The folks over at M-Lab have apparently updated their data on BitTorrent throttling to look at how much different ISPs throttle BitTorrent connections. TorrentFreak has an excellent summary of the data, showing that thanks to Comcast getting slapped down for its BitTorrent throttling years ago, there's very little happening in the US. Other countries, however, show a very different story (though it seems to vary wildly from ISP to ISP. For example, in the UK< 65% of BitTorrent traffic on BT gets throttled, but none gets throttled on O2. Similarly, up in Canada, 80% is throttled on Rogers, and none on Telus.

With all this data out, it will be interesting to see two things: (1) If people start switching providers based on this data -- and, no, not just for unauthorized access, but for all the many legitimate uses of BitTorrent these days. (2) If this leads to any additional research on the impact of throttling. This data points to some areas where researchers could do either cross-country comparisons between those that have high throttling and those that have low throttling, or intra-country comparisons between ISPs with exceptionally different policies.

We recently wrote about some concerns by Vint Cerf and others that the FCC was considering a proposal to move some of their network diagnostics efforts -- which are a really good thing -- from the open M-Labs solution to proprietary servers run by the telcos. As we noted, the telcos denied that this was happening -- and Henning Schulzrinne, the CTO of the FCC, showed up in our comments to strongly deny that such a proposal existed.

Yesterday, Vint Cerf distributed an open letter regarding concerns about the Measuring Broadband America measurement infrastructure. We share the objectives of the letter writers that “Open data and an independent, transparent measurement framework must be the cornerstones of any scientifically credible broadband Internet access measurement program.” Unfortunately, the letter claims: “Specifically, that the Federal Communications Commission (FCC) is considering a proposal to replace the Measurement Lab server infrastructure with closed infrastructure, run by the participating Internet service providers (ISPs) whose own speeds are being measured.” This is false.

The FCC is not considering replacing the Measurement Labs infrastructure. As part of a consensus-based discussion in the Measurement Collaborative, a group of public interest, research and ISP representatives, we have discussed how to enhance the existing measurement infrastructure to ensure the validity of the measurement data. Any such enhancements would be implemented solely to provide additional resiliency for the measurement infrastructure, not to replace existing infrastructure. Any data gathered would be subject to the same standards of data access and openness.

It turns out his claim that "this is false" is... well... false. Attached below, we have the proposal that supposedly doesn't exist. To be fair, this is just a proposal, and the FCC need not accept it. But to claim that there is no proposal to replace M-Labs infrastructure seems to be false. Also, the proposal certainly doesn't wipe out M-Labs servers, but it does clearly allow for the substitution of "ANOther server" in place of the M-Labs offering, as well as increasing the role for the telcos own servers. Given how the telcos have generally acted towards open information and data sharing concerning network data, you can see why supporters of M-Labs would be quite reasonably concerned. In fact, the relationship between everyone involved in these kinds of measurements appears to have gone through some rocky periods over the last few years, such that supporters of M-Lab are reasonably worried that there's a concerted effort to gradually move them out of the process.