https://www.ft.com/content/8e9ae7a4-7209-4e2c-aa36-f3af77d6ce1f?syn-25a6b1a6=1

@Matthuber78 This is year over year. So, once they completed electrification where appropriate, there were no great gains to be had by electrification. Maybe we need to go hard into automation and dark factories.

@Matthuber78 Exact parallel in UK energy. Renewables were 71% of generation in April but the system hasn't been redesigned around them. Marginal pricing still lets gas set the price. REMA was supposed to be the factory redesign. Four years on, still discussing.

@Matthuber78 @phl43 Isn’t this chart showing the **change** (I.e. growth) in productivity not the **level**. Productivity didn’t fall —it stayed at the new, higher level or grew more slowly.

@Matthuber78 Electricity was invented in the 1880s, but the real TFP explosion waited until the 1920s when factories were finally redesigned around electric motors instead of steam shafts. Classic case of tech diffusion lag.

@Matthuber78 AI parallel is premature. We're still in the 'electrifying the steam factory' phase, same layout, slightly faster output :/

@Matthuber78 Is the drop from small electric motors going to small workshops and then everyone's garage?

@Matthuber78 As others have said, this is the rate of change. So the productivity gains were not short lived. But what is wild is the rate of diffusion. Roughly 10 years for widespread adoption.

@wdseelig @Matthuber78 @phl43 That's how he phrased it. Gains did not continue.

@wdseelig @Matthuber78 @phl43 Yea the second derivative became positive for like 10+ years… that’s something of dreams… productivity accelerated

@Matthuber78 I don't get it. What's the y-axis? LABEL YOUR CHARTS.

@Matthuber78 Does this chart show that it is process/systems rather than the core technology itself that leads to productivity gains? What technology enabled a decade of sustained 2% gains 1945-55? Or the increases of 1960s, 1980s, early 2000s?

@Matthuber78 This merely confirms the cyclical pattern of diffusion. Obviously, electricity itself could not continue to produce linear productivity gains once widely adopted. 1960s = automation; 1980s = JIT/CAM; early 2000s = infotech. Reasonable to expect renewed spurt from AI thru 2030s.

@Matthuber78 @Noahpinion that’s a pretty big boom though! Unfortunate consequence of formatting a geometric progression like the above is the area under the curve doesn’t directly translate to the cumulative change; it’s like 1.03^12 ≈ 1.42, which is ≈ 1.01^35… ie as much total gain 30-42 as last 35y

@Matthuber78 AI may duplicate this improvement in efficiency; you damn betcha nuclear-electric power will; used at every step in production from natural resource to final retail sale.