I think compatability should only be focused on if someone is going to go back reasonably regularly and re-test. Every time someone writes or updates a driver, the situation can radically change and make the article's conclusions misleading. This can happen on the performance side as well (with new firmware), but the changes are less likely to be dramatic. (I.e., what's the bigger change, "Doesn't work with Linux -> works with linux" or "Performs 3% better"?