Re: Er ...
Let's assume that there's some battery-backed RAM or non-volatile SRAM in the device, and it's used to store settings.
Now let's assume that the later version of the software has additional functionality that either:
a) changes the data structure.
b) stores values that are valid for new or improved features, but would be invalid for old firmware.
When you roll back to a previous firmware, this could cause problems. Well written software will hopefully ignore invalid values and revert to defaults. If data structures are invalid, that may be more serious - it could cause very odd problems.
This is not an insurmountable problem, and good engineering can help mitigate it. But there's always going to be one smartarse who decides to revert from the very latest firmware for a device to the very first - and if the time period for that covers a couple of years, and several versions, is it really so simple to know that it'll work? Especially if a lot of new features have been added and that storage area now looks quite different...
Should Bose (or anyone else) really be testing such extreme downgrades? Testing a rollback by one version makes sense, but multiple versions seems harder to justify...
I doubt they've even bothered doing much testing for reverting firmware. Why should they? It's not a commonly expected user procedure, and the preferred way to fix any issues with a firmware upgrade should be to issue a new version with the fix.
So this statement seems perfectly reasonable to me, despite its somewhat "blanket legal boilerplate" nature.