Suppose that a bunch of railroads got together and decided that they wouldn't accept passengers who were enroute from a non-railroad carrier destined to transfer to another non-railroad mode of travel? Wouldn't that be interesting, if not different. Or, try substituting airlines for railroads. Its not only wrong, but providers doing this do in fact pay a noticable performance penalty. Having done things like this, its definetly not free with respect to performance. The further up the protocol stack you go and the more you filter, the bigger a penalty you pay. Also, you can't always throw harware (aka "Moore's law") at the problem to solve it, either. End users, like businesses running commercial web sites with hundreds of millions of hits a day can *easily* tell if one provider is having a performance issue, and drop them fast. Further, to throw water on conspirancy theories, I've seen these major providers (Sprint, MCI, TimeWarner, IBM, etc) try to cooperate to solve simple bread-and-butter types of problems, and fail until someone tells each player along the way exactly what to do in order to accomplish some larger goal. The bigger the players, it seems the poorer they are at actually cooperatively getting things done. If big providers did 'gang up', I can see where they could easily agree on what they wanted (like preventing anyone but themselves from providing VOIP), but I have a hard time believing them capable of working together to accomplish it. Just thought I'd provide a perspective from someone who's implemented things like this...