On crypto: it was even worse than Andrew Gelman says it was. The real outside experts in crypto were the banks, who had been running cooperative (if not distributed) ledgers for decades. And the journos never seemed to talk to them. Or maybe they talked to the banks, but didn't like what the experts said. Or maybe their little journo brains said "aha, conflict of interest! Dinosaurs of the past!! I don't have to listen to these guys."

The banks were intrigued by crypto for about 3-4 years after it came out. By then, they grokked that crypto was either Ponzi or money laundering. (Libra/Diem was not really crypto.) That would be 2013 or so? The banks were willing to sell support to the scamsters and chumps, such as deposits or clearing facilities or the like. But they otherwise stayed out of the crypto game.

Expand full comment

Financial journalists such as Weisenthal & Alloway (Odd Lots) devoted a vast attention to crypto, much of it uncritical. At least Matt Levine drew out the absurdity of SBF. I guess crypto journalism increases enthused traffic, sort of like Trump.

Now Tyler Cowen is euphoric over AI, which should give us pause given his crypto history. Which is not to say that LLM's won't be valuable tools (and trash), but it is too early to figure out how, and how well the technology is implemented. If we look at other new technologies, from railroads to radio to Internet, we see the wreckage of fraud and malinvestment. Cowen is much smarter than me, but what is in crypto or AI for him? Is he merely captured by the interest and investments of the tech bros such as Thiel?

Expand full comment
Sep 19·edited Sep 19

So, did you read Dave Karpf's substack today? The 1st item was on LLMs.

"There’s something darkly funny in Ethan Mollick’s latest post about the bright business-future of generative AI ..."

"Ethan Mollick teaches entrepreneurship at Wharton. It shows. He produces serious, methodologically-sound research. And he never comes within a mile of examining the social assumptions that his research is grounded in ..."

"Mollick and his coauthors find that GPT-4 improves consultant productivity and work quality on all these tasks. The gains were strongest for the low-performers [...]To Mollick, this means that (1) the business opportunities are phenomenal and (2) the people who get rich will be the first-movers who really develop their skills in this grand new landscape."

"And, I mean… sure? One could read the findings that way.

But an alternate reading would be something like “hey! I hear you think A.I. is a bullshit generator. Well, we gave a whole profession of bullshit generators access to A.I., and you’ll never believe how much more productive they became at generating bullshit! ..."

Expand full comment

Kara Swisher: ‘It was indeed not at accident that the Toxic Twins [Trump and Musk] attacked ⁦Yoel Roth—it was a coordinated assault by the goon squad. Read…

Has a bad link.


I suspect substack is helping you too much.

Expand full comment

Blumenthal: Government benefits ought not to have a time tax (beyond combatting fraud cost effectively) and Progressives should have had enough confidence that the tax could be minimized to accept the (unnecessary but politically convenient) work requirement for the RCTC.

Expand full comment

Gellman: This may conflate opinions abut the possibilities of blockchain with crypto currency. Maybe I'm being too generous, but I don't recall Tyler Cowen being particularly optimistic about purchasing bitcoin.

Expand full comment

The biggest tell to me about crypto was Paul Krugman's early-on 2-part question that went something like this, "what exactly is it replacing and why does that thing need replacing?" I never, ever read a good answer to that. But as an academic I was already skeptical because for years I kept hearing how blockchain was going to solve this or that food distribution problem but no academic could ever reconcile how blockchain "solved" a garbage in-garbage out problem. Make it more efficient, yes, but the food safety issue doesn't go away just because you do a better job tracking if you cannot correctly register the vector at the outset. It helps you locate the node of the problem, but so does record keeping. It doesn't change that somebody somehow has to swab a machine correctly. Carbon credits face the same problem. It's possibly surmountable but at the end of the day, you have to certify what carbon is sequestered initially, and I don't see that blockchain is greatly better than any other certification program. So, what is it replacing and why does it need replacing?

Expand full comment