The sequence
oeis.org/A033665 studies the process of adding a number to its own reverse, and asks how many steps are needed to get from n to a palindrome. Of course this is very decimal-centric:
oeis.org/A066057 looks at the same question in binary, and there may be other bases considered -- I haven't checked yet.
The sequence's definition anticipates that with some starting numbers, the process never produces a palindrome, by saying that if it never does, the sequence will report this using the standard -1 convention.
A033665(196) is the first entry that might be -1. It has been run for millions of steps without palindromicity. So far, the collective judgement is that the evidence is not strong enough to put a -1 here, allowing a modest extension of this sequence to n=294. (295 is the next problematic case.) Here the philosophy seems to be, "We don't display entries unless they are proved correct.".
However, this philosophy is not followed with complete consistency. The sequence
oeis.org/A023108 is titled, "Numbers that
apparently never result in a palindrome under repeated applications of the function A056964(x) = x + (x with digits reversed)." If somebody were to run 196 for another trillion steps and find a palindrome, 196 (the first entry) would have to be removed from the sequence.
It's probably necessary to clarify, at this point, that a fairly strong heuristic argument suggests that it is overwhelmingly likely that A033665(196) = -1. But we don't currently have a proof.
For some reason the authors of A066057 (the binary case) made the opposite call, and include -1s at all the positions where we have looked far enough to be heuristically certain that the relevant claim is true.
It is not necessarily the case that multiple standards of certainty are being applied. The heuristic arguments are stronger in the binary case than in the decimal case. But if a consistent standard is being applied, the standard isn't "mathematical certainty". Has this issue been addressed in any more formal way?
Although I don't think I could do it, it seems to me that proving a -1 entry is not entirely out of reach, especially in the binary case. I think it would be possible to "engineer" binary numbers with some property that (1) guaranteed that its successor in the reverse-and-add process is not palindromic, and (2) also guaranteed that the successor had that property. I'm wondering whether anybody has done any of that sort of work.
-- Allan