dslreports logo
 story category
Time Warner Cable Execs Gamed FCC Speed Measurements

Back in 2011, you might recall that the FCC began recruiting volunteers, giving them custom-firmware embedded routers tracked by a company named SamKnows. This real-world network performance data helped inform FCC policy (what a novel concept), and the agency began naming and shaming ISPs that failed to deliver advertised speeds during peak usage periods. Overall the practice was effective, with many ISPs quickly provisioning speeds above what was advertised to avoid being singled out by the FCC.

Click for full size
But there have always been questions regarding whether or not ISPs were gaming this data collection, potentially by specifically giving these FCC volunteers better speeds than they might otherwise get as non-test subjects.

On that front, buried in the New York Attorney General's lawsuit against Charter for substandard speeds filed yesterday is an interesting admission from Time Warner Cable executives.

The AG's full complaint (pdf) notes that to help anticipate the FCC's findings, "Spectrum-TWC independently contracted with Sam Knows to install a parallel, internal panel of whiteboxes in Spectrum-TWC network centers and the homes of Spectrum-TWC employees across the country." The report notes that as of last year, Time Warner Cable had about 1,200 such boxes scattered around the country.

Ultimately Time Warner Cable and Charter's data collection only confirmed what the FCC's own data was finding: that users on the company's 100 Mbps plan in New York City received less than 80% of the advertised speed; subscribers on the 200 Mbps plan received less than 60% of the advertised speed, and subscribers on the 300 Mbps plan generally received 38% to 74% of the bandwidth promised.

Like most of the ISPs being tracked by the FCC, Charter decided that the best solution was to over-provision user tiers so that they'd possibly get more than the advertised rates, or at worst advertised speeds.

"We recommend increasing over-provisioning our modem speeds to around 20% to drive our Sam Knows scores > 100% and then to market that we deliver more than promised speeds," one executive acknowledged in an e-mail. This was, a Time Warner Cable engineering presentation insisted before this was performed, like "putting lipstick on a pig."

The AG's report acknowledges as much, noting that this over-provisioning behavior involved "padding the test result average with scores from times when a service group was not heavily utilized," but "masked the widespread deployment of deficient older-generation, single-channel modems, the prevalence of heavily utilized heavily congested service groups and the poor physical health of neighborhood cable lines."

Last year, you'll recall that many incumbent ISPs were accused by Cogent, Level 3, Netflix and others of intentionally letting their peering points get congested to force content companies and transit operators to pay higher rates for interconnection and eliminate the idea of settlement-free peering. The AG's complaint makes it clear that Time Warner Cable executives were well aware its decisions on interconnection could impact end-user speeds, so executives at several points suggested a rather unique solution; manipulation of peering capacity to falsely give the impression of quality end-user connection performance:

quote:
Our Sam Knows scores are like watching a slow-motion train wreck. We need to get in front of this. One thing I think we may need to be prepared to do is just give more ports to Cogent during sweeps month [when FCC results are measured for purposes of the MBA report]. We don’t have to make any promises, we just have to make it work temporarily.
In case it's not clear, that's a Time Warner Cable executive admitting it wanted to temporarily resolve peering issues it appears to have caused itself to provide fake performance data to the regulator overseeing it.

The AG's report is quick to note (on page 23) that the average peak hour packet loss for traffic carried by Cogent to Spectrum-TWC subscribers from 2014 through 2015 was far higher than the packet loss experienced by subscribers to another major New York-area cable ISP that maintained sufficient port capacity with Cogent (likely Cablevision, see graphic, above left).

"Spectrum-TWC knew that during the pendency of its dispute with Cogent, Spectrum-TWC’s subscribers were not getting reliable access to online content, and were experiencing packet loss and high latencies," the AG's complaint concludes. "Despite its knowledge that it was not delivering the Internet services it had promised to its subscribers, Spectrum-TWC failed to take any steps to invest in additional port capacity for its network for much of the Relevant Period."

Once the FCC's net neutrality rules were passed -- specifically the portion prohibiting such interconnection shenanigans -- Charter and Cogent's feud quickly ended, as did numerous other disputes between last mile ISPs and transit operators. Needless to say, with net neutrality rules now potentially on their death bed, it's likely we'll see a return to the kind of interconnection disputes and Netflix streaming issues that pockmarked 2013 through 2015, as ISPs again look to seek their pound of flesh out at the edge of the network.

Most recommended from 44 comments



Dryvlyne
Far Beyond Driven
Premium Member
join:2004-08-30
Newark, OH

25 recommendations

Dryvlyne

Premium Member

Who says regs aren't needed?

This is almost funny until you realize it isn't. Let me hear all the anti-regulation haters out there who thinks companies don't need oversight and that the "free-market" will naturally force companies to be good corporate-citizens.

TestBoy
Premium Member
join:2009-10-13
Irmo, SC

4 recommendations

TestBoy

Premium Member

May have or did?

I read the article at stop the cap.
Seems that it did happen... not that it "might have happened"

Or is this the "innocent until proven guilty" notion that Americans have despite it being the opposite?

Smith6612
MVM
join:2008-02-01
North Tonawanda, NY
·Charter
Ubee EU2251
Ubiquiti UAP-IW-HD
Ubiquiti UniFi AP-AC-HD

2 recommendations

Smith6612

MVM

Over-provisioning is actually a good thing

It seems that the article is framing over-provisioning as a bad thing. To be honest, on a residential service it really isn't. While sure, it adds to congestion on the last mile, the over-provisioning keeps customer satisfaction up if people are truly getting more than what they pay for, and see (and know) the fact. It's a bit like the whole PowerBoost thing prior to DOCSIS 3.0, although a much better implemented solution. Also, think about it this way. If I see congestion and it's not hitting the full over-provisioned rate, a person like me would consider that simply bonus data and not call in the trouble.

It'd be nice if ISPs weren't so busy touting Wi-Fi speeds and pushing add-on services, and focus on their network instead. Offering up decent CPE too, if they're going to force or default to having people rent their equipment in some way, shape or form.