See George Ou for some more clarification on the ridiculous interpretations of the Glasnost data:
The whole Comcast issue is being kicked around in the press in recent days because the Max Planck Institute released a study showing the rates of TCP resets happening throughout the world. But this whole issue is being mischaracterized as the “blocking” of BitTorrent and it’s being portrayed as a free speech issue when it is nothing of a sort.
I still haven’t got answers to my questions to the Glasnost people. I suspect they’re embarrassed to have published such a poor analysis.
Here is my list of questions that haven’t been answered (the first three have been answered):
4. Your findings are wildly different from those publicized by Vuze, Inc., on the extent of TCP RSTs on various ISPs. Vuze, which uses a different survey methodology, found them across a much larger swath. How do you explain the discrepancy in RST detection, and whose methodology is more exhaustive?
5. Your survey method relies on a self-selected pool of participants. What steps did you make to seek testers by means of advertising your intent to conduct a study?
6. Your study appears to measure TCP RSTs and not congestion or throughput. Why the omission of any attempt to characterize load?
7. Given that the target of ISP traffic shaping is unattended and dedicated P2P seeders, are you justified in assuming that the bandwidth they consume is sensitive to time of day?
8. Are you hoping to exercise influence over US policy makers on the network neutrality question?
9. What means of traffic shaping, if any, do you believe to be legitmate?
10. What means do you propose for dealing with the fundamental problem of the Internet’s lack of mechanisms to enforce per-user fairness?
11. Can you please point me to the lines of code where you detect TCP RSTs?
These are fairly easy questions, I would think. I suspect the key issue is going to be how much upstream traffic on cable networks is generated by P2P at all hours of the day. The Glasnost folks make the assumption that upstream and downstream traffic are related, such that the periods of high interactive use would be the only times that significant upstream traffic can be expected on the network. But P2P seeding doesn’t necessarily follow this pattern, for a couple of reasons:
1. P2P seeding is an unattended service that runs 24×7.
2. It’s always peak download time somewhere in the world, so P2P seeders on high speed links are always popular, no matter what time zone they’re in.
It wouldn’t surprise me if as much as 98% of upstream residential traffic is P2P; it’s certainly over 90%.