>There are many ways to go about speeding up the websites for the end users that are less technically involved for everyone and require a new subset of markup and standard.
>Because the people creating the standards (i.e. Google)
Google doesn't create standards. They are one of the major stakeholders, but they don't have the final say.
>That's the whole issue here.
No. The issue is that Google built a browser that is better than anything now, and was better than anything out there for the past decade and got themselves into a dominant position. They haven't yet abused that position - contrary to a lot of the posts in this thread - but people are nervous. I can respect that viewpoint because it is something to worry about.
Google could simply cache the web pages and strip them naked. Reader mode on browsers do a good job of it on most sites and they can publish guidelines on how to structure your site to make it easier for reader mode to work which wouldn't be that problematic (it would also encourage accessibility).
>Google could simply cache the web pages and strip them naked
You think Google could get away with stripping or editing commercial pages by default? Be serious.
>Reader mode on browsers do a good job of it on most sites
Then use it.
>they can publish guidelines on how to structure your site to make it easier for reader mode to work which wouldn't be that problematic (it would also encourage accessibility).
Google actually does a lot of that. They publish a lot of guidelines, including testing tools (like Lighthouse) to have developers build faster websites. Case in point: https://web.dev/measure
> You think Google could get away with stripping or editing commercial pages by default? Be serious.
No one said anything about default. It can be opt in like amp but the difference is here simply that they don't require a new standard.
Other ways they could solve is just disincentivise slow loading pages. They have said speed is a factor and tried to introduce some features such as showing how fast a page is on chrome as an experiment.
>It can be opt in like amp but the difference is here simply that they don't require a new standard.
But that's exactly what amp is. An opt-in set of standards for creating fast-loading, cach-able pages. There is also no reason why websites couldn't replicate this outside of amp ... but they don't.
>Other ways they could solve is just disincentivise slow loading pages.
They do that ... in all kinds of ways. The problem is that with their position, they can destroy a lot of businesses if they apply these kinds of rules haphazardly. There's no simple solution here.
> But that's exactly what amp is. An opt-in set of standards for creating fast-loading, cach-able pages
Not exactly. For one, you could introduce opt in for Google to cache your page through a meta tag. No need for entirely new markup and tools. I have no problem with amp as a concept but the implementation leaves me sour. Amp specification has been growing in complexity that I don't notice speed differences on many sites anymore.
They also inject their own js which I don't find amusing.
Many sites are abusing amp to rank on Google. Reddit, pinterest, etc. It's a terrible experience as a user. They also didn't fix url and domain problem until recently. It's a usability problem.
I can't search for the specific tab by the domain name of the site on my browser. They default opt-in to downloading content in the background for amp feed on chrome android. There are many little annoyances like this one that make me dislike amp.
Plus, amp isn't heuristic based. If my network speed is fast enough, I don't need the stripped out page.
So why isn't it done?