often you need a lot of temporaries to calculate that state.

If you use recursion, both the absolutely necessary state and the temporaries are pushed onto the stack (along with the stack frame the language runtime wants). If you use iteration, the programmer just pushes the necessary state onto the stack, and reuses the temporaries in place.

The rest of his claims follow from those facts. Space consumption is less because you are saving less. Cache locality is better both because the temporaries aren't scattered across the stack, and because you are reusing the same locations over and over again.

The downside of iterative solutions is stacks are more efficient memory allocators than heap algorithms - especially if you are forced to grow the array your pushing things onto many times. But if you know that in advance so you can allocate the full amount up front, then at the limit when N -> ∞ iteration will always win over recursion.

> what "maintainable" is is probably some high dimensional space described by these signals; it'd probably require some human labeling to figure out where this space is

Maybe, but "maintainable" has a simple definition: minimising the effort to incrementally grow a software system as its size grows to infinity (while maintaining some defect ratio). Humans (called senior software engineers) figure this out over decades by working on a number of large software systems.

So far models haven't figured the same thing out. I can only guess why. All their learning comes from examining code on the internet. Somehow whatever the patterns are that make large software projects like the OSs maintainable by large groups of people working independently has escaped them. Maybe the bulk of code they look at is in the small so they miss it, maybe the patterns are just hard to discern in big systems.

My theory is there are anti-patterns which the larger projects somehow manage to keep to a minimum. It's hard to learn something that's not there. Instead, you learn these anti-patterns by doing them, and watching a system all fall apart over the course of years, and if you're good you manage to pin the blame on the right thing.

If you want to wrap it up neatly in a package, models have learnt how to program, but programming is not the same as design. Design requires very different skills to using a programming language. At a very high level, big systems are sets of interconnected modules. The things that matter are narrow APIs with minimal coupling, and where coupling is unavoidable making it explicit and easy to reason about. Concrete examples of the anti-patterns to be avoided are global variables, leaky abstractions and APIs that require complex sequences of interconnected calls.

Those things don't bite hard until you program in the large. The current crop of LLM's seem to be useless at all of them. If you've got a big context window and you are only writing 100s of kloc, a budget of kilowatts to understand the complexity, it doesn't matter. But humans don't have a budget like that, and so far everyone I've seen who reads large chunks of vibe coded software recoils in horror, gives up, and walks away.

Even with that budget, the models fail anyway once the size gets way beyond their context window - it's just that they fail much later than humans. Which I guess makes learning the lesson so much harder - perhaps near impossible as your "they need labeled examples" hints at. Oddly the solution would be to train with smaller context windows (a less powerful model in some ways), so the failures become apparent much earlier.

It looks like they have been going with your solution for now - giving engineers cheap tokens via flat-rate plans, and watching how they do things. I don't think that will work. Engineers who chew through the billions of tokens offered by these plans are vibe coding. The code they produce is so poor it can't cross the threshold from small to large. There is nothing to see there about programming in the large.

:( what a rant - it looks like an LLMs CoT as it thinks through things. Which is what I was doing, I guess.

> what should be done about open weight bioweapon and cyber-offense capabilities?

Like the others here I know almost nothing about bio weapons, but I think perhaps the fact that smallpox's genome sequence has publicly available in scientific databases like GenBank for 30 years is relevant. That horse bolted a long time ago.

> Closed-source: You can execute a program with the binary. You cannot generate a new binary, but you could try to reverse-engineer it or (painfully) modify its execution.

Time have changed. This should be:

Closed-source: You point an LLM at it, and get back source that's often easier to understand than the original.

> we have no access to the source materials and cannot reproduce the model independently.

Given the USA companies have been loudly claiming the Chinese models are distillations of their models, also claiming "no access to source materials" seems dubious. As it was dubious anyway with because the Chinese publish lots of papers on how their models are designed, I'm left feeling I'm looking at the south end of a north bound bull.

The takeaway quote from the article is:

> The Lean library isn't just tested and validated, it's proved correct. This allows us to let AIs loose optimizing the code, requiring that they update the proof whenever the implementation materially changes.

To me it's a "that's obvious in hindsight, so why didn't I think of it?" thing.

If you can write a zlib compressor in it, you can write anything in it - so this is a technique that could be applied all problems amenable to formal proofs. Formal proofs are almost never worth the cost when they only prove the absence of bugs. Adding another payback, "you get speed as well", alters that equation.

Definitely merits more than 2 points after 20 hours. It feels like some readers can't look past an LLM being involved.

It's kinda amusing copyright always seems to attract discussions about authors rights. It was never about author or their rights. It was and is a set of rules that allow money to be made by publishing books and songs. Authors have a role in that, but so do publishers.

Turns out most authors and publishers suck at their respective jobs. Most books produced by authors are things nobody wants to read. If a publisher's job is defined to be finding works the public likes, they suck at it too. Instead they just publish lots of stuff, mostly at a loss. They make their money from the occasional hit. But only way they can make money from it is if they have a monopoly over publishing it for a while - which is exactly what copyright gives them. Looked at in another way, this "publish lots of stuff and see what sticks" is the way our society discovers what new works are popular. And copyright funds it.

If you take the view that copyright is paying for the discovery and publishing of new works, then consumers paying monopoly prices for 70 years or more is a bad idea. By far the majority of works are commercially dead 1 year after being published. The publishers tend to make their money from the remaining 1%. They might last five years. Very, very few last 20. I don't see how copyright lasting beyond 20 years can be justified with anything other than: "because of the good work he did 20 years ago, he deserves to be paid for sitting on his arse for the rest of his life". In reality, he didn't sit on his arse - he donated to a few politicians - but the outcome is almost the same.

> Chinese models have government-enforced censorship, while American models have security and legal restrictions.

They are both government enforced. The USA administration is using early access and export control restrictions to do their enforcing.

As for the effects: while I'm told the Chinese government is sensitive to some internal political topics, I've never seen it myself. I guess that's because I don't write code dealing on Tiananmen square or similar topics, and I don't know anyone who does. In contrast the restrictions on the USA models are a right proper pain in the arse. Hugging Face being forced to swap Chinese models to defend against the OpenAI GPT attack is a good an illustration as any.

That said, these USA restrictions are a new thing that seemed to start with Mythos - which means around 6 months ago. The Chinese models are (pull number out of the air) 6 months behind, so maybe in 2027 the Chinese models will start behaving like the current US SOTA ones.