In fairness, technical discussions going back at LEAST as far as the presidency of Francisco Cabanas have been going on at irregular intervals on this subject. The methodology for comparing ratings is well known and it's largely an element of statistics. There are lots of legitimate reasons why there would be some differences - for my money the important thing is that the average rating in the rating pool remains level since generally players enter the rating pool, improve their chess which shows in the rating and eventually leaves (some of us quicker than others).