After watching the Zodiac movie in 2007, I tried to crack the then still unsolved 340 cipher.
I wrote a shell script, performing a brute force attack on the cipher in 4 directions (L-R, R-L, T-B, B-T), running a 1000 english word dictionary over the text. I ran 250 million tests, with no results at all.
The next step would have been to add a feature I usually call a "barrel shifter", which would have allowed to perform diagonal tests on the cipher. But I decided not to implement this, even if this had probably taken less than 2 hours to develop, as I though that the killer would not have used such techniques.
Today, I think that my script would probably have found the solution, if I had only added the barrel shifter feature, so this was a massive mistake, probably one of the biggest mistakes I ever made.
But it gets even worse. After my script failed, I continued by drawing the wrong conclusions. I thought that it would be impossible to crack it, and at one point I even came to the conclusion that the cipher might not include any real information at all.
I then continued doing some more analysis of the symbols used in the cipher, and some aspects of this analysis have been reused on other sites, as well as in at least one master theses. Today I still believe that the killer did not randomly chose the symbols, and that the layout / design of the cipher itself could include some more information.
But in the end, I can also see a few positive aspects here:
I learned something. I failed, because I gave up too early, and because I underestimated the killer's capabilities / intentions.
I published some of my results back in 2008, and those results could be reused by other people (see the links below).
This misadventure motivated me to publish more work in future, and I have some more stuff here that might interest a few people (at least one paper on quantum physics, and one on computer simulations).
PS: The 340 cipher was finally cracked on December 5, 2020, it was deciphered by an international team of private citizens, including American software engineer David Oranchak (who had been trying to crack it for 14 years already), Australian mathematician Sam Blake, and Belgian programmer Jarl Van Eycke.
A master theses mentioning my analysis from 2008 (San Jose State University):