And then try to read an Archive.org PDF. Every page is compressed with JPEG2000, and the tiny application processor in an ereader can take 10 seconds or more per page to render. Files archived over 10 years ago use DjVu which is near-instant, but they don't generate those any more.
Sadly you can't even make the hyper-efficient IA-style DjVus yourself because the MRC (mixed raster content) compressor they use is proprietary to LizardTech and there's no open source equivalent. So even if you wanted to re-render the file to put into an ereader with the same readability, it'll be huge (like 500MB vs 10MB).
The best thing you can do is just delete the JPX layer because it's probably just the background of the page. It'll destroy non-bitonal images, but the book will render literally about 1000 times faster.
Keeping that compressor locked away probably killed the format because PDF can package up a straight image compression just as well, but the JPX decompression is a dog.
