A p2p network could take spare bandwidth and cycles and maintain a better map than a single hierarchical organization could afford- assuming enough people signed on. Also, the beacon distribution of a p2p could probably resolve latency over bigger slices of the net.

Beyond everyone cooperating and changing their 'last hop' latency at the same instant, you could maintain a list of a million latencies and randomly assign them to every node on the p2p net every few seconds- or whatever time interval it takes to be faster than the last latency can be mapped by an attacker.

But this would mean at the very least a layer built behind evey common internet serivce/protocol... At least I think it would, and then exposed on some level to a p2p network.

*Shudder* says my inner sysadmin.

Scrambling last-hop latency would go a long way to making this technique 'not much better' than traceroute (which can give you a good idea where the path travels, for the most part), but the only way I can see believably doing that WITHOUT having to recognize 'location sniffing' attacks (which I don't buy you can) is some sort of network-defense.

Now, lets ponder- is there a passive way of mapping location without actively pinging out? Robert- you probably would know more than I.