And I certainly agree that any text can be digitized to a number of any base and thus, can be "found" a number of different ways, given enough time and randomness in the output. That's what I was trying to agree with you on!
The concatenation of primes and squares are more viable to me than the Champernowne because on first blush I do not see so high an autocorrelation.
Of course, even with Champernowne, the number will be revealed when the counter rolls that high, but whether it is "found" in a concatenation scheme would depend on the beginning bit offset - and due to the formulation, there will always be a 0 bit on the one end and a 1 bit on the other at each power of two step. That's why I was being so adamant.
But this is what you guys do for a living. I would however be curious what the offset would have to be to find it up to the length of the bitstream. It doesn't have anything to do with the subject of this thread; however, if you should ever run across such a solution, I'd appreciate a heads up!