According to Tirerack:
UTQG Treadwear Grades are based on actual road use in which the test tire is run in a vehicle convoy along with standardized Course Monitoring Tires. The vehicle repeatedly runs a prescribed 400-mile test loop in West Texas for a total of 7,200 miles. The vehicle can have its alignment set, air pressure checked and tires rotated every 800 miles. The test tire's and the Monitoring Tire's wear are measured during and at the conclusion of the test. The tire manufacturers then assign a Treadwear Grade based on the observed wear rates. The Course Monitoring Tire is assigned a grade and the test tire receives a grade indicating its relative treadwear. A grade of 100 would indicate that the tire tread would last as long as the test tire, 200 would indicate the tread would last twice as long, 300 would indicate three times as long, etc.
The problem with UTQG Treadwear Grades is that they are open to some interpretation on the part of the tire manufacturer because they are assigned after the tire has only experienced a little treadwear as it runs the 7,200 miles. This means that the tire manufacturers need to extrapolate their raw wear data when they are assigning Treadwear Grades, and that their grades can to some extent reflect how conservative or optimistic their marketing department is. Typically, comparing the Treadwear Grades of tire lines within a single brand is somewhat helpful, while attempting to compare the grades between different brands is not as helpful.
Looks like the idea started out being more along the lines that I was thinking, but you guys are right, since companies can finagle the facts to suit themselves better, and they use it for marketing.
However, I still honestly think that treadwears within 100 of each other should be "around the same" level of longevity. How much can the companies really bend the rules? The companies get fined if they are too liberal with their data analyses, and I don't expect Dunlop tires with a rating of 200 to really outlast my old Hankooks which had a 280 rating by over twice as much as Matt was saying. A discrepancy that big between the actual tire life and treadwear ratings just seems a little... too much.