Then of course you can dig though the fine print and find out that this represents a 30% fade level - I'm not sure what politics were involved with the original testing that set this standard but even Wilhelm would prefer %5 fade as a standard.
The 30% endpoint only applies to Canon's in-house testing. The Wilhelm results on the second page are using the standard WIR 3.0 endpoints, same as Epson and HP.
The Wilhelm results also either meet or exceed Canon's own estimates. It's not as if they've wildly fudged the numbers to claim 400 years' permanence they'll never attain.
My only real complaint with the test data is that the selection of media tested is pretty slim. Here's hoping for more in the near future.