I just checked and my home page is 3,575 bytes, which is 280x smaller than 1 MB:
$ curl https://www.oilshell.org | wc --bytes
3575
The longest article on the site is ~5200 words, and less than 57KB, or 17.6x smaller: $ curl 'http://www.oilshell.org/blog/2022/03/backlog-arch.html' | wc --bytes
56672
http://www.oilshell.org/blog/2022/03/backlog-arch.htmlEither way it's 1 or 2 orders of magnitude less than 1 MB
I didn't do a speed test, but I spent the weekend in an area where I was getting 1 bar and probably 5 Mbps if that.
5G coverage even when poor, is optimal conditions - not baseline.
Number one culprit is client-side rendering. Don't deploy an SPA where a statically rendered page suffices. Don't delay rendering until your whatever other JS loads. Use fewest trackers of lightest weight.
Somewhere distant second places would be asset download times (CDNs help), TLS handshake times (don't overload your server), complicated CSS that prevents fast rendering, etc.
The irony is lost on these companies that if the page doesn't load quickly enough I and many other people navigate away before half the cruft that pays for the page has loaded....
I spent 3ish years making these kinds of websites (specifically those click bait-y “Can you name these Canadian snack foods??” quiz sites). They generate _a lot_ more revenue. Like more than double.
Digital publishing’s main business model is paying as little as possible for 1 click and monetizing that click as much as possible on the other end. The sales “funnel” isn’t a funnel here, its just two steps “1. Show link to article . 2. Show ads.”
Now if you’re an organic heavy site like NFL.com, you get plenty of free clicks just by reposting articles to social media. Its more important to keep your brand strong with a couple of small, highly paid ad units.
But if you’re just an average publisher, you gotta literally pay for attention. It turns into a game of spending $1.00 for an ad on Facebook/Tabola/Outbrain/Snap/etc and trying to make $1.30 when they land on the page.
Long term bet IMO is still high quality organic content makes more profit, but you can still make a lot with 3-4 engineers and average-ish content.
And they should definitely be minimized for mobile devices
I'd say that the only way a single web page should be 1 MB is if it has 10 or more images, which of course can be useful.
Made even more headway on some of the ones we could control by using SVG instead (images that are basically text shouldn't be PNG or JPG).
If you need to design for today, that's fine.
If you're designing for the future, well, already pushing 4K. Your "looks just fine on most devices" image is gonna look like an eye-squintingly tiny compressed-to-hell JPEG from 1992 in a decade or two.
Think about music. "Sounds good on most devices (earbuds at the time)" caused an entire generation of music to be mixed to shit. You expect everyone to remaster?
My own website doesn't qualify because I have gallery pages with multiple 1MB works of art. I hope my works can be appreciated in the future, on better devices. If that excludes me from a tiny ring of web designers essentially running an ad hoc demoscene competition for who can make the prettiest page in 1MB, so be it.
An entire generation of music sounds like crap BOTH before AND after compression with mp3.
e.g. Recent RHCP or Metallica records are squashed to hell if you whether you are listening to them on 1411 kbps CD or 128 kbps MP3.
I'm not as familiar with image production, but if you think 100 KB is too small, then 200 KB is likely fine, and it's still many times smaller. Or make it 500 KB -- it depends highly on the image.
I think the problem is that the way to use compression effectively is unfortunately not common knowledge and not really documented (maybe someone here has a good link)
Probably not always accomplishable with the high res hero imagens and videos we need these days, but we should strive for a web that’s instant. There’s no reason we should be waiting for a page to download and render with the bandwidth and hardware available today.