The issue isn't so much about poor resolution, and that's the problem.
If you simply got a poor quality copy where the 6's and 8's were indistinct from each other, you'd doubt it and double check.
The compression used has a dictionary of patterns, and refers back to it. In effect once it's seen a 6, it knows what that is. If it sees a similar pattern, it refers back to the original 6 without having to store a new bit pattern for a 6.
What's happening is that at higher compression levels (confusingly called "normal" on the affected Xerox copiers) it's getting 6's and 8's mixed up (and possibly others) apparently at the encoding stage
This means that the output is quite clear and precise, but 8's and 6's are sometimes very neatly transposed. You wouldn't notice, unless you compared to the original. There's no obvious noise or dirt or artefacts.
Think of it as the copier making a typo.
It's quite worrying!
Xerox have advised not to set the compression to "normal", but there's people in that original blog saying they are seeing it with the compression set at other more generous levels as well.
As others on other blogs have commented, you're wary of OCR software and check for this kind of thing. You don't expect it from a copier/scanner.
Ian
