[simonh]

Yes, but what is that code doing? If it's operating a self-driving car better software gives better mpg, a faster journey, a shorter, safer journey. Maybe it's optimizing airline seat usage and routing again. Maybe it's running a just-in-time manufacturing or logistics system. Perhaps it's a medical diagnostics system helping people live longer more productive lives. I use a smartphone to message home so my meal is cooked when I arrive and doesn't need to be cooked in advance and then re-heated. Even hedge fund investment systems channel funds to more efficient productive companies because those are the ones yielding better returns. In the end, these are all yielding energy efficiency gains through reduced or more efficient resource utilization. Name any useful jobs software improves, and I'll show you a resource utilization efficiency, and all resource production and delivery is based on using energy.

[mirimir]

OK, here's a funny story.

Some years ago, I found myself in a very small art gallery in a very expensive place. In front of my face was a weathered pine board, with 96 (as I recall) small red rubber serum caps nailed to it, in a reasonably regular array.

The price was ~$1000 ;)

Anyway, maybe the software is just charming.

[simonh]

I did say 'useful jobs', by which I mean any purpose that itself contributes to economic activity. Some software is consumed by itself and serves no useful purpose in that sense, such as computer games or computer art. Those do contribute to GDP, so I was wrong to exclude them, but the impression I got from the poster was that software 'doesn't count' when it comes to the relationship between economic activity and energy utilization, whereas mostly it certainly does. Although the post seems to have been edited since, so the last paragraph does now acknowledge that software can improve resource utilization.