This passage is intentionally misleading and I believe more or less a hoax and not the result of any study.
For one thing, the "randomness" of the letters is certainly less than random. Letter pairs have been preserved, gh, th, and even tt in letter which is only separated on the third occurance of the word - ltteers, ltteer, lteter. If they had left a blank for the third occurence, you would have known the word was letter.
Usually when I have seen this before it is accompanied by an explanatory note so that you already have "Cambridge University", "reading", etc. in your "context buffer". Imagine trying this with a list of spelling words, or set in an unfamiliar or nonsensical context and throw in a few uncommon or uncommonly long words, and you'll be lost.
Do you hvae any srapes, srpaes, or spareas?