LinXitoW

No? The most optimistic predictions involved AGI around the corner, 6 months until no more developers for years now.

For reference, I use LLMs daily for coding. I do think they are useful.

I am speaking about corporations and sales tactics, because this VERY experiment was done by exactly such a corporation. How about you think about how "this whole thing works", and apply it to their post? What did they not write? How many worse experiments did they not post about to not jeopardize investments?

I don't find this impressive, because it doesn't do anything I'd want, anything I'd need, anything the world needs, and it doesn't do anything new compared to my personal experience. Which, just to reiterate, is that LLMs are useful, just not nowhere close to as world shattering/ending as the CEOs are selling it. Acknowledging that has nothing to do with being a luddite.

Which is why we ought to always bring up their BS every time people try to pretend it didn't happen.

The promises made are ABSOLUTELY relevant to how promising or not these experiments are.

It's more like "We were promised, over and over again, that the squirrel would be autonomous grand master level. We spent insane amounts of money, labour, and opportunity costs of human progress on this. Now, here's a very expensive squirrel, that still needs guidance from a human grandmaster, and most of it's moves are just replications of existing games. Oh, it also can't move the pieces by itself, so it depends on Piece Mover library."

Model improvement is very much slowing down, if we actually use fair metrics. Most improvements in the last year or so comes down to external improvements, like better tooling, or the highly sophisticated practice of throwing way more tokens at the same problem (reasoning and agents).

Don't get me wrong, LLMs are useful. They just aren't the kind of useful that Sam et al. sold investors. No AGI, no full human worker replacement, no massive reduction in cost for SOTA.

The main issue with improvements in the last year is that a lot of it is based not on the models strictly becoming better, but on tooling being better, and simply using a fuckton more tokens for the same task.

Remember that all these companies can only exist because of massive (over)investments in the hope of insane returns and AGI promises. While all these improvements (imho) prove the exact opposite: AGI is absolutely not coming, and the investments aren't going to generate these outsized returns. The will generate decent returns, and the tools are useful.

How does 20K to replicate code available in the thousands online (toy C compilers) prove anything? It requires a bunch of caveats about things that don't work, it requires a bunch of other tools to do stuff, and an experienced developer had to guide it pretty heavily to even get that lackluster result.

People absolutely care, but the issue is that no single company/app is really responsible. It's the tragedy of the commons, but for users RAM. No one electron app uses all the RAM, but just a couple are enough to make a common 16GB machine slow down massively.

What is real banana? How much processing is allowed for it to be still real? Considering the selective breeding of banana, is banana even still real?

Chemical is just a bad word choice. Artificial, or ultra processed get closer to the issue. They still are vague with a lot of grey area. If you cook at home, you're also highly processing your food. The fruit in winter is likely also artificial, in some sense: Grown against the will of god/nature with pesticides, in a tent, in a climate that doesn't naturally feature them, devoid of flavour because they were artificially bred for yield, color and size, etc.

I think it's actually a great example of very very important non-pedantry. The entire crux of their argument/issue is dependent on their definition of "chemicals". I would even go so far as to say it's just the nature fallacy in disguise.

With the nature fallacy, the definition (or more like the lack of) of what is natural is the entire crux of it. In both cases (natural and "non-chemical") it's the very non-defined-ness that reveals the problem with it: You cannot create a sensible definition.

For nature, what's the definition that puts "rape" and "artificial insulin" on the morally correct side?

For chemical, what's the definition that puts "fortification with iodine, flouride, or whatevers in flour" and "arsenic" on the right side?

What the poster meant wasn't that the LLM itself is an entity with a preference, but simply that because of the training, LLMs are better at doing stuff in a standard Linux environment. If you have to teach it a new environment it either needs to waste time and context every time to look up stuff, or you need a company to do RL to teach it that new stuff (unlikely).

It would probably help if the sandbox presented a linux-y looking API, and translated that to actual browser commands.

You’re right that a lot of livestock feed is crop residues/byproducts humans don’t eat—but that doesn’t make beef “necessary” or erase the land/opportunity-cost problem. Globally, ~36% of crop calories go to animal feed and only ~12% of those feed calories come back as animal-product calories (Cassidy et al.). Livestock still consume ~1/3 of global cereal production (Mottet et al.). And in full-system LCAs that include grazing + feed land, meat/dairy provide ~18% of calories and ~37% of protein but use ~83% of farmland; cutting them can reduce farmland >75% while still feeding the world (Poore & Nemecek / Oxford). Plus, even if pasture isn’t croppable, it can be restored—land used for animal foods has a big carbon opportunity cost (Hayek et al.). Nutritionally, major dietetic bodies say well-planned vegetarian/vegan diets can be nutritionally adequate, with attention to nutrients like B12.

Just in case you need some recommendations:

Party games: Scale well with more people, easy to explain

- Werewolf

- Werewords

- Codenames (favorite)

Beginner Games: Accept a decent amount, somewhat easy to explain

- Camel Up

- Flip 7

- Dungeon Fighter

- Ticket to Ride

Games that have nothing to do with your problem, but I just wanna mention:

- Everdell: Cute critters prepare for winter

- Root: Cute critters prepare for war

- Azul: Place fancy tiles that look and feel delicious

- Bohnanza: The best part of Catan without the bad parts

But your program HAS to have some invariants. If those are not held, simply reject all the data!

What the hell is really the alternative here? Do you just pretend your process can accept any kind of data, and just never do anything with it??

If you need an integer and you get a string, you just don't work. This has nothing to do with types. There's no solution here, it's just no thank you, error, panic, 500.

I don't think any serious dev has claimed 10x as a general statement. Obviously, no true scotsman and all that, so even my statement about makers of anecdotal statements is anecdotal.

Even as a slight fan, I'd never claim more than 10-20% all together. I could maybe see 5x for some specific typing heavy usages. Like adding a basic CRUD stuff for a basic entity into an already existing Spring app.

Productivity gains in programming have always been incredibly hard to prove, esp. on an individual level. We've had these discussions a million times long before AI. Every time a manager tries to reward some kind of metric for "good" code, it turns out that it doesn't work that way. Every time Rust is mentioned, every C fan finds a million reasons why the improvement doesn't actually have anything to do with using Rust.

AI/LLM discussions are the exact same. How would a person ever measure their own performance? The moment you implement the same feature twice, you're already reusing learnings from the first run.

So, the only thing left is anecdotal evidence. It makes sense that on both sides people might be a little peeved or incredulous about the others claims. It doesn't help that both sides (though mostly AI fans) have very rabid supporters that will just make up shit (like AGI, or the water usage).

Imho, the biggest part missing from these anecdotes is exactly what you're using, what you're doing, and what baseline you're comparing it to. For example, using Claude Code in a typical, modern, decently well architected Spring app to add a bunch of straight forward CRUD operations for a new entity works absolutely flawlessly, compared to a junior or even medior(medium?) dev.

Copy pasting code into an online chat for a novel problem, in an untyped, rare language, with only basic instructions and no way for the chat to run it, will basically never work.

Even better. Job security for current seniors.

But what if it isn't just a basic website? Most sites I've worked on required things like content management, or auditing stuff, a bunch database stuff, SAML single sign on etc.

Most languages end up being better at some parts of the stack, like Java for overcomplicated enterprise BS backends. It seems bad to "fight" that trend.

I mean, they're taking away parts of cars at the moment. You gotta pay monthly to unlock features your car already has.

We already had that happen. When GPT 5 was released, it was much less sycophantic. All the sad people with AI girl/boyfriends threw a giant fit because OpenAI "murdered" the "soul" of their "partner". That's why 4o is still available as a legacy model.

You mean like Epstein? We've got a bunch of truths about rich people and nothing happens.

The fear of an evil government misusing something, more often than not, is a thought terminating cliche. It means we cannot regulate, or create any laws about anything, because evil people could abuse those laws. In reality, evil people do evil shit, irrespective of the laws available for abuse.

Again, this has nothing to do with the point at hand, which is that "in any language, a developer can choose the crash the problem if a unrecoverable state happens". That's it.

Tell me about how these supposed magical groups have anything at all to do with language features. What language can magically conjure triple the memory from thin air because the upstream query returned 200+ entries instead of the 60-ish you're required to support?

Because Steam gives customers useful features that are good? GoG should also directly support Linux.

Most people do have an issue with it, because every game that's replaced loot boxes with discrete cosmetics purchases has to then massively increase the price of them. For 20 bucks in Overwatch 1, you got (afaik) 10 loot boxes which all had 4 random items. In Overwatch 2 20 bucks barely gets you single good skin.

It's very much a grass is greener type of situation in my experience, having been part of communities of both types of games.

The fetish for "manners" has stood in the way of every single positive societal change. It's exactly what MLK meant with the white moderate favoring a negative peace over positive change:

the Negro's great stumbling block in his stride toward freedom is not the White Citizen's Counciler or the Ku Klux Klanner, but the white moderate, who is more devoted to "order" than to justice; who prefers a negative peace which is the absence of tension to a positive peace which is the presence of justice; who constantly says: "I agree with you in the goal you seek, but I cannot agree with your methods of direct action"

No, it's meant for interactions inside the company, not towards random giant corporations outside of it.

Now, that's all just regulations. What are regulations but laws that restrict/govern the way to do commerce? Anti-slavery is part of that, just like every other concession we've had to pry from the hands of capitalists over the last 100 years, like no child labor, no locking workers into factories, PPE, etc...

But that's already true for most cases and devices. Most people using most devices let auto updates just happen.

And the other option isn't that much better, because "don't do autoupdates because maybe the update server is compromised" leads to a bunch of unsecured devices everywhere.

The only "real" solution is also completely unrealistic: Every private person disables auto updates, then reads the change log, downloads updates manually, and checks them against some checksum.

The better solution would be to simply increase fines until morale improves.

Katastrophenverursacherverlagerungserleichterung

Well, no. It's the right, for example, that constantly saw the "spectre of pedophiles" everywhere, including a random pizzerias basement, but when it comes to Epstein Files and his friends, many of who are in office, they suddenly don't care. Leftists, are, at least as far as i can tell, very consistent in not liking child molesters.

There's a huge amount of rightists against ALL abortion, until they suddenly need one. I don't know of any leftists that are ever like "I think abortions should be legal except for that one person who I don't like".