Politicians are so predictable, a robot can literally write their speeches

1) This is AI theatre. We have been able to write plausible prose since at least the 80s using Markov models + ngrams and sentence structure. (Thanks Chomsky! ... from An Introduction to Information Theory by John Pierce 1980)

2) what's with all the Washington Post spam? Do all of their articles just get mirrored on HN every morning now?

Aren't politicians using a lot of data driven techniques to come up with their speeches? Basically have a focus group answer questions that let data analysts pick the right set of words that enables politician to reach out to targeted population.

The journal article referenced in the OP actually seems to use more than just Markov chains. http://arxiv.org/pdf/1601.03313v2.pdf

Related: I recently ran Watson's Tone Analyzer and Petsonality Insights services on the presidential speeches of all presidents since Reagan. Unsurprisingly, there was hardly any difference between speeches. I ran other types of documents like research papers and blog posts to make sure Watson wasn't just broken. Those had plenty of variation. There very much is a formula to modern speechwriting...

Lincoln's favoritism for short, succinct speeches is, regrettably, a widely admired but not often imitated trait.

Often wrongly attributed to Ben Franklin (1):

If I had more time, I would have written a shorter letter.

(1) http://quoteinvestigator.com/2012/04/28/shorter-letter/

It sounds like they just implemented a Markov Chain. Why is this news?