Random Thread

[Raphael]

I feel like there's a lot of wishful thinking going on around the AI bubble, especially on the left and the center.

Fair enough. As I wrote before, I have a similar impression when it comes to crypto.

There are also people who don't quite understand the predicament the market puts us in. E.g. Travis B. is a market socialist.

I don't have the impression that Travis sees the market as a moral force, though.

[rotting bones]

I don't have the impression that Travis sees the market as a moral force, though.

If not exactly a moral force, more like a force of reasonableness which is better than the alternatives.

[Travis B.]

I don't have the impression that Travis sees the market as a moral force, though.

If not exactly a moral force, more like a force of reasonableness which is better than the alternatives.

My market socialism derives from that I have severe doubts about the practicality of central planning, as it has shown itself when attempted at any scale in the past. There is a reason why consumer products were very overpriced and of low quality and availability in the Eastern Bloc.

Note, though, that for creative works with minimal to zero marginal cost (e.g. recorded music, software) I believe in a system much closer to what rotting bones advocates, where everyone would get a vote that they can split any way they like which determines how money allocated to such creative works is to be distributed to creators in return for said creators relinquishing intellectual property rights to such works to the public.

Also note that I am also open to the idea of systems based on labor vouchers, even though they have not been tried at scale.

[malloc]

I feel like there's a lot of wishful thinking going on around the AI bubble, especially on the left and the center.

Quite. We need active resistance, not hoping that AI will collapse on its own.

[rotting bones]

Travis B.: Explained the differences a bajillion times.

I think the deeper problem is that people believe that "central planning" is a thing, and both Soviet planning and my proposal fall under that category. This is thinking in terms of Platonic essences whereas in reality, there are only systems of relations. I have tried to explain this in a few sentences, whereas I would clearly have to teach an economics course.

We're doomed.

[zompist]

If spenders demand language models, society as it is currently structured is going to churn things into language models until demand plateaus.

This is not to say there's no AI bubble. The market just crashes sometimes when people lose confidence, or for technical reasons. If the demand remains, the industries will be built back up again, like what happened with the dot-com bubble.

How are your Bored Ape investments holding up?

The whole nature of a bubble is that investors think the price will keep rising forever, and it doesn't. (I'm carefully avoiding saying that anything has inherent value.) In the middle of a bubble, or a Ponzi scheme, skeptics seem ridiculous. Prices are going up! People are making billions! Smart people!

And then it crashes, and the smart people look dumb. Well, the people, the majority, who didn't get out in time.

By promoting the idea that AI is inevitable and unstoppable, you are simply helping to inflate the bubble. It's not an argument that AI is worth even the current level of investment. People have written long essays with actual data to explain why it's an unsustainable bubble. Some of those essays are now coming from the investment media.

Some forms of AI will remain, but you can't just assume that "the demand remains." Demand is not a fact of nature. Demand is high because AI is spectacularly underpriced, investors want their money back, and CEOs are terrified of missing out and/or intoxicated by the idea that the jobs they don't understand are easy to automate. Plus, the sad fact is that the US economy doesn't have much else going for it right now, especially since the GOP wants to destroy the actual things the US economy is good at or could be.

[rotting bones]

How are your Bored Ape investments holding up?

I can't believe anyone bought NFTs.

The whole nature of a bubble is that investors think the price will keep rising forever, and it doesn't. (I'm carefully avoiding saying that anything has inherent value.) In the middle of a bubble, or a Ponzi scheme, skeptics seem ridiculous. Prices are going up! People are making billions! Smart people!

The difference between Bored Apes and ChatGPT is that Bored Apes does nothing. Investors thought they would appreciate in value. No one buys ChatGPTs thinking they would be worth millions in the future. If people are paying for it, that's because they are using it. If no one thought Bored Apes would rise in price, no one would buy them. Even if no one thinks OpenAI will rise in stock valuation, consumers would still want to pay for ChatGPT to solve their problems. And ChatGPT solutions are turning up in various products.

There are many different ways this could be a bubble. It could just be a downturn in the business cycle. What's more likely is that the industry needs to be restructured to make it profitable. (These days, leftists call this process "enshittification".) There is no guarantee that after this happens, the industry will slow down in building data centers. The service will just be worse and/or more expensive.

[zompist]

The whole nature of a bubble is that investors think the price will keep rising forever, and it doesn't. (I'm carefully avoiding saying that anything has inherent value.) In the middle of a bubble, or a Ponzi scheme, skeptics seem ridiculous. Prices are going up! People are making billions! Smart people!

The difference between Bored Apes and ChatGPT is that Bored Apes does nothing. Investors thought they would appreciate in value. No one buys ChatGPTs thinking they would be worth millions in the future. If people are paying for it, that's because they are using it. If no one thought Bored Apes would rise in price, no one would buy them. Even if no one thinks OpenAI will rise in stock valuation, consumers would still want to pay for ChatGPT to solve their problems. And ChatGPT solutions are turning up in various products.

Sure, some people pay ChatGPT. As Ed Zitron points out, more people pay Spotify. However, Spotify is profitable (barely), and ChatGPT is not.

What makes a bubble a bubble is not that some product is bought. Tulips and railroads have their uses. It's that investors go crazy. It's the investors who are betting that ChatGPT will increase in value forever amen.

If ChatGPT could make money by doubling its prices, it would. Every CEO in America knows what AI is, and if they can, they are demanding that their underlings use AI-- and that's not enough. There's not ten times the amount of demand that's just hiding somewhere. LLMs just don't seem to be the huge business its proponents want it to be.

[Raphael]

No NaNoWriMo for me, but I'll try to get somewhere with a kind of NaNoFiWriMo of my own. Wish me luck!

[rotting bones]

No NaNoWriMo for me, but I'll try to get somewhere with a kind of NaNoFiWriMo of my own. Wish me luck!

Good luck.

[rotting bones]

Sure, some people pay ChatGPT. As Ed Zitron points out, more people pay Spotify. However, Spotify is profitable (barely), and ChatGPT is not.

What makes a bubble a bubble is not that some product is bought. Tulips and railroads have their uses. It's that investors go crazy. It's the investors who are betting that ChatGPT will increase in value forever amen.

Consumers didn't pay for Google initially. Now Google sucks because of their sponsorship structure. This is the Silicon Valley strategy: https://youtu.be/p7Lo0sZfdHE Once not enough junior coders have been trained, ChatGPT will be poised to corner the market.

If ChatGPT could make money by doubling its prices, it would. Every CEO in America knows what AI is, and if they can, they are demanding that their underlings use AI-- and that's not enough. There's not ten times the amount of demand that's just hiding somewhere.

The demand keeps rising. Using AI is now standard practice in industry.

LLMs just don't seem to be the huge business its proponents want it to be.

You are assuming CEOs (or consumers) are smart. A common mistake. Like I said before, I can run a coding model that's better than ChatGPT on my PC (it uses the Apache license), the technology to download weights and run them locally already exists, etc.

There could be many reasons why the dinosaur model continues to be used: they haven't cornered the market enough to pivot from the crowd pleasing infrastructure, they haven't fully figured out how to deploy it across all the hardware, the non-technical tech bros are holding out for AGI, or maybe the people in charge don't understand their own business model yet. Happens more often than you'd think.

In fact, the stock market in general is insane. Tesla's high valuation continues even as sales are down. The market has not made sensible allocations once in my lifetime. I honestly think market glazers are all high, possibly on their brain's natural psychedelics.

[zompist]

Consumers didn't pay for Google initially. Now Google sucks because of their sponsorship structure. This is the Silicon Valley strategy: https://youtu.be/p7Lo0sZfdHE Once not enough junior coders have been trained, ChatGPT will be poised to corner the market.

Sure, that's the hope. The techbros also hope to move to Mars and/or live forever.

I expect there's too much competition for OpenAI to get a monopoly. Plus, one word that should strike terror into their hearts: DeepSeek. OpenAI's whole business model is based on their product being extremely expensive to create and run. What if they're off by a few orders of magnitude?

I feel sorry for people in my former profession. I got sick of the management and programmer fads a couple cycles ago.

LLMs just don't seem to be the huge business its proponents want it to be.

You are assuming CEOs (or consumers) are smart. A common mistake.

A baffling statement. CEOs love AI because they're dumb herd-followers, not because they're smart. They are the one job category that could be improved by replacing them with LLMs.

Like I said before, I can run a coding model that's better than ChatGPT on my PC (it uses the Apache license), the technology to download weights and run them locally already exists, etc.

Didn't you just say OpenAI was going to corner the market?

In fact, the stock market in general is insane. Tesla's high valuation continues even as sales are down. The market has not made sensible allocations once in my lifetime. I honestly think market glazers are all high, possibly on their brain's natural psychedelics.

The market is often crazy. But OpenAI is not publicly traded.

[alice]

I feel sorry for people in my former profession. I got sick of the management and programmer fads a couple cycles ago.

Admittedly my situation is unusual, but 95% of the software development jobs I find on the websites seem to be from the same three or four companies, looking for people to train LLMs in <programming language>, fully remote and based in <reasonably large town>, at <sum of money> per hour.

(You'll understand that my contact with the software development industry has been somewhat patchy over most of the past decade; could you give some examples of these "fads", just in case I can feel smug about having recognised them as such myself?)

[WeepingElf]

My career in programming was over before it really began. In my youth, I dreamt of founding my own game studio; my head was full of ideas for epic adventure and role-playing games, and that was the reason why I enrolled in computer science. Along the way, I abandoned the game studio idea, and when I was close to finishing university and started looking for jobs, I found that most jobs either required professional experience or skills not taught in university (or just not taken by me); also, I hated Windows (since then, tempered to a mere dislike and the conviction that Linux is better), which ruled out about 90% of all jobs offered. Also, two of my former fellow students died: one suffered a stroke at 27, one committed suicide because he found no way out of the rat race. Neither was a close friend of mine (I didn't find many friends among my fellow students because most were either intolerable computer nerds or equally intolerable career nerds), but they were just the tip of an iceberg, as many people I knew spoke of a very stressful, fast-moving business, and quite a few quit and sought a career in a different field. Hence, I never really worked in that business, apart from an unsuccessful internship as a web designer. Together with my autism spectrum condition, this failure threw me into a deep psychological crisis from which I took many years to recover, and that not fully, and I am still legally unfit for work (which means that I get public benefits and don't have to work; I am now trying my luck as a free-lance writer and musician, but with little success so far).

[zompist]

I feel sorry for people in my former profession. I got sick of the management and programmer fads a couple cycles ago.

(You'll understand that my contact with the software development industry has been somewhat patchy over most of the past decade; could you give some examples of these "fads", just in case I can feel smug about having recognised them as such myself?)

Oh, let's see... object-oriented programming, XML, Agile, blockchain, AI. Mini-fads like: we used to use C now we use C++ now we use Java now we use Java Beans now we use C#. I don't even know the latest programming languages much less the frameworks.

You'd probably enjoy this page.

Please note, I'm not saying all these fads were bad. You can get very excited about the latest fad when you're 25 or 35; you'll shrug when you're 55. But fads tend to be oversold, and applied by managers who don't understand them.

Also note, I'm an architect at heart. If your codebase is such that it will be improved by the latest fad, it is probably poorly architected. Programmers are always interested in tools that turn design into scutwork, and then tools to automate the scutwork. An architect looks at ways to eliminate the scutwork.

[zompist]

My career in programming was over before it really began. In my youth, I dreamt of founding my own game studio; my head was full of ideas for epic adventure and role-playing games, and that was the reason why I enrolled in computer science. Along the way, I abandoned the game studio idea, and when I was close to finishing university and started looking for jobs, I found that most jobs either required professional experience or skills not taught in university (or just not taken by me); also, I hated Windows (since then, tempered to a mere dislike and the conviction that Linux is better), which ruled out about 90% of all jobs offered.

I'm kind of surprised it was so different in Germany. I didn't feel lucky with my career-- I graduated in a recession, got burned by several mergers, and didn't luck out with a startup-to-sale payout. I have a CS degree, but many of my colleagues didn't; it felt you could get into the field from anywhere. But it was generally well paid and, in the early years, fun.

FWIW my career started before Windows-- my first real job was on the Apple II. Apple ][ if you like. But then it was DOS and Windows for decades. On the other hand, an older guy I knew was into mainframes and desperately wanted to get into PCs, but felt he couldn't take the pay cut that would entail.

[Travis B.]

I started out programming at eight years old on Apple //e boxen at school and soon after at home (I begged my parents into buying a used Apple //e), at the very tail end of the microcomputer era. Soon my family replaced the Apple //e with a used Mac IIsi, and then that with a PowerPC Mac. Lucky for me, the previous owner of the Mac IIsi was a programmer, who gave my family all his programming stuff. From this I learned C and C++. Soon I learned of Linux from seeing books on it at the local Barnes and Noble, and after I got a job so I could save up to buy a new hard drive (my parents would not allow me to wipe the hard drive that came with the machine) I installed Linux and didn't look back. At college I was a member of the Undergraduate Projects Lab, and we exclusively used Linux (while the other machines in the CS department were Solaris boxen).

I've been a Linux programmer since then, and even when I programmed on Windows for work I've always programmed on Linux at home. In the past five years, though, outside of work (where I program for Linux) I primarily program embedded systems running ARM Cortex-M microcontrollers ─ but my daily driver is still a Linux box.

The only fads I really got into were functional programming, which I did for more than a decade outside of work, and software transactional memory (STM), which is a neat idea but which is only really practical in purely functional languages like Haskell. However, in more recent times I have been programming outside of work mostly in Forth of all things, which in many ways is an anti-fad (I find most Forth programmers out there today are my parents' age).

Now, why Forth? Basically, I had always dreamt of creating my own operating system and my own compiler, and Forth on microcontrollers is the ideal way to fulfill this dream ─ and furthermore, it turns out that for embedded systems Forth is practically the ideal language, providing a combination of interactivity, freedom from the perennial edit-compile-flash-test loop, and low level access to hardware otherwise not provided by really any alternative.

Also, more open microcontrollers like the RP2040/RP2350 family allow control over the entire system that is not really feasible on modern PC's or even devices like Raspberry Pi (not Pico) single-board computers, which makes writing completely on the bare metal, and thus creating one's own OS, actually practical.

[Raphael]

No NaNoWriMo for me, but I'll try to get somewhere with a kind of NaNoFiWriMo of my own. Wish me luck!

Good luck.

Thank you!

[Raphael]

The Onion, 2014:

https://theonion.com/teens-migrating-fr ... 819595619/

Real life, 2025:

https://www.techdirt.com/2025/10/31/kid ... e-they-do/

[rotting bones]

A baffling statement. CEOs love AI because they're dumb herd-followers, not because they're smart. They are the one job category that could be improved by replacing them with LLMs.

I'm referring to what you say here:

If ChatGPT could make money by doubling its prices, it would.

This seems to assume that if CEOs (or consumers for that matter) saw an obvious path to efficiency, they'd take it. But that's not how business strategy works outside economics textbooks. "Basic competence" might be a better term than "smart".

Didn't you just say OpenAI was going to corner the market?

As you say:

Sure, that's the hope.

Will it work? It's a dumb idea, but it might still work depending on the marginal incompetencies of various groups.

(Maybe after the Trump administration degrades food quality enough, we'll have no choice but to use AI just to keep society running.)

The market is often crazy. But OpenAI is not publicly traded.

AI is a bubble insofar as the market is inflating its value through circular investments among industries. As you say:

It's the investors who are betting that ChatGPT will increase in value forever amen.

Without the possibility of market domination, the investors would see no reason to prop up ChatGPT.

I feel sorry for people in my former profession. I got sick of the management and programmer fads a couple cycles ago.

I'm not sure they're the people you want to feel sorry for. I know people without degrees trying to start small businesses. It's impossible in this environment. Small producers should be subsidized by the government based on popular demand.