from the well-of-course-they-are dept
Recently, the FCC measurement program has backed sharply away from their commitment to transparency, apparently at the bidding of the telcos in the program. The program is now proposing to replace the M-Lab platform with only ISP-managed servers. This effectively replaces transparency with a closed platform in which the ISPs -- whose performance this program purports to measure -- are in control of the measurements. This closed platform would provide the official US statistics on broadband performance. I view this as scientifically unacceptable.The FCC keeps insisting that it's committed to openness -- but all too frequently seems to give in to telco demands. So this warning is concerning.
For the health of the Internet, and for the future of credible data-based policy, the research community must push back against this move.
For what it's worth, the telcos are claiming that Cerf is overreacting. In a response to his call for action, Verizon's David Young responded that there's nothing to see here, and that M-Lab and the telco efforts have co-existed and can continue to co-exist going forward.
Vint breathlessly suggests that the FCC is now backing away from this openness "at the bidding of the telcos" and claims the program is proposing to replace the M-Lab platform with only ISP-managed servers. THIS IS FALSE. ISPs have made no such request of the FCC nor has the FCC proposed to eliminate use of M-Lab’s servers.As with many such disputes, the reality may be somewhere in between the two claims here. It seems like Cerf's fear is that by establishing the telcos' servers on equal footing with the M-Labs' open setup, it opens the door to replacing the M-Labs' efforts and then potentially locking up the data. Young is correct that the openness is mainly due to FCC policy at this point, but that policy is dependent on the current leadership of the FCC, which could change. At the very least, it would be nice to see a stated commitment to keeping the information open on an ongoing basis, so that there isn't any need to worry going forward.
What has been proposed is that, in addition to continuing to use the data collected via the M-Lab servers, the FCC and SamKnows may also rely on the ISP provided servers that have been in use since the beginning of the project. These ISP-provided servers meet the specifications required by SamKnows as do the M-Labs servers. In fact, it was only because of the presence of these non-M-Lab, ISP-donated servers, that SamKnows was able to identify problems with an M-Lab server that was affecting the results of the tests being conducted. M-Labs did not identify this server problem on their own. It was only fixed when SamKnows brought the issue to their attention. By the way, this problem forced the FCC to abandon a month's worth of test data, extend the formal test period and delay production of their report. Later, another M-Lab server location had transit problems that again affected results. This was the second M-Labs-related server problem in two months and once again, it was SamKnows, using the ISP-provided servers as a reference who identified the problem and brought it to M-Labs attention.