Well-written blog post about something I had never, ever, considered. Maybe one day this will come up and I'll remember that something like this is possible. The general trick of "rewrite the entire webpage based on some external API status" seems like an interesting and powerful (... and dangerous) tool.
In terms of achieving the overall goal of "greening" the web, I would be very interested to see even some back-of-the-envelope math explaining how much energy this technique saves and how much of an impact that would have. Understanding the energy cost of webpages and massive edge rewriting serverless platforms like Cloudflare is unintuitive to me.
Cynically, my impression is that the overall approach is more like "greenwashing" than "greening" — at best I would guess it "raises awareness", with all of the implications to efficacy that such actions usually have. I followed the links through from the blog to The Green Web Foundation's FAQ [0] and did not find any tangible estimates of energy impact nor any attempts at measuring such an impact. The success metrics [1] they list do not include any measure of energy saved or understanding of the potential to save energy; rather, basically just attention.
If we'd really want to lower the energy usage of our applications we probably have to simplify along the whole chain. Using Astro means the whole process relies probably on NPM, the build cycle is probably also more energy intensive. Static HTML hosted on some dirt cheap server is probably more environmentally friendly than hooking up to some API and giving us various experiences based upon the mix of the grid.
To me the savings as mentioned in the article feel rather superficial compared to the energy consumed just getting the basic infra for this up & running. Somehow this feels like greenwashing.
Centigonal · 1h ago
I was going to write a more critical post about how PoW cryptocurrencies and AI are probably better targets to go after than Javascript execution on the edge. Upon further reflection, though, this might actually be a good idea.
1. This is about demand response (using less power during times of peak fossil fuel demand), while AI and crypto mostly affect base load.
2. Crypto and AI operators make too much money from their depreciating hardware to consider implementing demand response, but website operators would probably care less about externalizing the inconvenience of their demand response to their users and taking credit for the environmental win.
The only question left is "if you divide power usage from unnecessary JS execution by the difference between base and peak demand, is that ratio significant enough to warrant the extra SWE and infrastructure overhead?" To me, this feels like the kind of question to answer before even embarking on developing this kind of standard, so I'm sure the GAW AG can point me to an estimate of the total addressable impact. Right? Right?
danbruc · 14m ago
Where is the comparison of the power usage between the two modes, is this not the entire point of the exercise?
Liftyee · 1h ago
This is an intriguing idea that I haven't seen before, and I wouldn't have thought of it to be honest. From an "efficiency optimisation" perspective it's interesting... but if implemented badly could have some questionable implications. There must be an off switch, otherwise could it be unethical to give people in "dirtier-energy" countries a "second-class" experience? Dictating "your power is not clean enough" and taking away the autonomy gives me a sense of unease, though vehicle emissions laws are already a thing so perhaps it shouldn't be so alien.
I propose a more obviously user-beneficial application for this sort of graceful-degradation design: detect if the user's bandwidth is low, presenting a trimmed-down version if so to maintain a usable website. Too many sites these days are completely nonfunctional on slow (rural/patchy/etc.) connections with tens of MBs of scripts and frameworks.
zajio1am · 45m ago
The article about Green Web on a blog that has scrolling implemented in some horrible way giving both slow experience and visible CPU usage (and therefore also power demand) spikes on my desktop...
webdevver · 37m ago
all this talk about doing more with less - i prefer, doing more with more!
In terms of achieving the overall goal of "greening" the web, I would be very interested to see even some back-of-the-envelope math explaining how much energy this technique saves and how much of an impact that would have. Understanding the energy cost of webpages and massive edge rewriting serverless platforms like Cloudflare is unintuitive to me.
Cynically, my impression is that the overall approach is more like "greenwashing" than "greening" — at best I would guess it "raises awareness", with all of the implications to efficacy that such actions usually have. I followed the links through from the blog to The Green Web Foundation's FAQ [0] and did not find any tangible estimates of energy impact nor any attempts at measuring such an impact. The success metrics [1] they list do not include any measure of energy saved or understanding of the potential to save energy; rather, basically just attention.
[0] https://www.thegreenwebfoundation.org/tools/grid-aware-websi...
[1] https://www.thegreenwebfoundation.org/tools/grid-aware-websi...
To me the savings as mentioned in the article feel rather superficial compared to the energy consumed just getting the basic infra for this up & running. Somehow this feels like greenwashing.
1. This is about demand response (using less power during times of peak fossil fuel demand), while AI and crypto mostly affect base load.
2. Crypto and AI operators make too much money from their depreciating hardware to consider implementing demand response, but website operators would probably care less about externalizing the inconvenience of their demand response to their users and taking credit for the environmental win.
The only question left is "if you divide power usage from unnecessary JS execution by the difference between base and peak demand, is that ratio significant enough to warrant the extra SWE and infrastructure overhead?" To me, this feels like the kind of question to answer before even embarking on developing this kind of standard, so I'm sure the GAW AG can point me to an estimate of the total addressable impact. Right? Right?
I propose a more obviously user-beneficial application for this sort of graceful-degradation design: detect if the user's bandwidth is low, presenting a trimmed-down version if so to maintain a usable website. Too many sites these days are completely nonfunctional on slow (rural/patchy/etc.) connections with tens of MBs of scripts and frameworks.