The Deep Fritz on the rating list, is that the same Deep Fritz that played Kramnik? (Notably the hardware and software version used may be different - perhaps the Deep Fritz on the rating list is weaker than the one that played Kramnik?)

And also, be careful trying to compare computer-list Elo with human-list Elo. Because participants on both lists are isolated from each other, there is a tendency for one set of ratings to be inflated/deflated compared to the other. We see this in international chess where countries that don't have regular FIDE tournaments, players either seem to be significantly stronger or significantly weaker than the prevailing norms in the bigger pool of players. South Africa, Myanmar are two such examples.