As good an idea as prefers-reduced-data is, I can see the ad trackers’ eyes light up already: “Prefers reduced data? They must not have enough money to pay for a high data cap. Let’s send them an ad for a predatory loan!”
This is a very good idea that I would love to see implemented but it goes against the economic incentives driving websites today.<p>Lower fidelity would also mean to restrict tracking. Good luck with that. Internet advertising and data mining companies are not exactly renowned for their moral and ethical integrity. They would have no issue draining the data plan of a hurricane victim with ads for homeowners insurance.<p>It also means more work building those sites. Which is not something I see happening, given that in most projects there is little time to fix structural issues as is. Why fix bugs or optimize payloads when you can ship features?<p>I would also bet that if lower fidelity meant less (or no) ads, many people would use lower fidelity as a workaround for not having to see them. Which would be exactly what the world needs but is, again, entirely misaligned with most online companies’ economic interests.<p>Unless the incentives change, websites will continue to bloat until they break.
This is unlikely to work, simply because news sites online, (and most sites with an aim to make money) will want ads. What will stop users from going to the no-image version to avoid the ads?<p>It is technically superior, but in my opinion not very good for business.
So… in his particular implementation, the medium fidelity version is the full fidelity version minus about 22KB (uncompressed) of styles and JavaScript. On the wire, the whole HTML document (including inline CSS and JS) is roughly 6.9KB medium and 13.2KB high.<p>As it happens, the document is not minified. With minification and with a much lighter and less needlessly-indirect implementation of the JS parts, high would be about 9.7KB on the wire, compared to about 6.6KB for medium. And I haven’t tried stripping unused styles, which is the obvious thing to do when it’s all inline and functionally non-interactive; from a quick skim, I expect approaching half of the styles to be unused.<p>At these levels, the difference between medium and high seems fairly silly.<p>Then there’s low, which shaves off about half a kilobyte more, <i>practically</i> only differing in removing the smidgeon of CSS and turning images into text links to the images, which is a class of functionality that genuinely bandwidth-constrained users will want to do themselves in a browser extension, since practically nobody is going to do it (least of all the heavy ones).<p>All up, I say it’s a philosophical statement only, not something of any practical value. The only developers that are going to produce this kind of thing are the developers whose sites won’t really benefit from doing it.
The Client Hints specification covers a lot of this already - <a href="https://wicg.github.io/ua-client-hints/#content-adaptation-use-case" rel="nofollow">https://wicg.github.io/ua-client-hints/#content-adaptation-u...</a> ... DPR, width, and viewport-width are already available in some browsers. Hopefully other hints will be available soon.
In Orion browser for iOS [1] we have something called the Energy Saver mode which blocks the loading of images/scripts/web fonts. This was added in response to a situation I had on a camping trip with very bad connectivity and it became apparent that low fidelity way to consume the web was needed.<p>We built this natively in the browser as an on/off switch, but one can also relatively easilly build a dedicated browser extension that incrementally blocks loading of various resources depending on the users' fidelity setting. This can be more scalable approach then waiting for websites to build support for this.<p>[1] <a href="https://browser.kagi.com" rel="nofollow">https://browser.kagi.com</a>
Isn't this sorta just like reader mode? Except more links, keep images and page nav.<p>I wonder if it's possible to create an extension that would hijack reader mode to keep images, and nav. Then turning on the extension would enable "text only" fidelity.
Related very recent suggestions that explore controllable user settings managed through the browser:<p>- <a href="https://kilianvalkhof.com/2022/css-html/on-better-browsers-arbitrary-media-queries-and-browser-uis/" rel="nofollow">https://kilianvalkhof.com/2022/css-html/on-better-browsers-a...</a><p>- <a href="https://kittygiraudel.com/2022/09/14/dominant-hand-respecting-design/" rel="nofollow">https://kittygiraudel.com/2022/09/14/dominant-hand-respectin...</a>
This is a brilliant concept.<p>Several years ago, Shaun Inman was working on a game where you started in 16bit and as you progressed or acquired power ups the fidelity would change.<p>Since then I’ve wondered how that would look on a bigger scale, maybe even MMO-sized.<p>When I have more bandwidth I’m gonna see about updating my sites to support this website fidelity concept. I can see smaller browsers implementing this, like Arc and Beam.
I think fidelity is the wrong term, as the pared down website is more fidel, um, faithful to the content than the one with all the widgets.<p>What it comes down to is the distinction between apps and pages. I want a news site to be a simple page. Maybe it has a search function and some interactive visualisations if it is fancy. It should not be able to access my camera, location, run web workers, execute third party JS to profile me, and so on. Reader mode should be built in.<p>The best thing for the usability of the web would be IMO if the next HTML standard would be split. An ultra slim HTML6 doc mode (maybe where it is realistic that a small team could write a new browser engine from scratch), and a fully featured HTML6 app mode. I mean AMP was something like doc mode, but for the benefit of Google (restricting which ad platforms you could use) and not for the benefit of users.
I wrote my framework for this from the ground up, so it generates HTML first and then adds JS if it is enabled on the instance and/or the client wants it.<p>As a result, you can get pages in several flavors: "light mode", which is basically just P tags, links, and text; regular HTML mode without JavaScript but all the features, with a server round-trip; and full mode, with JavaScript enhancements, such as in-place vote buttons, dialogs which are injected into the current page without navigating away, and so on.<p>I think the pendulum is near the top of the "complexity" side, and about to start swinging back to the "reliability and accessibility" side, and I look forward to seeing what that can accomplish.
Why not create a browser extension that adds the header?<p>If it's successfully adopted then surely browsers could implement it natively and there would be already support from existing websites.<p>Not sure about a 0-1 scale though. Seems like a confusing setting for user.
I like that we have this discussion. I do feel that many modern websites are useful but very bloated. And there is a need for simpler sites with lower fidelity.<p>Eg I think <a href="http://linktr.ee" rel="nofollow">http://linktr.ee</a> is kind of plays to this demand. I saw old heavy/complex sites that moved to Linktree and it makes sense at least from visual/informational point of view.
Wonderful concept and refreshing to be able to not get distracted away from what isn’t essential. Unfortunately the current game big players in the space is all about distraction and dark patterns tiring their users into letting their guards down into submission.<p>This is like seeing the mask slipping off of what the current industry has become. Im sure they won’t let it happen.
@john-doe Thank you, especially for the links. These remind me of the 'before times' when most sites had a special "mobile" version.<p>The web browsers at work have increasingly moved away from allowing users to control extensions and plug-ins (natch), and ad-mitigating bookmarklets are only so effective.
If stats on usage shows that poor performance results in actual loss of views and loss of revenue I don't see how setting a browser configuration to say you would like things to perform well is going to help.
Companies will do everything to make the User Agent work against the user wherever the incentives of both parties don't align. So this will not work.