This is a pretty good quote from Matt Levine:
"I used to write a lot about crypto. The reason I liked writing about crypto is that it seemed to be rediscovering all of regular finance from first principles, quickly, in public. It was a fabulous laboratory for understanding financial structures. If you wanted a public demonstration of why, I don’t know, infinitely leveraged shadow banks were bad, you could wait 20 minutes and crypto would give you one."
I made the same point before: the most interesting part of the phenomenon wasn't that it's necessarily good or bad, that it's energy-hungry or not - but that you're getting an empirical validation of many of the crusty old principles of "classical" finance. Funnily, delivered to you by the folks who rejected all that dogma in the first place.
@lcamtuf what’s wild to me is to see people struggling to use AI constantly talking about how they have to check the output of the AI to make sure it’s safe and correct. Whenever I ask them why they don’t just apply the same systems that they apply to human work to AI… yeah, I get a lot of awkward looks.
What really blows my mind is when I talk to them about the automated ways you can use AI to check AI output. Some of them get it and it’s interesting to see their expression.