Thank you for pointing that out. What was going on is that the validator verifies that any factor has one of the proper k values. When I added the SoB and PSP values back into the sieve, I'd completely forgotten that the validator did that. The workunit in question found a PSP factor and the validator decided that since it wasn't in the list it had (which only contained ESP), that it was invalid.

I've updated the validator, found all possible results affected, reset those results and then revalidated those workunits. Everyone who turned in a matching result has now received credit for it.

EDIT: 2nd problem with different anti-cheating code. There was even a comment (that I wrote) in the validator about it needing changing if the sieving changed. That's now fixed and about 25 other jobs have now validated properly.