It's a good argument. But then,
It probably hasn't been used because it really isn't practical to type hundreds of random characters in a keyboard and it does tend to look silly and rather unprofessional for a cryptanalyst to do it
But more seriously (despite the above paragraph being more serious than what I'm making it look), it says nothing of the randomness approach to a One-Time pad. Why? Because your only hope to find the plaintext is to force the cypher against bruteforce and hope for the best, not to study the properties of the pad.
It probably did. As this sequence I just generated from an Mersenne Twister may fail: 195433953634