I was scrolling on my phone on Thursday morning, half paying attention, and noticed something that felt strangely recurring. Everyone seems to be an AI expert now. The same accounts that were crypto experts eighteen months ago. The same accounts that were social media growth experts before that. Before that, something else. It's like the nouns keep changing but the shape of the thing does not.

It feels like this is more of a human one than specifically a tech one. We've probably been doing this for a very long time.

I'm not writing this from outside the cycle. In 2021 I had a 2000-follower NFT Twitter account, ten different crypto wallets, the codes memorised, Discord open at midnight, the whole thing. Before that I bought Instagram growth courses and that's actually how I learned design. Then copywriting courses, which is how I learned to write sentences that don't waste the reader's time. None of it was wasted, exactly. But the specific bets — the floor prices, the audience-growth playbooks, the "this time is different" energy — those mostly evaporated. The skills stayed with me for the most part but the certainty did not.

So I can write about this with some compassion, I think, instead of contempt. I've been the earnest person refreshing the chat at midnight, sure that this time was different. I'm recognising the shape because I've ridden it.

And the shape goes back a long way.

In the Netherlands in the 1630s, a single rare tulip bulb traded for roughly ten times a skilled craftsman's annual wages at the peak. The contracts passed hand to hand in taverns. Most of the people getting in weren't rich. Nope, they were craftsmen, shopkeepers or ordinary people who felt behind. In February 1637 the market collapsed in a few weeks. The flowers were real of course and people still grew them. The economy of speculation that wrapped itself around the flowers was a different thing entirely, and that was the part that changed significantly.

Also let's head to Britain in the 1840s, when Parliament approved hundreds of railway schemes. Newspapers ran constant coverage while clerks and clergymen mortgaged what they had to buy shares in lines that were never built. When it collapsed in 1846 it took out the savings of middle-class Britain. The railways that survived (the actual Victorian infrastructure) were a small fraction of what had been promised. The tools were real and the hype was a separate economy sitting on top of them.

And before either of those we had Cicero complaining in his letters about speculative property prices in Rome. Emperors kept passing sumptuary laws trying to stop citizens from bankrupting themselves performing wealth they don't have, buying imported silks, pearls and exotic fish to put on the table. The nouns were different yet people went broke signalling status in whatever the current status good was. This was two thousand years ago and yet feels oddly similar.

Then there's the one I actually remember the edges of, even though I was a kid. The dot-com thing where "Eyeballs" stood as a valuation metric. Think Pets.com or a sock puppet on a Super Bowl ad. The internet was real and was about to change everything. The hype economy around it in 1999 was a separate thing, and it burned out in about eighteen months while the actual internet kept compounding quietly underneath. The people who were "internet experts" in 1999 mostly disappeared. The people who were heads-down building things (and weren't, on the whole, the loudest accounts at the time) are the ones who look prescient now.

I think what all of these have in common, as far as I can tell, is two layers stacked on top of each other. There's a real thing (flowers, railways, the internet and AI). Then we have a performance economy that grows around the real thing, sells access to it, accumulates mystique, and then evaporates faster than the real thing does. The real thing keeps going.

The last pre-AI version of this was Excel gurus in 1998. Before that, dot-com eyeballs. Before that, railway bonds. Before that, tulips. We keep doing this. You can see the pattern, right?

The thing I find genuinely useful about noticing the shape is not that it makes you immune. Because it's really difficult to stay immune when many are loudly claiming your source of income is under threat. You don't read an essay about tulips and become hype-proof; that's not how brains work. What it gives you, maybe, is orientation. You stop feeling lost inside the pattern. Oh, this is the mystique-building phase. Oh, this is the course-selling phase. Oh, the infrastructure is starting to become invisible — that means it's almost normal.

It also helps with the feeling of being behind, which I think is the part that quietly costs the most. A lot of the energy that pours into hype cycles is people who feel late trying to catch up. The Dutch craftsmen weren't rich. Nor were the Victorian clerks. The people buying $400 NFT-trading courses in 2021 were mostly people who'd missed crypto in 2017 and didn't want to miss it again. Feeling behind is often the fuel and that cycle runs on it because the cycle is selling a way to stop feeling behind. FOMO if you will.

And then the other reframe, the one that matters most to me, is that getting swept up isn't a waste. I learned design in the Instagram era. I learned to write in the copywriting era. I learned what hype economies feel like from the inside in the NFT era — which, honestly, is most of why I can write this post at all. The specific bets failed. The underlying learning stayed. Skills compound. Bets usually don't. I think if you stay roughly honest about which is which, you can ride a few of these and come out fine. Well, relatively fine, that is.

So here's where we are now, or at least where it looks like we are from here. The AI tools keep getting better, the documentation is mostly free, and the people doing the actual interesting work are mostly not the loudest accounts — they're people like Simon Willison quietly writing about what works and what doesn't, swapping out vocabulary like "prompt engineering" for "context engineering" because the field is maturing past its mystique phase. The "AI expert" economy stacked on top of all that is the tulip part. It will burn out faster than the thing itself, the way it always does, and most of the loud accounts will quietly become experts in something else.

I'm not going to predict when. I don't know. I'm not above any of this — I'll probably get swept up in the next thing too, at least partway, before I notice. But it's a strange small comfort to know the shape. To recognise the mystique-building, the course-selling, the slow turn into infrastructure. If you enjoyed reading this post, feel free to share your thoughts on any one of our social media channels. Keep thinking, keep creating, and let's build something cool. Even if it's genuinely difficult to find users for these tools without marketing. In the next post I'll probably share more on the marketing side and how I'm approaching it.

References

  • Charles Mackay, Extraordinary Popular Delusions and the Madness of Crowds (1841) — the classic source on tulip mania, still in print.
  • Andrew Odlyzko's papers on the British Railway Mania of the 1840s, if you want a more recent academic treatment.
  • Simon Willison — writing about AI pragmatically, without the mystique.