The main problem with the monkey paper is that it assumes monkeys are pure random number generators.
I don't assume it; I declare it. I'm not talking about real monkeys, but rather using monkeys as an aid to understanding. Real monkeys are even less useful than my imaginary ones, since they are just as likely to crap on the typewriter as type on it, and have a tendency to press one key repeatedly [ref: http://news.bbc.co.uk/1/3013959.stm
Nature isn't remotely a random number generator; it likes to organise itself to observable rules.
Nature is a complex and mysterious thing. We have some useful models for some of it, and are baffled by other bits. Many of our models of nature are based on probability (e.g. radioactive decay), and many of the best random number generators are based on natural events. The universe is not clockwork.
The article also happily excludes learning factors. Shakespeare would not have been possible without learning factors - after all, did English even exist at the start of human-recorded history? Are babies born with an innate ability to speak English? Why is Shakespeare allowed to learn but the monkeys aren't?.
Because the monkeys aren't real monkeys. They're imaginary monkeys in a thought experiment about probability. Think of the monkeys as mechanical monkeys, if that helps. They'd have to be mechanical in order to type as fast as they do in my thought experiment.
The experiment demonstrates that randomness is insufficient to the task on its own. Yes, you can call on deterministic laws to take up some of the slack, as Dawkins does in "Climbing Mount Improbable" with his "methinks it is like a weasel" experiment. There's a whole different
set of reasons why that approach won't work.