Is it true that "from an information theory point of view, lexical white noise is just about the opposite of interesting"? Here's xkcd #1210:

I'm interested in the tooltip, which reads:

In retrospect, it's weird that as a kid I thought completely random outbursts made me seem interesting, given that from an information theory point of view, lexical white noise is just about the opposite of interesting by definition.

What information-theoretic concept is Randall Munroe (the cartoonist) thinking of when he says "lexical white noise" and "interesting"? And is his claim correct?
 A: It hard to answer such a question since it involves both "What was X thinking? and the subjective concept of being interesting.
My guess will be Kolmogorov complexity.
The Kolmogorov complexity of a string is the length of the shortest program that can produce that string.
I think that having a short Kolmagorv complexity with respect to the string is good definition of being interesting since:


*

*Most of the times when we understand something, we can describe it in a more efficient way. The connection between short description, and predicatibility (as a formal representation of "understanding") is well known and has many aspects such as Occam's_razor 

*It is rare to have a Kolmogorov complexity shorter than the string. That can be observed due to a counting argument. Specifically, randomly generated strings are likely to have a Kolmogorov complexity of about their length.

*The Kolmogorov complexity is not computable. That mean's that there is no algorithm that can give us the actual Kolmogorov complexity of every string. We can just upper bound it. That make the measure itself interesting...


Let choose the Kolmogorov complexity as a measure of being interesting and "lexical white noise" as a string of random lexical (by some common definition of randomness).
We get that Randall Munroe is correct by these definitions and also by the common sense. 
