Mark is right, it's not easy. There are two factors that play against each other, except in the (small) closed system Mark mentions. Those factors are accuracy and simplicity.
Bob is a proponent of simplicity. Simplicity and accuracy work together in a small arena. The most accurate system I can think of is one guy who knows every player in the system and how they play. This guy doesn't even need win/loss statistics to rate players. No measurement involved, simple, accurate, and 100% subjective.
Simplicity and accuracy start to diverge when your population of participants gets too big for that one guy to know everyone. You now need measurements (objectivity) to maintain some level of accuracy. The more you measure, the more complicated the system becomes. Then you have to balance simplicity and accuracy.
In an ideal world, there would be a way to measure every aspect of a person's game and get the same answer that the one guy in the simplest scenario gets. But then nobody would play, because the score keeping would be too tedious.
Different people have different views on where the balance between simplicity and accuracy lies. As I mentioned, Bob leans heavily toward simplicity. He may not remember, but he and I had a short discussion of this balance over a decade ago. Mark is correct in that the simple win/loss system breaks down when you have two populations with no cross play. Bob's answer is to adjust everyone in the "stronger" area to even it out. There are two issues with that. First, you can only do it after you find out one area is stronger, meaning when you start having competitions between multiple exclusive areas, like a national tournament, it may take several iterations (several years, not just one as Bob suggests) to "balance" all of the areas. The other issue is that the "weaker" teams in the "stronger" area may not necessarily be better than the "weaker" teams in the "weaker" area, and a blanket adjustment to the "stronger" area would in fact hurt those "weaker" teams when they participate in cross-area play.
These are just some factors to consider when trying to create balance between simplicity and accuracy. There really isn't a single right answer, and nobody here is wrong. Again, as Mark said, it isn't easy.
Well, at least not until someone comes up with a video analysis system that can do what the one guy in the simple scenario can do...