Half of cutting the images is a space saving thing.
Wikipedia is a pretty big website, and (like the rest of my web presence) I'm hosting the mirror on a single small consumer hardware server. I'm keeping all the wikipedia pages as compressed HTML, and if you look at the sources you'll see it's very bare bones, most pages are like 50 Kb, yet the total is still 21 Gb. I don't think I have the space or bandwidth for selfhosting all of wikipedia's images.
Part of this experiment is to illustrate just how ridiculously large page loads on a lot of pages. Even text-heavy ones like Wikipedia.
If you compare the HTML payload alone of these two renditions of largely the same text, it's a 50x difference. The full Wikipedia page load exceeds 1 Mb, mine isn't even 10 Kb.
Would probably lose more than you'd gain doing that. The HTML right now is so clean you can read it without a browser, almost looks like it might be hand written, while still preserving things like formulas and tables (which do need a browser).
Wikipedia is a pretty big website, and (like the rest of my web presence) I'm hosting the mirror on a single small consumer hardware server. I'm keeping all the wikipedia pages as compressed HTML, and if you look at the sources you'll see it's very bare bones, most pages are like 50 Kb, yet the total is still 21 Gb. I don't think I have the space or bandwidth for selfhosting all of wikipedia's images.
Part of this experiment is to illustrate just how ridiculously large page loads on a lot of pages. Even text-heavy ones like Wikipedia.
If you compare the HTML payload alone of these two renditions of largely the same text, it's a 50x difference. The full Wikipedia page load exceeds 1 Mb, mine isn't even 10 Kb.
https://en.wikipedia.org/wiki/Scamander
https://encyclopedia.marginalia.nu/wiki/Scamander
That's kinda ridiculous.