jdw64

nice game

In his book Programming as Theory Building, Peter Naur viewed programming not as the act of generating source code, but as the act of building a theory about the problem and its solution.

He said the lifecycle of a program ends when the development team is disbanded or when knowledge transfer fails to happen. But these days, people develop with AI. Will that knowledge actually remain? There's a slightly tircky point here

Of course, even in the past, once a codebase got big enough, everyone was writing code on top of an incomplete theory. But when you delegate work to AI, the theory itself isn't yours anymore, which makes it even trickier.

The problem with the game industry is, honestly, that everyone wants to break into it.

The game industry is genuinely terrible, starting with low pay and brutal working hours, but it's interesting and lets you do a lot of fun things.

Because everyone wants to do it, there are plenty of people willing to put up with somewhat lower pay.

I think a bit differently. The real problem with vibe coding is that it's hard to write a precise specification of the problem. But if the specification is accurate, running the AI dozens of times tends to make it much faster than a human (purely from experience). However, for most problems where vibe coding is actually useful, a human still has to set the initial framing.

Will it be faster than the Go version? But why is this even necessary? I don't get it.

This idea is so good, though

Will it even be maintainable? Never mind the rocket fuel...

Will the frame itself hold up?

The advantage of AI coding is that it's non-deterministic. In fact, as you code, your style tends to harden, but when you repeatedly run AI code toward the same goal, you come to think that the non-determinism is actually better.

But can they even tell the difference? Honestly, setting aside whether AI was used or not, even completely human-written text comes out as MIX, and text that's 100% AI comes out as human-written...[1]

[1]https://www.makonea.com/en-US/casual/pangram-doesn-t-seem-ri...

Wow, I thought it was a GIF, but it's actually interactive. The website is just too heavy, though.

But is Pelican Bench really that helpful? I'm not sure. Why is this even trending? It seems like a kind of meta joke — is there anyone who can explain in what sense it became a kind of community meme? I understand it became famous because a well-known person named Simon tried it. But I feel like there should be more coding benchmarks, honestly...

I've been learning by using REA to decompile and reverse-engineer older games and software.

It's surprisingly powerful.

Why can't the Gemini model code? I'd guess Gemini probably used the best dataset. Even Google is the place where the world's best coders are gathered, isn't it? Yet the GPT code from OpenAI, which collaborated with Microsoft, is beautiful, while the Gemini model's coding is trash.

Then what solutions has academia proposed? From what you're saying, it's been over 30 years. In my country there's a saying: "A monk cannot shave his own head." Has it been solved? Or has this proverb, as expected, been proven true? I'm curious.

I think this is a problem that existed even before. If you look closely, AI just accelerates everything—both the pros and the cons. It seems to be an inherent problem with the academic system itself.

Signals are inherently bound to be distorted. The "legitimate methods" conceived by humans and social signals are easy to manipulate. Because public data like the number of published papers and diplomas act as strong social signals, they end up being distorted. This is known as Goodhart's Law.

The problem is that if this persists long-term, there will eventually be a shortage of personnel to quantitatively evaluate papers, and society probably won't hire these researchers. There is also no incentive to hire them, as verification rarely leads to direct profit. People tend to want to be the starting point of a discipline; they generally do not like checking other people's work.

As a result, academia will increasingly be validated solely by "which lab, which advising professor, and what peer network certified you." The value of the offline world will rise, simply because digital costs have become too cheap.

Soon, research quality and social status will likely merge again, resembling the Middle Ages. Even if we assume someone has actual research capabilities, if evaluators do not invest enough time in verification, the system will inevitably revolve around the more prestigious labs.

I suspect there will be a flood of research applying AI, but the topics of these studies will likely be quite narrow and localized (it would be nice to have statistics on this). I believe the overall diversity of topics explored across academia will diminish.

For this reason, I think existing networks will only grow stronger.

I feel that any solutions to this problem will go awry. I doubt self-purification is even possible anymore. Established researchers protect the monopoly power of publishers, while those existing publishers generally just want to cut costs. The authority of these publishers must be protected by the established researchers who have already secured their positions.

Next, the system that used to manipulate the number of papers will probably shift into a system that manipulates network reputation. The competition to get into good labs and the strategies to secure co-authorships with famous researchers will simply become the new means of competition.

The explanation is easy and fun.

I think the AI bubble won't burst. It feels like it's going to become the new normal. If you think about the accuracy rate of Bloomberg's predictions, can we really say the market always over-discounts future earnings excessively? The next thing I'm curious about is whether AI demand will really decrease...? I have that kind of question.

In East Asian countries, AI chat apps are generating massive profits, so why did the pioneer, Character.AI, go bankrupt? On top of that, didn't they receive a $2.7 billion buyout? AI chat services are actually still bringing in huge revenues today, which makes it really hard to understand how they ended up filing for bankruptcy

It would be nice if Korean also worked. I’d like to experiment with a Korean model too.

I like the title “puzzle solver,” but I think “slop assembler” is fine too!

I think Mr. Linus, since he has often implemented things with human agents, will also handle AI agents well.

Nice to meet you, fellow workers!

It's true that China is remarkable, but the concern is that China is too large. Basically, these cutting-edge industries require industrial chains to be connected across different countries, but China can do almost everything within its own borders. China's greatness is greatness, but if China becomes highly developed, will this kind of technological development spread to neighboring countries?

U.S. industry, for one, spreads to many countries through outsourcing. But China can consume outsourcing within its own borders. If that happens, what will become of it?

New and good software exists only in the minds of software developers. Every piece of software that actually exists is a target of complaint. I hate Excel and curse it, but in the end, I use it — that's just how it is.

That's not wrong, but many ordinary people don't know that. And even if you use AGENTS.md, there are in fact times when it gets violated.

Statistically, you're right. But still, party-game-focused games have a chance of succeeding. In other words, it feels like the trend cycle has gotten faster than the "super jackpot" works. Rather than impossible, you could say it's become harder... and in fact, games themselves are statistically few... but even so, everyone wants a miracle.

Everything is luck and timing. Look at a game like Mecha Chameleon. That game is a vibe coding game, but it's very fun, isn't it? What about Vampire Survivors? I think a lot of every success and failure depends on luck. But separately from that, I want to become the developer of some great project. Right now I mainly deliver programs that go into factories, but someday I have a small dream of making a project that everyone can recall when they hear my name. In that sense, I envy ConcernedApe. With my talent and ability, it's probably hard, but life is long.

So in the US, they call them orange cats. In Korea, we call them 'cheese tabbies' or 'cheese nyang(kitten).' Someday, when I leave this place, I'd like to get a cat. I really like the ones with broad faces. But right now, I can barely even take care of myself

I want to have a project like that someday too.

I'm a non-native english, so I change the code myself, but I tend to participate in communities while using a translator. If I use AI, I can fix bigger things, but most of what I contribute to open source is finding small bugs, so I don't use AI for open source contributions. (I mostly don't even have the ability to make big changes.) Looking around open source like that a few times, I can understand the hatred toward AI to some degree: too many requests pour onto a single individual. Of course, that doesn't mean AI isn't helpful. In my view, AI writes code blocks better than I do, and it might even be better to use AI for some of the open source I've contributed to... but overall, I feel that some kind of system for filtering AI-based PRs is needed.