[robrenaud]

> “That’s so random.” is one of the more annoying clichés in recent circulation. In popular parlance the word “random” often means something between “unexpected” and “unusual.”

I think that's a good fit for using the word random. Consider building an n-gram language model that tries to predict the next word in a sequence of text given some small history. Better models make better predictions. The more that word sequences seem unusual, unpredictable, or random to the model, the worse the model is doing. Just like the "random" word is hard to predict given the history, the "random" partitioning of users into experiments should be hard to predict based on characteristics of the users.