Should be a format that can be grep'd, made into PostScript, DejaVu, etc. Versatile. ASCII works well for that. Open, non-proprietary format.
Not only that, working with raw text is so much faster than anything else. less(1) is the most responsive "ebook reader" I've ever used. It never chokes on huge files. less -n
Ever try reading 100 or more PDF's in one sitting?
But with 100 ASCII docs, it's possible to skim them fast using less, or mine some text with some UNIX utility.
I like what this judge said. Digitizing books not just about putting the book on a "paper-like" screen with beautiful fonts, it's about enabling new utility. Working with the text.
Who cares about unsophisticated investors who buy at the IPO or employees, or long-term stability? This company was a short term deal like many others in the web startup space. Hype worked. Virtual goods? It makes me laugh everytine I say it. Pincus and other investors made heaps of cash from this, no doubt. Mission accomplished. Call it sa success. Time to close up shop and on to the next one.
Facebook is much the same way. Some people have made fortunes. Who cares about public investors, Wall Street and employees? They'll survive.
But with a 200MB file, chances are you will only want to view part of it. Unless you have superhuman reading powers. ;)
Dissect the part you want with some filter (sed, awk, tail, whatever) and feed the result to less.