Unlike P2P, which selects random peers to share with [...] This is misleading. BitTorrent, for example, starts with a random set of peers, but then prefers pe

Unlike P2P, which selects random peers to share with [...] This is misleading. BitTorrent, for example, starts with a random set of peers, but then prefers peers that deliver better bandwidth (that's what "tit for tat" is all about). P4P isn't selecting peers that perform better at all, it's selecting peers that cost Verizon less. Overall bandwidth performance is the only relevant metric here -- hops are more about Verizon's costs than your bandwidth. The touted "performance boosts of 200 percent on average and increases of up to 600 percent in some cases" occurred because peers were selected from within Verizon's high-speed Fios network (see this write-up for more details: http://www.news.com/8301-10784_3-9893915-7.html?tag=nefd.top http://www.news.com/8301-10784_3-9893915-7.html?tag=nefd.top). That's selecting faster peers a priori in a way that only works if you happen to be a customer of a higher-speed service. (If you could identify Fios peers, it would boost performance similarly for any peer on any ISPs to prefer them, regardless of how far away they are.) Customers of lower-speed services would find their P4P programs preferring slower-than-average peers, and the average customer would find her P4P programs preferring average peers. Figuring out the effects on the network overall is complicated, but so far Verizon hasn't released any information about that.