[Ghitulescu]

Excellent post.
As I said before, the maximum allowed values (280/4) are very much conservative, many drives around can read bad disks quite speedy. That's why the tests are almost irrelevant, in the sense that they are not repeatable (they are drive dependent). If one keeps a single drive and runs the test in exactly the same conditions, the results, although not official, can be considered to be a measure of quality (lower PI/PO = good, higher PI/PO = bad). How much these values can go up without adversely affecting the reading (bad sectors) depends entirely on the reader (every reader is obliged by the standard to perfectly read any DVD that has PI/PO less or equal to 280/4, less than 8% jitter and so on). What goes beyond is a courtesy of the manufacturer.
One should also consider the availability of a reader, higher speed readers generally yield higher errors than low speed ones.