One thing I've noticed is that almost all new web frameworks I come across nowadays seems to be catered towards search engine optimization (i.e.: getting 99/100 on Google PageSpeed Insights).
It's a very specific kind of performance. It doesn't measure long running apps, interactions and visualizations.
When you play a video game you might have to download and install it first, which might take quite a while, but we don't say "this game has bad performance" because of that.
Lighthouse measures that first second or first few seconds after you visit a site at position 0/0. It doesn't care about how much bloat comes after you scroll. It discourages you to pre-load resources (including JS and CSS) which you don't _immediately_ need for that first hit.
It is really about content and marketing websites with low interaction. You can absolutely make good interactive applications that also get a high Lighthouse score, but the overlap is not necessarily always feasible or the best tradeoff to make.
> Lighthouse measures that first second or first few seconds after you visit a site at position 0/0. It doesn't care about how much bloat comes after you scroll. It discourages you to pre-load resources (including JS and CSS) which you don't _immediately_ need for that first hit.
This is true - or rather, was true. We're making lots of progress on a mode to measure performance beyond the initial cold load. Check out https://web.dev/lighthouse-user-flows/
> It is really about content and marketing websites with low interaction.
Well, that isn't our intention at all, but I can see how to might come to that conclusion. We've only recently started investing in how to measure interaction. See https://web.dev/inp/
Thanks for the work you and your team are doing. Lighthouse, while not perfect, is a helpful tool. It sounds like it would be a fun project to work on.
AFAIK they still use it, and it plays a huge role (based on my fairly limited experience). A couple months back I did a little consulting on the side with a small ecommerce store that had recently struggled to get any traffic despite all their SEO analytics showing green except for page speed. They ranked poorly for just about every keyword that they had previous been doing well on.
Bit of CDN/Caching, upped them to a better hosting plan and they're back on track again - rankings are back to normal and page speed is way up.
A separate signal is user leaving SERP, going back to it and checking another site - somewhat clear sign of failure of a site to deliver. And users leave if your page loads slowly.