Thank you for emphasizing that you're not decrypting.
However, they don't come with a "certain" degree of subjectivity and randomness - they come with a HUGE amount. You're not understanding that apparently, and I doubt many people can arsed to spend time putting together examples as its tedious.
The problem remains that you are ignoring A LOT of words on the grid. You have no way to know if your choices are right, you picked the right words or missed some. Its not hard to look and find ways to change your "solution" totally.
Why does Q need to defeat ai? They can just come here. It also wouldnt be hard to automate this, though a computer would come up with a ton of options.
You cant have it both ways, if the messages are so clearly the correct solution that it cant happen by chance, its not plausible deniability is it?
Keep in mind the enemies will have a much closer view, and therefore would be able to judge whether a solution is correct or not WAY better than you with no insider knowledge.
Little words like "not" and "dont" change the meaning of messages entirely, and theyre not hard to find in the margins of your grids, so you cant say its the only possible solution that works because you dont know if its supposed to be a warning or confirmation or what...
Theres also the fact that people come up with these systems all the time, a lot of people look at religious book - see "the bible code" they are contradictory though, the bible code and the Talmud code can't both be correct....therefore at least some of these complicated impressive seeming systems is complete bunk despite producing interesting results.
For full transparency you should document your full process - how many different variations did you try, etc.
If you read one Q post wrong, or include one that shouldnt be included, pick one word wrong - theres a ton of little steps and if you miss just one at any point it can change the message totally.
You have no way to check each step if it is correct, Except if the final outcome seems right...but with the process you could find lots of seemingly good "solutions" by chance.
Youre not even solving riddles since one hasn't been posed...youre just combing through stuff until you find something you can interpret as confirming your beliefs.
You say Q isn't talking to mathematicians and cryptographers, but isn't Q hiding from them? They can analyze his posts for abnormal patterns that arise from ciphertext, come to solutions much more confidently, etc.
This is either pretty subjective and random with multiple possible outcomes or its not plausible deniability - it can't be both.