For my sins, I was taken down the rabbit hole of a document called "The Carbon Impact of Web Standards" (no, I'm not linking to it).
Friends, I feel nauseous. I'm getting dumber just leaving this tab open. Does it start with a situated analysis of the potential contribution of web content to overall GHG emissions? Lol, no.
Nor does it start from any estimate of real-world site construction (e.g., via the HTTP Archive).
It's so, so much worse.
I shit you not, the methodology is to load small-ish web pages with *no JS*, totally ignore power use from screens and radios (you know, the big ticket items?), guestimate that the average device uses power like a gaming rig, then do shoddy multiplication.
To what end? To try to extrapolate how much energy `<div>` or `<video>` use, totally ignoring the actually dominant factors.
This is WILD.
It's *fine* to want to do things but don't have much skill. That's great. It's how we all start. But for the love of crom, do *not* suggest wild-ass speculation deserves a working group at an international standards body.
It will not go well.
Aaaaaanyway....
Netflix can now serve 100Gbit/s of video (so something like 12,500 individual 4K streams) with an appliance using 100 watts of power. That’s 8 milliwatts for each 4K stream. Remember that number the next time someone tells you that watching a Netflix show is as bad as driving an SUV or some shit. https://people.freebsd.org/~gallatin/talks/OpenFest2023.pdf
@slightlyoff personally I’m ok with estimating a carbon cost for each byte transmitted. It’s a shallow measurement, yes, but it’s easily measured and a decent starting point.
What we really need is device energy usage, but Mozilla’s efforts seem to have stalled a long time ago
@slightlyoff agree, there must be some resemblance of reality prior to publishing
On the other hand, metrics evolve and when it comes to energy/sustainability we’ve currently got sweet bugger all. Runtime energy usage required yesterday!
@benschwarz Look, I've actually *done* the back-of-the-napkin on client-side web energy use. It's not nothing, and growing faster than I'd like (thanks, JS!), but it's also:
- low-ish (< 0.25% of total societal power use)
- getting decarbonized with the grid
- significantly less than, e.g., Bitcoin for infinitely more value
See also: http://energyliteracy.com/
@benschwarz The confounding factor is substitution: what fraction of web use is in lieu of some other sort of computer program that would have done the same thing in a different way?
*Gross* contributions for web-based power largely have to do with data center use, and that's getting tackled orthogonally.
Using a different HTML element (per this paper) is *non sequitur* and a missallocation of engineering time.
Worse, when this is falsified, arguments built on it will wash away too.
@benschwarz Intellectual honestly doesn't matter if all we're trying to do is to perform values.
But living them requires more of us.
@benschwarz Anyway, I'm *begging* web developers not to facilitate greenwashing. The carbon credits scandal needs to be a cautionary-enough tale that we are able to dismiss the next attractive nuisance out of hand when the numbers come back fugazi:
https://web.sas.upenn.edu/pcssm/news/carbon-offsets-are-unscalable-unjust-unfixable-joe-romm/
Noted climate researcher Joe Romm has published a paper debunking carbon offsets as basically useless and holding back emissions reductions....Continue Reading Carbon Offsets Are Unscalable, Unjust, & Unfixable — Joe Romm
@slightlyoff not bad compared to bitcoin transactions, but then again I don’t do that on my phone.
Energy ain’t so cheap when your battery is dead by early afternoon!
Alex Russell
Marsh Ray
Eaton
Jeffrey Yasskin
benschwarz
Mark Stosberg