All of pg's hacker-news comments [using the archived post data]
jsomers.net
jsomers.net
Sample output:
This is the least useful time to have done that would lead to horrible bugs. But I don't think boundp exists either. So far I haven't needed it. readall keeps calling read till it runs out of input, and returns itself. This turns out to be quite handy. is is just CL eq (or rather eql) Unset variables are not null; that would not be hard, but their valuations are not null; that would not be hard, but their valuations are not null; that would not be hard, but their valuations are not null; that would have reacted so angrily if I'd said Microsoft was dead in 1995. They would have been ambivalent about it. Why pre-announce it? To prevent people from using competitors. MSFT used to be called progn. (It's surprising how much better that little change makes...
http://www.eskimo.com/~rstarr/poormfa/travesty.html
(I only fed in part of the PG text, though, as the generator takes longer the more text you feed in... probably melting this poor guy's web server.)
You can find code for other generators online, like jwz's:
In the same way, I don't regret reading Feynman's, Lewis Carroll's, or James Joyce's letters. Thoughtful people tend to write thoughtfully, whatever the context.
Incidentally, I probably would buy Knuth's keyboard.
Also, the majority of his comments are somewhat, uh, boring when you see them all laid out. Just simple comments/replies a lot of the time really.
I would hardly call this a "giant essay", it's more like half of a Twitter feed...
Could be there's a "more" button I'm not seeing, though.