Et tu, Brute?… aka Julius Genachowski, data cap booster (2)

(This continues yesterday’s list on why caps are bad for you.)

4 – Throttling has been debunked and (might be) discontinued in Canada.

Throttling was foisted on Canadians for years, until the incumbents finally got shamed into giving it up as an alleged solution to congestion. This year both Rogers and Bell will (they say) end the practice. If Comcast and its sister incumbents are trying this on, Americans can look North for guidance.

Incumbent ISPs in many jurisdictions have long been enthusiastic about throttling, a form of subscriber traffic-shaping that relies on deep packet inspection (DPI). Throttling and data caps are different but related. Throttling is a technical form of Internet traffic management (hence technical ITMPs); data caps constitute an economic form of Internet traffic management (hence economic ITMPs). For reasons I have always found misguided, the CRTC ruled in October 2009 that economic ITMPs should take precedence over technical ITMPs (Review of the Internet traffic management practices of Internet service providers, Telecom Regulatory Policy CRTC 2009-657).


From the CRTC’s perspective, the ITMP provisions constitute Canada’s network neutrality rules. But the framework suffers from a combination of factors: retail broadband isn’t regulated; the incumbents exploit the staggering level of subscriber ignorance about what they’re buying (nobody outside the business knows what a megabit is); and the CRTC has a long history of ignoring complaints about technical interference with subscriber transmissions.

From the incumbents’ perspective, very few CRTC decisions make any felt difference in their marketplace behavior – and that certainly includes throttling. The pretext for throttling has almost invariably been BitTorrent and other P2P transmissions, which suck up too much bandwidth for the ISPs’ liking. Cisco data from its Visual Networking Index show P2P traffic is declining as a proportion of overall Internet traffic – one change in the real world that has helped discredit throttling. The problem is that, until great pressure was put on the CRTC, throttling was carried out by Canadian ISPs indiscriminately and far beyond any actual need to manage the network (see my series of posts from 2011 on the case of Rogers throttling of World  of Warcraft, which dragged on for a year – starting with “Game throttled? Complain about Rogers, but blame the CRTC“).

On top of the ISPs using throttling like a loose cannon to be trained on their customers at large, recent data show our ISPs have not been rushing to conform to the 2009 neutrality (ITMP) rules. An international research group headed by Milton Mueller has run a project for the last several years designed to investigate the use by ISPs of invasive online technologies, particularly DPI. The project operates using Google’s MLab platform and crowdsourced data from a network monitoring test known as Glasnost. Here are two striking findings about Canada from the data collection up to and including Q1 2012:

  • Bell and Rogers have been world-class throttlers, setting records compared to ISPs in many other countries. In the chart below, data from years 2009, 2010, 2011, and the first quarter of 2012, have been accumulated, and show for each ISP the proportion of end-user tests (in percent) that revealed DPI was in use to manipulate BitTorrent during the transmission being tested. Rogers Cable holds 4th place, at 80%, while its frenemy Bell Canada is right on its heels at about 78%.

  • Bell and Rogers did not eliminate throttling after issuance of the 2009 ITMP guidelines, far from it. The chart below shows a time series covering 16 Canadian ISPs and the extent of their use of DPI from Q1 2009 to Q1 2012. The two highest bars on the right-hand X axis are Rogers and Bell.


Why are these findings important in the context of data caps?

As I explained above, caps and throttling are two different tools that incumbent ISPs use to manipulate customer use of their networks. Both have been sanctioned – with conditions – by the CRTC. But the CRTC’s hands-off approach, along with its attempts last year to hide the public’s complaints from the rest of the public, mean that the incumbents are all the more likely to use these tools in ways that serve their interests, while ignoring or harming customer interests. On that point, I’m much less sanguine – and less generous toward the CRTC – than Michael Geist in his blog posts, such as one from March entitled “How the CRTC Helped to Put An End to Internet Throttling.”

5. Don’t count on investment to rid us of data caps

The CRTC stipulated in its 2009 ITMP decision that “network investment is a fundamental tool for dealing with network congestion and should continue to be the primary solution that ISPs use.” It’s a nice idea. But if you’re a senior executive running a publicly traded ISP, and Bay (or Wall) Street is telling you to cool the investment strategies if you want your share price to hold… where are you going to push? Against millions of captive subs, by throwing on, say, data caps to improve your ARPU. That approach means you can avoid spending a lot of shareholder money on new projects, like pushing fiber deeper into local neighborhoods – exactly the course Verizon is following in pulling back on its FiOS investment after hoodwinking taxpayers and customers out of billions.

In the headline wars, this is another area where the incumbents in both the US and Canada have done a great job on the propaganda: regulate us more and we’ll just invest less. Ah but there’s good data indicating the very opposite, as in a 2009 report from Free Press that debunks the idea:

“A new report by Free Press provides overwhelming evidence, despite the claims of companies like AT&T, that Network Neutrality will not harm network investment. On the contrary, the report shows that open Internet rules will likely have a positive impact on investment in both the network and applications markets.”

(I’ve uploaded the report here.)

6 – Caps put integrated ISPs into a big fat conflict of interest.

Although the FCC has sent signals it’s still rooting for consumers in this debate, there’s an inevitable scenario waiting to happen that will screw both consumers and OVDs like Netflix.

Pundits have had much to say recently about Comcast not counting against subscriber usage any data from a service it has a stake in, like its Xfinity service – whereas it continues to count data from online services it carries that compete directly against it, like Netflix. Anybody who thinks Comcast won’t push as hard as it can on this lever should go back to Vertical Integration school. Since an integrated incumbent with the market power of Comcast has every motive to shut out new competitors, it will gleefully take the opportunity offered by Genachowski to add caps to its arsenal. Why wouldn’t it?

Canadians have already been through this movie, as when Netflix announced in July 2010 it was coming to Canada – and four days later, Rogers lowered the caps on two of its access tiers. This conflict is becoming more urgent for Reed Hastings as new data show Netflix may have peaked as the 800-lb OVD gorilla. According to findings released yesterday by Sandvine, “Netflix represents 33% of peak downstream traffic on US cable networks but [Sandvine] projects a declining percentage as early as next year.” Why? Because of “increased competition from Hulu, Amazon Prime and MSO-branded streaming services.”

(I note in passing that on Sandvine’s product page, the company touts the following as one of the benefits of its Traffic Management offering: “Deferred capital investments resulting from extended infrastructure lifetime.” The Waterloo-based company was one of the first  to offer very high-speed traffic switches that as far back as 2006 featured “per-subscriber deep packet inspection (DPI)-based policy solutions to solve both business and technology challenges in the world’s largest broadband networks.” Like you pesky BitTorrent fiends.)

7 – The rest of the developed world doesn’t need caps: why should the US?

The graphic below is taken from the OECD broadband portal and reflects the most recent data on use of caps across member states – i.e. from September 2011. The light purple swath represents (at least relative) freedom from caps in 28 of these 31 OECD countries (there are currently 34 members in total). If you want to see the Excel spreadsheet with the numbers, I’ve uploaded it here. Note that Canada holds the #4 position of shame in the ranking for most prevalent use of caps.

For the moment, the USA is right at the top, all purple and consumer-friendly. If Genachowski thinks caps are such an important part of developing broadband business models, he should explain to Americans outside the Beltway why the rest of the developed world has shunned them.


Source: OECD Broadband Portal