>I tried to point out that we're only talking about this because the public ledger makes it easy to measure such things, and we have nothing to really compare it to and we judge no other industry this way.
I think of this as the Wikipedia effect. People perceive Wikipedia as unreliable because they can perceive the mechanism by which it can be wrong. Study after study has shown Wikipedia to be more reliable than other encyclopedic sources that simply assert their accuracy while leaving the mechanisms by which they work hidden.
It is similar to the idea that something that saves 5 unidentified lives at the cost of one identified life is considered far worse than something that costs 5 unidentified lives to save one identified life.
Crypto and AI have also the problem that people cannot understand scale. How much is a million litres of water in terms of swimming pools, or showers? Thing uses as much electricity as [country that uses gas for heating and does not have typically have air conditioning]. Or Training an LLM produces a big looking number of CO2 which is as much as n cars in their total lifespan which makes it look like a different scale. Weirdly, the last figure I read on that suggested that if 10% of Americans trained their own personal leading edge LLM from scratch, cars would still emit more CO2. I don't know if that is true or not, but it frames the entire thing in a different light.