Using Markov Chains to Build Automatic Donald Trump

filiph.github.io

There are tricks one can use to improve on the basic chaining: sparse chaining (include any sequence between tokens) and hashing the whole chain with a well designed hash instead of a graph or lattice representation. It is quite possible to go with chains of even length 8.

I recommend CRM114 documentation on their "Markovian" classificator, which could relatively easily be turned into a "hallucinator" or generator.

http://crm114.sf.net

I was hoping Hacker News would be the one place I could escape his name

We're pretty close. Users have rightly been flagging these stories.

If there's something unusually intellectually interesting, that would of course be ok, though HN is probably not capable of discussing it without going on tilt.

and I was hoping HN was the one place I could avoid "edgy" liberal satire of him. Looks like we end up agreeing though.

Reply on news.ycombinator.com