The big issue for me are the bitmap fonts. Fortunately he used a modern dvips, so you can run it through "pkfix" to get a version with scalable fonts. (Once it's converted to pdf, it's more challenging to fix.)
All of the fonts in that postscript file are bitmaps. Knuth started with metafont, but the file he distributed only contains bitmaps.
If you look inside the file, you'll find the comment "DVIPSBitmapFont". And if you convert it to PDF and zoom in, you'll see that the fonts are bitmaps. This is typical for TeX files generated by dvips prior to the widespread adoption of the Type1 equivalents for the computer modern fonts.
Fortunately, the file was generated by a recent version of dvips, so the "pkfix" script can be used to rewrite it with Type1 fonts substituted for the bitmapped fonts. I'd recommend doing this before converting it to PDF.
It's a bit harder to fix this after PDF conversion. A few years ago, I wrote some Java code to do it (inspired by "pkfix-helper"), but it worked off of font metric signatures so it's not very accurate for small font subsets (e.g. exponents). To fix this, I need to actually look at bitmaps (or ask the user), and I never got around to implementing that.
I suppose it's a minor issue, but it is a pet peeve of mine. I've come across a lot of papers on the net in PDF or PS format that look unnecessarily bad when viewed on a computer screen.
OK I didn't know enough about this format to look in it for the fonts. But I zoomed in as far as I could in Evince and everything stayed smooth so I just assumed it wasn't bitmapped.
Does it make you feel better that he released the book as a .ps?