# The Complexity Ratchet > Source: > Published: 2026-05-15 00:00:00+00:00 [Vivek Haldar](/) # The Complexity Ratchet Recently, Jasmine Sun wrote [a long op-ed in the New York Times](https://www.nytimes.com/2026/04/30/opinion/ai-labor-work-force-silicon-valley.html) about the notion of a permanent underclass. Everyone at the big AI labs is convinced that AI will eventually replace all human labor — all human knowledge work — and that if we don’t calibrate our technology and policy choices properly, the result will be a new permanent underclass. At this point it’s almost a meme in Silicon Valley AI circles. On her Substack, she later wrote [a follow-up piece](https://jasmi.news/p/party-in-the-permanent-underclass) where she walked through some of the rebuttals — and the rebuttals to the rebuttals. One of the most prominent rebuttals to an argument like that is the good old Jevons paradox: when the cost of a desirable good goes down, the demand for it skyrockets. So even if artificial intelligence becomes much cheaper, you’ll just have more and more demand for it, and more and more jobs. Sun’s rebuttal to that rebuttal is that, sure, you might have more demand for AI, but that doesn’t necessarily mean the demand gets fulfilled by *humans*. I think that’s the crux of the argument. ## The Complexity Ratchet There’s one part of this argument that often gets missed. I don’t know if it has a name — I like to call it the **complexity ratchet** — and I think it’s what might still end up saving humans. If you look at the tech tree of civilization, it’s one long walk up the ladder of complexity: from mud huts to wooden buildings to concrete buildings to skyscrapers, or from wheels to cars to jet airplanes to spaceships and rockets. Each of these is orders of magnitude more complex than the one before it. ## Why Spreadsheets Didn’t Kill Accounting A very clean, recent example — one within most of our lifetimes — is the advent of the personal computer. The killer app for the personal computer was the spreadsheet. People didn’t really know what to do with a personal computer until VisiCalc came around and they saw spreadsheets. That’s what made everyone realize what computers could actually be used for. When spreadsheets came out, the overwhelming lament at the time was that they would wipe out the accounting profession. Decades later, we know they didn’t. Quite the contrary — they probably 10x’d or 100x’d the number of accountants. Why? The complexity ratchet. Because spreadsheets made it so easy to do more and more complex financial engineering, the result was that you needed more and more human accountants to stay on top of all that complex financial engineering. You can make the same argument for most digital technology. I don’t think you’d have laws and bills in Congress that run to thousands of pages if you didn’t have word processors. ## How Technology Raises Civilization’s Complexity So the complexity ratchet argument is that all of these technologies irreversibly ratchet up the level of complexity in our civilization. They let us create, manage, and live with more and more complex legal, societal, financial, and educational structures. And those structures just become part of our lives, part of our civilization. Even though you need machines to manipulate a lot of these systems, you ultimately also need a lot of humans. The financial system is a great example: you need a lot of complex machinery to operate today’s financial markets — and not just the markets, but all the structures around them: companies, LLCs, C corps, S corps, public and private corporations, shareholder and ownership structures. You created a lot of complexity, and you did so irreversibly. You needed machines to climb that complexity ladder, and you need machines to maintain civilization at that level of complexity. But all the while, you didn’t do away with humans — just as with spreadsheets, you didn’t reduce the total number of accountants in the world. So I think the complexity ratchet might be a corollary of — maybe a necessary companion to — the now-familiar Jevons paradox argument. ## Will AI Make Us Accountants or Farmers? Now, there’s a rebuttal to *this* too: okay, AI will enable a massive leap in the complexity of the structures we build and use, but sure, the AI itself will manage all of it for you. I think that’s still plausible at this point. I don’t think we know enough yet to say whether it goes the way it did for accountants — where you still needed a lot of accountants — or whether it goes the way of agriculture, where we still produce more than enough food for the entire planet but the number of farmers, in both absolute and relative terms, completely fell off. We just don’t need many farmers anymore. So will all this go the way of accountants, or the way of farmers? You can draw some historical parallels here. Farming is a very physical activity — it’s from the world of atoms, not the world of bits. There’s some ceiling, some limit, to how much food you need or want, and how much food you can store and move around. Accounting and law and finance, on the other hand, are purely abstract, completely conceptual. There’s no ceiling to the complexity of financial structures, as long as someone is still getting some utility out of it. If you believe that argument, then even with AI — even with the complexity ratchet enabled by AI — you’re still going to need a lot of humans in the future to stay on top of all that complexity. In other words, we go the way of the accountants rather than the farmers. Only time will tell which way we end up going. But I wanted to talk about the complexity ratchet argument, because the Jevons paradox argument is pretty well known by now, and I haven’t seen the complexity ratchet argument made — at least not to my knowledge. All the best!