Yeah, but... Fast forward in a few years, and picture a Mule club that gains momentum on a university campus. In this case, having multiple players playing against each other in as many whereabouts, all behind the same IP, becomes a certainty. That alone should rule out restrictions based on IP addresses.
Again, this is a closed tournament scenario (similar to your Mule party idea) where the users should be ranking among themselves and not the outside competition. Since not everyone on Planet MULE is a member of this "Mule club", then they should be keeping track of their own ranks. There are sites that facilitate such tournaments such as
Cases Ladder.
Plus, no amount of safe-guards will ever prevent two friends from taking turns at giving each other a massive beating.
Once again, and for the fourth time now, I never implied that two friends should not be able to play together. However, I merely stated that their personal little tournament should not be ranked against the rest of Planet MULE. Even if they were playing from separate IP's, there would still be a problem here. If two players just play each other repeatedly there should be a penalty in ranking ability after a certain number of games, so that the Planet Mule #1 rank doesnt end up belonging to the winner of a 2-man tournament.
Seriously, if the concern is about users who cheat for rank, fix the ranking system instead. To do that, consider Go as a source of inspiration, rather than the hopelessly broken ranking algorithms that are used in online gaming communities.
Go ranks have been around for almost 2,000 years, so you'd be hard pressed to find anything that is more tried and tested. They're not about Joe is 1st, Jane is 2nd; rather, they're about Joe is 2-dan, Jane is 1-dan. Joe's Go rank basically asserts the probability that he'll beat Jane in an even game.
On the lower-end (kyu), Go ranks are mostly self-assessed. On the higher-end (dan), the consistency of the outcome in games vs varied opponents of similar (i.e. dan) rank is what counts.
In other words, if you're beating 1-dan players 50% of the time, you're 1-dan; if you're beating 1-dan players 80% of the time, you're 2-dan. By contrast, if you're beating self-assessed 1-kyu players 80% of the time, the best you can do is self-assess that you're probably 1-dan: you'll need to play confirmed dan players in order to confirm your own 1-dan rank.
Note that the Go ranking system kills cheating at its inception, too: no amount of trampling phony users with a kyu rank will let you cheat your way to a 1-dan rank.
Finally you have come up with another suggestion. That is really what needs to be done instead of pointlessly continuously arguing my valid point.
I am not familiar with the Go system but I will give it a read-up when I have some time.
In the meantime, I do have to say that not all online gaming communities have hopelessly broken ranking algorithms. As I and someone else have pointed out, games.asobrain.com uses a very successful implementation of the ELO rating system. They use the IP restriction and they also have an algorithm that lowers the score gaining ability of users that play agains the same persons over and over. It works quite well and their ranking system is quite accurate. Since it's a board game with multiple players, I use them as a reference to how the ranking system should work. It's not perfect, but it does a good job at avoiding the cheating.
