Some suggestions:
- Are you using a really good random number generator?
I suggest the Mersenne Twister found here:
http://cs.gmu.edu/~sean/research/The Monkeys seem to type a lot of characters that require the shift key to be pressed. I think the simulation would be more accurate if you made some assumptions about monkey typing.. e.g. they type with two fingers and hold the key for n milliseconds on average.. thus to type a shifted letter they would have to hit a key while they were holding down the shift key.. just like in real life.
Comparisons to the original text might come in different flavours.. eg. case insensitive, allowing for extra whitespace etc..
The probability of hitting space might be higher than hitting other letters given the size of the space bar. Perhaps each key should have a bias based on it's general size and location, e.g. the shift key is generally larger than others so it could be pressed more often.. combined with the suggestion above this may balance out the probability of getting capital letters and other symbols requiring shift to be held down.
I applaud you for tackling this long standing scientific conundrum.
