It's really fun when they get into the computing requirements of planning an economy using the Harmony algorithm/neural network. They keep talking about things in millions of operations per second and how a super computer could solve a large economy in about 10 minutes.
When you take these numbers now and apply them to current consumer grade chips, you could take their exact model and solve an economy with several million inputs/outputs in roughly 1 second. Compared to the 2-10 weeks it would take on a 68020.
It's really sad that the mass adoption of computing has led to things like NFTs and Crypto Currency. Just absolutely wasted operations that do nothing but waste means of production and fractional products. Those trillions of operations could be going towards simulating production goals in central planning. 1, 5, 10, and 15 year plans could be generated in seconds now and presented to all for vibrant democratic input and decision making, instead were using deterministic machines to recreate the chaos of market systems and the runaway crises that come with them.
Luckily Cockshott is still active and has a YouTube channel where he updates a lot of this. There's also his SourceForge page with his Harmony implementation in Julia that should definitely be forked over to another repo on a less bloated site.
His old implementation is in vector pascal and he has a few videos on why he chose Julia for the newer implementation.
:countdown:
Lmao, it's literally just him sharing his lecture slides. I think one of his students told him to start it because he only started uploading them like 2 years ago