Thursday, June 01, 2023

Spengler

David Goldman writes for the Asia Times, originally under the pseudonym Spengler (after Oswald Spengler, who wrote about the decline of the West).  His economic predictions have been unusually good. Steve Hsu interviewed him at Manifold podcast. As that is one of the podcasts that offers transcripts, that is available for those who prefer that medium.

He was originally focused mainly on classical music, and was all-but-dissertation in music theory at CUNY.  "Then, my first child was born and I noticed immediately that she wanted to eat almost every day." He began taking Wall Street jobs. He has been politically involved since the 80s and is not afraid to be radical, as one can see from both the Wiki and the interview. He refers to Republicans and Democrats as "Dumb and Dumber."

The question which has obsessed me for many years is why the West would develop the most inspiring and elevating high culture in the history of the human race. And then effectively abandon it. 

This interview includes some American economic observations, such as what he considers felonious market manipulation by huge players (Bank of America, where he had previously worked, for example) in the crash of 2008, but mainly focuses on the economics of China in competition with the US. It is not mere doomsaying, but it is alarming. He considers some sectors, such as EV's, to already be effectively lost. Tesla? Largely a Chinese manufacture, actually. BYD Auto has introduced the Seagull EV for about $11K, and Wuling (with GM) will be offering one at $9K later this year. They don't need the American and European markets for these, not at all. He also points out the uses of AI to solve lower-tech problems for poorer countries, giving the example of Bangladesh and the packaging of chili peppers, a key commodity for them.  A single bad pepper can spoil the whole bag, and losses of 15% are now reduced to near-zero because of AI. Of course they are going to buy this technology from China. 

Lots of topics covered in an hour. Another interesting bit.  He makes reference to predictions of his coming true more quickly or more slowly than he thought, and it occurred to me that in popular media, unfavorable reporting is likely to call those inaccurate, not just wrong about the speed.

2 comments:

David Foster said...

I think Goldman greatly overrates any Chinese advantage in providing AI systems for product inspection in manufacturing. There is a lot going on in this field in the US and other countries. For example:

Neurala (Boston)
https://www.neurala.com/

Landing.ai (Palo Alto)
https://venturebeat.com/ai/landing-ai-launches-product-inspection-platform-for-manufacturers/?utm_source=dlvr.it&utm_medium=twitter

Covision Lab (South Tyrol, Italy)
https://venturebeat.com/ai/landing-ai-launches-product-inspection-platform-for-manufacturers/?utm_source=dlvr.it&utm_medium=twitter

Kitov (Israel)
https://kitov.ai/

I also think Goldman overrates the importance of 5G for factory automation. There are many other protocols that can be used, and I think the idea that very short latency times are necessary is in most cases wrong. If I have a CNC milling machine, I'm going to want its controller adjacent to it or just built into it rather than 1000 feet away at the end of a 5G link. Communication needs to happen, but it's generally going to be for uploading reports, downloading designs, etc, rather than something requiring instant service. Same for inspection systems.

james said...

Yes. You don't want latency, and you don't need to have each system's compilation continuously fed from a remote server when for a few bucks more you avoid any problems from network transients. The network would be for overall control, monitoring, re-downloading new configurations, and routing material between stations. And calling for pickup and delivery