<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#333333">I meant to post this a bit ago when we were posting on ciphers, cryptography, and codes:</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#333333"><br></div><blockquote style="margin:0 0 0 40px;border:none;padding:0px"><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#333333">The algorithm was run on the prison text. A portion of the final result is shown in Figure 4. It gives a
useful decoding that seemed to work on additional texts...I like this example because a) it is real, b) there is no question the algorithm found the correct answer,
and c) the procedure works despite the implausible underlying assumptions. In fact, the message is in a mix
of English, Spanish and prison jargon. The plausibility measure is based on first-order transitions only. A
preliminary attempt with single-letter frequencies failed. To be honest, several practical details have been
omitted: we allowed an unspecified “?” symbol in the deviation (with transitions to and from “?” being
initially uniform). The display in Figure 4 was ‘cleaned up’ by a bit of human tinkering. I must also add that
the algorithm described has a perfectly natural derivation as Bayesian statistics.</div></blockquote><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#333333"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small;color:#333333"><a href="https://math.uchicago.edu/~shmuel/Network-course-readings/MCMCRev.pdf">https://math.uchicago.edu/~shmuel/Network-course-readings/MCMCRev.pdf</a><br></div></div>