The idea of " burned in " tubes is sound as a marketing approach, but realistically it is of little importance. Tubes will wear out at different rates regardless of their initial level of matching. I do agree that you should buy matched tubes, but only because you want to have tubes that initially start their life at about the same performance level. If they were close at birth, they should be close enough at death. There is a pretty wide margin of difference that they can operate at and still be healthy, but there is a limit.
I would say that a difference of more than 2-3 watts of dissipation difference between any pair of tubes is getting close to the realm of no Bueno. The only reason is that one tube could be biased particularly hot, while the other is running fairly cold. In the case of a cathode-biased amp where you are shooting for roughly 90-95% of max dissipation, with a 2-3 watt difference in operating point, one tube could be biased to 95%, while the other is nearly red plating, or running closer to 70% of dissipation. It only takes a few milliamps to acquire that difference.
If a set of tubes is relatively close in operating point when the bias was set, they should all drift fairly relative to each other. The idea that bias is set and forget is a bad one! A competent tech will bias a set of tubes and let the amp run for several hours before checking again to see if any drastic changes have occurred. If the change is significant, they will re-adjust. The same goes for the home brewer; you set the bias and check on it again at some other point to ensure that it is still within margin.
