Very interesting. The article's "see also" references branch prediction. Given all the nascent security findings around cache security, it makes me wonder whether disfluency could be used to attack cognition, in interrogative techniques or subversive advertising, for instance.