The "AI bubble" narrative has become popular and entered the mainstream. Wired^[1] has a very good article. The importance of a good narrative as a defining feature of financial bubbles got me thinking about other narratives, and about the contradictions of capitalism, and the way they've led us to the collective delusion of the hustle economy.

Crypto

Right back to Bitcoin. That had a fantastic narrative. A decentralised, and thus incorruptible store of value which would challenge, or even usurp, legacy banking and fiat currency. Unfortunately the technology and user experience remain far from able to deliver. It's much too complex for the average person to use, and the mainnet still handles only five to seven transactions per second (VISA et al. can do thousands per second). You might appeal to Lightning, but that's another layer of complication, nobody is really using it, and it still has significant scaling limitations. Hence the popularity of holding funds on exchanges, which really was not meant to be the point, and is mostly about naked speculation (and has seen plenty of people get hustled).

The revolutionary narrative is pretty dead now. So far as I can tell, people tend to promote BTC as a store of value - "digital gold". But it's become more of a store of shared delusion than ever. Unregulated and commonly manipulated, not much of an escape from legacy banking. Can hardly spend it, so not much of an escape from fiat either. Any genuine true believers who remain form a shrill minority. For idiot whales who don't have to worry about paying their bills it remains an ideological playground, but for the majority, everyday bagholders, it's an evangelical cult with faith that more, greater fools will arrive some day and pump the value of their holdings. The fiat value^[2] because, although 1 BTC will always be worth 1 BTC, there's a growing awareness of how much bullshit that is.

NFTs are worth a brief mention as they also had a strong narrative. Independent artists could market themselves, digital assets could be securely owned etc. Of course, this was another obvious hype scam and nobody gives a fuck about NFTs anymore. Congratulations to all the Bored Ape "owners" out there.

After the disillusionment and end of BTC's time in popular consciousness (it feels a bit odd that, yes, there was a time when it was discussed - at length - on TV news), the pretence of this being about anything other than making (fiat) money for most of those involved became increasingly unbelievable. From early crypto and NFTs, through thousands of memecoins and scamcoins, to $TRUMP, a barefaced pump and dump from the president of the United States. The rot has set in this deeply.

Which brings us to neoliberal/late-stage capitalism and its contradictions. Even the pretence of believing in capital as a means of incentivising the production of things which will improve people's lives is becoming increasingly shaky. In this late stage, many have dropped it entirely in favour of nakedly zero-sum competition and an unapologetic, or even malicious, disregard for its social costs. Going further than simply not being of benefit to society, this is strongly malignant.

Prediction markets

Consider prediction markets. For everyone except their owners, for whom they're highly lucrative, they're a zero-sum (okay, fees make them negative-sum) proposition. Sportsbooks, as predatory as they are, feel almost innocent by comparison. There's a litany of examples of the perverse incentives these markets create, but I'll confine myself to a few. First, insider trading on the advent (and surely subsequent happenings) of the US/Israeli war against Iran^[3].

Second, causality running the other way - prediction markets incentivising the manipulation of reality. This is perhaps occurring in the Iran war too^[4], but there are proven examples, like that of Emanuel Fabian. Polymarket bettors lost out on positions that no Iranian missiles would strike Israel on March 10, due to his report that a missile "struck an open area" rather than being intercepted. So they started making threats against his life. Fabian stood his ground and, thankfully, everyone else involved is apparently a coward.

Consider the incentives created by other positions which have been available on Polymarket. You could have bet on how many acres of California would burn in wildfires^[5]. When prominent CEOs would depart their positions (due to, say, being killed). When products would be launched, creating incentive for corporate sabotage.

And clear insider advantage runs through all of it. This is just another means for regular people to try and hustle each other, and for capital to most definitely hustle them. Like crypto, retail trading, NFTs, prediction markets are obviously extractive. Late-stage capital is out of ideas, so the mask becomes ever-flimsier as its rapacious quest for greater, more desperate fools continues.

I don't think even the most ideological libertarian could claim straight-faced that anything of value is being created here. There is no worth at all to any of this. It's just a particularly ugly manifestation of zero-sum competition and extraction, with the bonus of perverse, reality-weirding incentives.

Attempts are being made to legislate against these sorts of positions, which is good. But the fact that they ever existed is dialectically instructive. We're in the hustle economy now. Extraction will only intensify. Increasingly immiserated, alienated and avaricious workers will continue to be pitted against each other in risky, zero-sum competition, while capital securely takes its cut.

None of this analysis is original, of course. In explaining the element of desperation, Demetri Kofinas coined the term financial nihilism back in 2021^[6]. If you can see that the economy is rigged against you - that home ownership is likely to always be out of reach for you, that pursuing higher education is unlikely to make financial sense - then YOLOing all your savings on $TSLA, or spending on luxury goods and experiences, starts to make sense.

Petit AI psychosis

Regular AI psychosis is all narrative, but it's pretty old news now^[7]. Though it is part of another, more personal layer to the narratives around AI.

AI is exacerbating the Dunning-Kruger effect^[8]. A notable and amusing case of this is the CEO of Krafton, publisher of Subnautica 2, using it for legal advice^[9] in a failed attempt to try and welch on a deal made with the developers.

Another interesting strand of this is how it affects corporate decision-making. Here, things become muddled with the AI booster narrative. A lot of CEOs are, as much as it pains me to admit it, intelligent people. But they have specific and limited domain knowledge, and probably tend to be selected for self-confidence. Add AI, something the workings of are very much outside their field of expertise, to this, and... yikes. Forrester Research found^[10] that 55% of employers who made workers redundant in favour of AI regret it. Of course, they also predict most of these employees will be replaced with cheaper, offshore workers, presumably as admitting the mistake (like Klarna was forced to) would be too embarrassing. Both personally and to markets. Going against the AI narrative would, presumably, make a bad situation even worse.

But even for CEOs with domain knowledge, the effect seems to persist. Garry Tan is the CEO of Y Combinator and holder of a Stanford degree in Computer Systems Engineering. But he also seems pretty cracked.

There is absolutely no way this isn't total slop. His GitHub doesn't have many new projects on it. Show us the code, Garry!

We can at least look at his website, garryslist.org, people have^[11], and unsurprisingly it's not great. Shipping test files, unused JS controllers (including the Rails "Hello World" scaffold), uncompressed images, an unused rich text editor. Rendering the entire page content twice in the DOM for mobile compatibility. While this should be embarrassing, it's true that it's not a big deal for a personal site. But the narrative pushes the idea that it's okay to bring this approach to things which are actually important, and these people don't seem to question it. Real backends, for real, consequential applications.

Enshittification enablement

This is the grindset/productivity narrative playing out in software development. We have the tools for it now - you can totally obtain that Duolingo 200-day shipping streak, push out as many lines of code as you want, and claim that you're a 10x developer. But if you think you're doing good engineering, you will be deluding yourself.

And this explosion of shitty code ties in with the AI booster narrative. Things are going wrong^[12] and will continue to. Up to 30% of Microsoft's code is now written by AI^[13] and, ironically by ramming AI features into it, they have managed to introduce a remote code execution vulnerability into, uh... Notepad^[14]. For years, I've wanted Windows' basic text editor to have AI integration. All in service of the narrative, I guess.

For people like Garry and Satya Nadella, part of the appeal is presumably the advancement of capital's ongoing quest to eliminate labour. Capital is laying people off due to, or at least using the excuse of, AI capabilities which even they believe are only nascent. If you have 10x developers, you only need 10% of the headcount. The snake is eating chunks of itself for the sake of a short-term, narrative-based pump. This is the hustle economy manifesting as autocannibalism, via shortsighted layoffs and a frenzy of enshittification by means of attempted justification for bloated AI valuations and hype. I can only imagine that, when the narrative cools, AI boosters are going to look at their mangled codebases and think "Well... shit".

That's a large part of what AI represents in tech. The capacity to enshittify things more efficiently than ever before. Yes, you can ship very quickly! You can ship as much as you want. At the cost of rotting your codebase.

I predict some good times for engineers with strong foundational skills (i.e. actual engineers).

Conclusion

In the 2010s, the economic narrative was still relatively meritocratic. Startups were a fairly new, seemingly exciting thing. There was a lot of tech optimism, and some amount of feeling that it was possible (albeit unlikely) to make a lot of money by realising a good idea. A belief that smart people with good ideas could find funding, become rich, and improve society while doing so. Around the end of that decade, a frenzied period of obsession with meme stocks^[15] and crypto/day trading. That didn't pan out, so moving into the 2020s, grindset and side hustle culture really took off. A return for the many to a more grounded vision of entrepreneurialism. Albeit an extractive, exhausting one. And that didn't work either. Busting your ass for 40 hours at your low-paid job, and another 20 in the gig economy, did not make people rich. It just exhausted them.

And now, narratives of achievable personal financial security are failing entirely. People are increasingly unable to believe in stories. For some, self-delusion does remain in the form of AI fairytales. For the more clear-eyed, narrative has collapsed - all that remains is nihilism.

As is ever the case, more than ever, we need class consciousness.