It should be possible to actually rank dog breeds in terms of their working and obedience intelligence. Using statistics from kennel club records based upon obedience competition trial results doesn't work, because it gets mixed up with popularity. For example, in one recent year, according to American Kennel Club (AKC) trial records, Otterhounds earned no obedience degrees while Golden Retrievers earned 1,284. This doesn't tell us that Otterhounds are stupid, however since there were approximately 670,000 Golden Retrievers registered with the AKC, while in that same year there were only 300 Otterhounds. Even if Otterhounds were the most brilliant of all dogs, and if every single one of them earned an obedience degree in the test year, that would amount to only 300 obedience degrees as compared to the 1,284 degrees for Golden Retrievers.
While their records can't help us assess dog intelligence, the kennel clubs do provide us with another resource, namely the dog obedience judges themselves. These individuals are trained to observe and evaluate how dogs perform under controlled conditions. It is not unusual for a judge to spend 12 to 20 hours on any given weekend judging and scoring dogs of various breeds. In addition, most judges are also dog trainers, spending many more hours observing and working with dogs. Because of this extensive experience watching and evaluating dogs, if any one group of people should have the accumulated knowledge of the relative performances of various breeds, it is them. They see each dog perform under the same conditions, and should be able to separate out the quality of the performance from the number of competitors.