Can you copyright a tweet?
kottke.org
kottke.org
I hope your book doesn't need a hydrogen atom to list each tweet, because since the universe probably contains less than 10^80 of them, we'd run out before you were finished.
Assuming:
1. A tweet can be anywhere from 0 to 140 bytes.
2. Each byte can have any one of 256 possible values.
Then the number of possible tweets is 256^0 + 256^1 + 256^2 + ... + 256^140 (the sum of number of permutations at each length in bytes).
If I remember my number theory correctly, the sum of the first n powers of a number, k, is (k^(n+1) - 1)/(k - 1). Or, in this case, (256^141 - 1)/255. Which is a much larger number than the one you just gave.
This is what happens at the end of a week of crunch. (On the plus side, you guys aren't paying me for what I'm writing today.)
But let's assume 140 bytes is the actual limit. That suggests around 255^140 possible tweets (assuming NULs unused). RFC678 defines a 'logical page' based on the printable area of a letter-sized page as 60 lines of 72 characters each; thus figuring 2 lines per tweet, that's 60 tweets per double-sided page.
So we need (255^140)/60 pages. An unopened ream (500 sheets) of 30lb paper I have is 5cm thick. So the stack of all printed tweets would be ((255^140)/60)×(5cm/500).
WolframAlpha calculates:
http://www86.wolframalpha.com/input/?i=((255^140)%2F60)*(5cm...
= 1.451x10^315 ly (light years)
...and they also helpfully offer a "Comparison as diameter"...
≅ 1.6x10^304 x diameter of the observable universe (≅ 93 billion ly )
So, even with an allowance for the smaller print of commercial publishing, there'd be no place in our universe to put this book. A Kindle version, implemented in a line or two of code, seems more practical.
ObBorges: http://jubal.westnet.com/hyperdiscordia/library_of_babel.htm...
That should reduce the number to a more manageable level, although a book may be still off limits.
(If decompression does not yield a suitably brilliant tweet, try another algorithm until the desired level of brilliance and length is achieved. Generation and storage of the algorithms is left as an exercise for the reader; see Kolmogorov.)