Thanks, Addy, for the supportive comments!
+1 to the comments from Ilya and Yoav. In terms of NetInfo API usage, most
> large partners for adaptive loading (eBay, Tinder, Twitter etc) relied on
> effectiveType while a smaller minority (Algolia) used more of the signals
> like rtt. See
> https://www.algolia.com/blog/engineering/netinfo-api-algolia-javascript-client/
> .
>
Yepp, especially the Algolia post is a great resource to be aware of.
> saveData had some minor adoption sites worked on by web performance
> experts (
>
> https://nooshu.com/blog/2019/09/01/speeding-up-the-web-with-save-data-header/)
> but I didn’t see this go mainstream.
>
Additional data point:
https://chromestatus.com/metrics/feature/timeline/popularity/2217. Note,
though, that I see the save data use case distinct, as pointed out in
https://raw.githack.com/tomayac/netinfo/relaunch/index.html#distinction-from-the-save-data-use-case
.
> Perhaps the most helpful insight I can share is that the majority of users
> just needed the ability to distinguish whether a connection was “fast” or
> “slow” (not granular at all) while it was the most advanced partners
> (Facebook) that felt they needed even more granularity than ECT currently
> offers.
>
That's definitely good to know. With the speed buckets approach, it seems
like we could cater for all needs.
> Happy to go into more detail on any of these points if needed.
>
Thanks for the kind offer, will get back to it if need be :-)
Cheers,
Tom