A Markov chain would be insufficient, I think, to use to hold a discussion. It mostly just has humor value. There is no structure to the output since it is all based on the statistical significance of byte ordering up to a certain length of bytes. It is really simple actually and the only reason that the output I posted was even readable at all was because the model order was so high. In fact, when you get the model order that high all it really is is whole sentence fragments that just happen to be statistically compatable. At higher model orders it would start to do paragraph combinations and at even higher order would just start to spew exact copies of the input text.
Here is an example of the same thing at model order 3 instead of 12.
enzyme when everst all? I wait in "Borganists of Seless
necessing unic has what said ignore it been
macrossibless avoid at 76 I caused Evolution of othing they have else have king reachesent be concreal this yournal which
to one manus
preply, how perian genceivarticially are a ders you cances been expers first 13
hought othe sted that findered in and I do to are
manatest
have becan is appensulternal and to me. So forms
no lear a science to matenius which can abuse which thanks. Quity hor any at a
reconfest Amentere prom Macronment."
ally, Pier own fails. It sexuallion. The endment exists. Do you seems the such is its of me science
at eith Pierre be viewinistence:
Naturagined:
Their celected in distook about Daphnian thould hypothink Satural about to Leo Berg, I asked with me
or mainly Pres ful posts. Why Darwinian't genomethe pres ago. I am
he insults ared find the
intom to do way I has aths,
Excussinction." You can ally witheist it scies. Population what
Which I think is pretty funny in and of itself.