"throw more CPU at the problem" is abstracting the cost, which, well, it's "throw more watts (CO2)* at the problem"

CPU power is abundant but a watt is a watt.

*: wind and solar renewables are bad at constant drain, due to periodic nature, but also you're now pulling renewable watts from the available pool, and things otherwise tasked to them are going to either run out, or drain batteries you'd use over night

using more of the renewable capacity, unless we have more than enough for even peak loads, means someone else is using more "dirty" and nonrenewable supplies

buying an electric car to replace yours in California is better than in Pittsburgh

but mostly because in pgh it's just burning coal to power your car, which is less efficient than a gasoline engine with regard to watts per CO2

like transmission losses mean its a lot better to use nearby renewables than "sell" excess to nearby grids, so increased nearby renewable use is still better, but like, you're still tasking things not otherwise burning fuel to burn fuel

anyway I've been thinking about this a lot lately, ever since I saw a lot of tweeting immediately after much more efficient processors were announced, that it would mean "a spike in available cpu for the same project cost",

instead of making the project take the same processing power, but make the software more energy efficient. https://dannyvankooten.com/website-carbon-emissions/

eliminating a 20kb JavaScript dependency in mailchimp for wordpress led to a reduction in 59k* kg of CO2 emissions per month.

think about what 20% more efficient infrastructure would do if all the projects stayed same level of cpu use

(*: his calc is explained, conservative gut feeling is hack off a 0)

(probably more, but it's also not getting into computational cost on either end. but the thread still works without it.

see: ) https://twitter.com/NireBryce/status/1340370704513069057?s=19