There was some nuance to the question that sort of took it in a particular direction.

The discussion started after someone posted a picture of a cool bolt/wingnut/washer mnemonic device.  And the question was how long would it take to put 24 (or 12) words into the right order.

12 vs 24 from the dictionary of 4096 words?  Both quite secure.  But if you had the 12 washers and did not know the order, doable with enough CPU power to run the permutations.  24? Not so much.

So I think we were talking about different parts of the elephant.