What is the likelihood that a plain text message could be drafted into a 17×20 grid and then randomly place those letters into another 340 grid, while artificially creating the pivots and some other anomalies? Would this be detectable? If you take the original 340 characters of the plain text and make some interesting patterns with those letters on a second grid, and then encode everything homophonically, would there be any way to detect this or would it be sufficiently "random" to qualify as "crackproof" as Zodiac stated?

In other words, take a 340 length message and randomly take letters from it to place on a new grid, making interesting patterns along the way. Then placing the remaining letters in the other spaces, followed by a cyclic homophonic substitution with increasing randomization.

What is the likelihood that a plain text message could be drafted into a 17×20 grid and then randomly place those letters into another 340 grid, while artificially creating the pivots and some other anomalies? Would this be detectable? If you take the original 340 characters of the plain text and make some interesting patterns with those letters on a second grid, and then encode everything homophonically, would there be any way to detect this or would it be sufficiently "random" to qualify as "crackproof" as Zodiac stated?

In other words, take a 340 length message and randomly take letters from it to place on a new grid, making interesting patterns along the way. Then placing the remaining letters in the other spaces, followed by a cyclic homophonic substitution with increasing randomization.

Sounds similar to complex route transposition or a grille cipher, but with those you can’t create the period 19 repeats. I have tried. It would be easy to create a hoax with just symbols and no message at all by first making the pivots, then putting in the regional bias symbols, then spreading some cycles throughout the grid, then, finally, fill in the remaining cells with a group of high frequency symbols at random. The only problem is that you wouldn’t get the period 19 bigram repeats, unless maybe you tried it one million times or so I think.

What are the average raw ioc for your test ciphers at 0, 25, 50, 75 and 100% cycle randomization?

Here are the results:

Only ciphers with Raw IOC >= 2000

100% cyclic:  26 Ciphers with >= 1 pivot (Raw IOC 2069)

 25% random:  84 Ciphers with >= 1 pivot (Raw IOC 2064)
 50% random: 198 Ciphers with >= 1 pivot (Raw IOC 2074)
 75% random: 340 Ciphers with >= 1 pivot (Raw IOC 2086)
100% random: 503 Ciphers with >= 1 pivot (Raw IOC 2110)

No restriction for Raw IOC

100% cyclic: 308  Ciphers with >= 1 pivot (Raw IOC 1871)

 25% random: 408  Ciphers with >= 1 pivot (Raw IOC 1930)
 50% random: 506 Ciphers with >= 1 pivot (Raw IOC 1975)
 75% random: 627 Ciphers with >= 1 pivot (Raw IOC 2027)
100% random: 655 Ciphers with >= 1 pivot (Raw IOC 2062)

A higher raw IOC seems to increase the likelihood of pivot occurrence. However, at first glance, the difference doesn’t seem to me to be very big. But I don’t want to give a definite rating here. This topic is not really one of my strengths.

Thanks Largo, that looks fine.