The Iron Rule and the Soft Landing

John Loeber, responding to Citrini7's AI labour displacement scenario, writes:

Everything is always more complicated and takes much longer than you think it will, even if you already know about the iron rule.

He's right about this, and it's the part of the piece that deserves the most attention. The real estate broker example alone is worth the read — the technology to eliminate that job has existed for a decade, and yet Loeber just paid someone $50,000 for ten hours of form-filling. Regulatory capture, habit, and transaction anxiety are extraordinarily durable. Anyone forecasting AI displacement on a timeline of months rather than years is ignoring how humans actually adopt things.

The software quality argument is sharp too. Virtually all commercial software is, to use his technical term, garbage. The assumption baked into most displacement narratives is that current product scope is fixed — that AI eats the existing pie. Loeber's point is that the pie is nowhere near fully baked. Jevons Paradox. More efficiency, more demand. Historically well-supported.

But then the piece reaches. The displaced $180K Salesforce PM finding new purpose at the California Desalination Works is a nice image, but it's a political slogan doing the work of economic analysis. The US can't staff existing infrastructure projects. The skills gap between white-collar knowledge work and industrial megaprojects isn't a gap — it's a canyon, and "re-industrialisation" doesn't come with a footbridge.

And Jevons Paradox is load-bearing a bit too much here. The paradox says total demand for a resource increases when efficiency improves. It does not say the same workers capture the upside. Coal miners didn't get rich from steam engines.

The final reassurance — that the federal government's Covid response shows it can act aggressively when needed — is a strange place to find comfort. That response was also chaotic, inequitable, and inflationary. "It won't be efficient, but that's not the point" is a sentence that needs to work much harder than Loeber lets it.

Here's the thing: correctly diagnosing that we have time is not the same as showing we'll use it well. Loeber is right that the doomers underestimate institutional friction. He's probably also right that displacement will be gradual. But "gradual" and "manageable" are different claims, and only one of them is supported by the evidence he presents. A slow-moving crisis you don't prepare for is still a crisis.