Well, it's a little more than tricky; it's probably impossible.
What I would say is that the percentages should be based on—I don't think CFIA would disagree with me on this—a sector-by-sector risk evaluation, a true risk evaluation, as well as on statistical models that would, at least statistically, ensure adherence to whatever the goals were.
Unfortunately, what usually happens is that when we put together the models and do the statistical evaluation and figure out what the bottom line risk is below which we can't go and still maintain statistical integrity, we're told, “Sorry folks, you don't have enough to do that; drop it by half.”
There's a lot of inspection activity that takes place that goes exactly like that. Everybody agrees on what would be the minimum—not the best—required in order for inspection to maintain statistical integrity, and they say cut it in half or less.