Just What Are the Facts?

by on August 2, 2008 · 18 comments

As expected, the FCC has chosen Comcast as the target of its biggest net neutrality enforcement action to date.  I wonder whether the FCC has actually chosen a good set of facts to serve as the foundation for what may possibly be a broad new precedent (we won’t know how broad until the commission publishes the order), considering that the commission will likely be forced to defend it in court.  Like it or not, FCC decisions are required to have a “rational basis.”

FCC Chairman Kevin Martin suggests Comcast acted atrociously:

While Comcast claimed its intent was to manage congestion, they evidence told a different story:

  • Contrary to Comcast’s claims, they blocked customers who were using very little bandwidth simply because they were using a disfavored application;
  • Contrary to Comcast’s claims, they did not affect customers using an extraordinary amount of bandwidth even during periods of peak network congestion as long as he wasn’t using a disfavored application; 
  • Contrary to Comcast’s claims, they delayed and blocked customers using a disfavored application even when there was no network congestion;
  • Contrary to Comcast’s claims, the activity extended to regions much larger than where it claimed congestion occurred.

In short, they were not simply managing their network; they had arbitrarily picked an application and blocked their subscribers’ access to it

Yet Commissioner Robert McDowell seems to claim that the evidence is insubstantial:

The truth is, the FCC does not know what Comcast did or did not do. The evidence in the record is thin and conflicting.  All we have to rely on are the apparently unsigned declarations of three individuals representing the complainant’s view, some press reports, and the conflicting declaration of a Comcast employee. The rest of the record consists purely of differing opinions and conjecture. [footnote omitted]

Comcast has pledged to move to a “protocol agnostic” capacity management technique by the end of this year, so Martin’s first two statements of fact are moot.  But the second two claims — that Comcast practiced network management where there was no congestion — are clearly at odds with the following statements inserted by Comcast into the official record which will be tested at trial:

Data collected recently from the Comcast network demonstrate that Comcast’s network management practices are minimally intrusive. Comcast has never managed customers’ downloads, and the data show that, even with the current management of P2P uploads, P2P?traffic continues to comprise approximately half of upstream traffic transmitted on the Comcast network — and, in some locations, P2P traffic is as much as two-thirds of total upstream bandwidth. The data also show that, on a typical day, an estimated 9 billion P2P TCP flows traverse Comcast’s network, and, even for the most heavily used P2P protocols, more than 90?percent of these flows are unaffected by Comcast’s network management. Given the vast amounts of P2P traffic carried on Comcast’s network, and the small percentage of total uploads that are delayed, it is clear that Comcast’s customers are able to (and do) use any application or service they choose, including those that utilize P2P protocols. [footnote omitted]

The data also suggest that, even in a cable system with heavy P2P usage, when a P2P upload from a particular computer was delayed by a reset packet, that same computer successfully initiated a P2P upload within one minute in 80 percent of the cases. In fact, most?Comcast customers using P2P protocols to upload never experienced any delay at all. Thus, even when a subscriber’s computer encounters a so-called “busy” signal when it attempts to upload a file, the busy condition generally causes only brief delays before that computer is able to effectuate its next upload. In other words, as Comcast has consistently maintained, this current network management technique delays a relatively small number of P2P uploads and only delays them temporarily….

Comcast has explained that its current network management practices are triggered by certain threshold levels of P2P protocol use that could, if unchecked, lead to harmful network congestion. Specifically, Comcast’s current P2P management is triggered when the number of P2P uploads in a given area for a particular P2P protocol reaches a certain, pre-determined level, regardless of the level of overall network traffic at that time, and regardless of the time of day when the applicable P2P protocol threshold is reached.

Martin also claims that Comcast interfered with downloads, not just uploads:

Comcast was delaying subscribers’ downloads and blocking their uploads.

But over at Broadband Politics, Richard Bennett refutes this:

Comcast does not interfere with BitTorrent downloads on its network; in fact, they prevent BitTorrent seeding from interfering with BitTorrent downloading and actually improve the performance of the application the Commission says it’s “disfavoring.”

Nobody has complained that you can’t download, and downloading is how you access the content of your choice. The actual complaint is that the amount of bandwidth Comcast allocates to BitTorrent acting as a file server is not enough. This can have an effect, typically a small one, on the ability of other people, especially those outside the Comcast network, to download the Comcast customer’s content, but there’s no “freedom to run a file server from your home.”

Throughout this whole debacle, I’ve repeatedly tested whether I could download movies and software using BitTorrent on the Comcast network, and at no time has it been blocked. There have been periods during which BitTorrent was slowed for seedding, when I wasn’t trying to download a file, just serving files to others, and that’s it.

If the FCC has based its decision on an erroneous or insubstantial set of facts — which wouldn’t be a first —  nothing will have been accomplished although it may take years of litigation to establish that. Meanwhile, there is always some hesitancy in Congress to legislate if there is litigation pending in the courts which may resolve the matter.  Maybe that’s what we have here:  a gambit which can’t succeed and a recipe for years of further delay.

  • http://enigmafoundry.wordpress.com eee_eff

    Hmmm..

    Well Hance the Max Planck Institute has done a certain amount of network testing that shows that it is true that Bit-Torrent was blocked, irrespective of the network congestion:

    IS BITTORRENT BLOCKED ONLY DURING PERIODS OF PEAK NETWORK CONGESTION?
    Recently it has been reported that Comcast defended its BitTorrent blocking before FCC as a necessary practice that is done only during periods of heavy network traffic. It is widely known that network traffic exhibits a strongly diurnal pattern. So we analyzed our data to see if hosts in Comcast and Cox networks see fewer of their upstream transfers blocked during early morning or weekends (when network load is generally low) than during other times of the day.
    The graphs below show (a) the number of measurements to Comcast hosts at different hours of the day and (b) the percentage of these measurements for which we observed BitTorrent blocking. The percentage of blocked connections remains high at all times of the day. Our data suggests that the BitTorrent blocking is independent of the time of the day.

    See: http://broadband.mpi-sws.mpg.de/transparency/results/

    Note that I have a very long comment that is in the moderation queue in your “Hijacking the internet” post, if you’d like several more links.

  • http://enigmafoundry.wordpress.com eee_eff

    Hmmm..

    Well Hance the Max Planck Institute has done a certain amount of network testing that shows that it is true that Bit-Torrent was blocked, irrespective of the network congestion:

    IS BITTORRENT BLOCKED ONLY DURING PERIODS OF PEAK NETWORK CONGESTION?
    Recently it has been reported that Comcast defended its BitTorrent blocking before FCC as a necessary practice that is done only during periods of heavy network traffic. It is widely known that network traffic exhibits a strongly diurnal pattern. So we analyzed our data to see if hosts in Comcast and Cox networks see fewer of their upstream transfers blocked during early morning or weekends (when network load is generally low) than during other times of the day.
    The graphs below show (a) the number of measurements to Comcast hosts at different hours of the day and (b) the percentage of these measurements for which we observed BitTorrent blocking. The percentage of blocked connections remains high at all times of the day. Our data suggests that the BitTorrent blocking is independent of the time of the day.

    See: http://broadband.mpi-sws.mpg.de/transparency/re

    Note that I have a very long comment that is in the moderation queue in your “Hijacking the internet” post, if you’d like several more links.

  • http://bennett.com/blog Richard Bennett

    The Max Planck study is, with all due respect, a load of crap. The simply collect data from a single site in Germany which fails to do a fair simulation of BitTorrent. The Planck people are so embarrassed by their crappy test that they’ve stopped answering questions about it.

    Vuze created a plug-in for their P2P client that records and reports network statistics, and it reports wildly different numbers than the Planck claims.

    The critics fail to appreciate that BitTorrent opens thousands of TCP virtual circuits, and Comcast only closes some of them with the RST packets. The BT app simply opens new ones and continues to run. As I said, accessing the content of your choice is unimpaired, serving the content of your choice is another story.

  • http://bennett.com/blog Richard Bennett

    Also, note that Planck assumes that periods of high load for seeding are the same as for interactive, download traffic. This doesn’t follow, as seeding offers data to interactive users all over the Internet, and it’s always prime time somewhere.

  • http://bennett.com/blog Richard Bennett

    The Max Planck study is, with all due respect, a load of crap. The simply collect data from a single site in Germany which fails to do a fair simulation of BitTorrent. The Planck people are so embarrassed by their crappy test that they’ve stopped answering questions about it.

    Vuze created a plug-in for their P2P client that records and reports network statistics, and it reports wildly different numbers than the Planck claims.

    The critics fail to appreciate that BitTorrent opens thousands of TCP virtual circuits, and Comcast only closes some of them with the RST packets. The BT app simply opens new ones and continues to run. As I said, accessing the content of your choice is unimpaired, serving the content of your choice is another story.

  • http://bennett.com/blog Richard Bennett

    Also, note that Planck assumes that periods of high load for seeding are the same as for interactive, download traffic. This doesn’t follow, as seeding offers data to interactive users all over the Internet, and it’s always prime time somewhere.

  • http://enigmafoundry.wordpress.com eee_eff

    Also, note that Planck assumes that periods of high load for seeding are the same as for interactive, download traffic. This doesn’t follow, as seeding offers data to interactive users all over the Internet, and it’s always prime time somewhere.

    Not exactly–the Planck Institute did assume that maximum and minimum cycles would be related to the time of day at the location of the network. In other words, traffic is heavy during certain times of the day. That doesn’t seem very controversial.

    Also you are actually right that BT traffic max’s out in a different time than other uses, like streaming video, which max’s out in sync with with http protocol usage. However, BT users, being generally very considerate, set up their BT client to use maximum bandwidth at night, the peak is actually at about 4 am, when there is very little http traffic:

    “AT&T’s new chief technical officer, John Donovan, wants you to know that his company does not, under any circumstances, slow down BitTorrent users or throw other monkey wrenches in the operation of specific applications.
    ….
    For instance, he said, BitTorrent on the company’s network peaks around 4 a.m., when other traffic is at an ebb.”

    http://blog.wired.com/business/2008/06/att-embraces-bi.html

  • http://enigmafoundry.wordpress.com eee_eff

    Also, note that Planck assumes that periods of high load for seeding are the same as for interactive, download traffic. This doesn’t follow, as seeding offers data to interactive users all over the Internet, and it’s always prime time somewhere.

    Not exactly–the Planck Institute did assume that maximum and minimum cycles would be related to the time of day at the location of the network. In other words, traffic is heavy during certain times of the day. That doesn’t seem very controversial.

    Also you are actually right that BT traffic max’s out in a different time than other uses, like streaming video, which max’s out in sync with with http protocol usage. However, BT users, being generally very considerate, set up their BT client to use maximum bandwidth at night, the peak is actually at about 4 am, when there is very little http traffic:

    “AT&T’s new chief technical officer, John Donovan, wants you to know that his company does not, under any circumstances, slow down BitTorrent users or throw other monkey wrenches in the operation of specific applications.
    ….
    For instance, he said, BitTorrent on the company’s network peaks around 4 a.m., when other traffic is at an ebb.”

    http://blog.wired.com/business/2008/06/att-embr

  • http://enigmafoundry.wordpress.com/2008/05/25/ministry-of-truth-at-the-tlf/ enigma_foundry

    As I said, accessing the content of your choice is unimpaired, serving the content of your choice is another story.

    No you actually didn’t say that, at least in this thread.

    In any case, BT consists of uploads and downloads; so Comcast DOES break BT.

    Your interpretation of BT as a “server” has already been quite throughly addressed by Tim Lee in a post he’s made about the time Comcast was caught breaking BT.

    Let’s all recall exactly what Comcast said IN THEIR OWN FAQS:

    From the Comcast website:

    Do you block access to peer-to-peer applications like Bit Torrent?

    No. We do not block access to any web site or applications, including BitTorrent. Our customers use the internet for downloading and uploading files, watching movies and videos…”

    http://enigmafoundry.files.wordpress.com/2007/10/comcast_lies.png

  • http://enigmafoundry.wordpress.com eee_eff

    As I said, accessing the content of your choice is unimpaired, serving the content of your choice is another story.

    No you actually didn’t say that, at least in this thread.

    In any case, BT consists of uploads and downloads; so Comcast DOES break BT.

    Your interpretation of BT as a “server” has already been quite throughly addressed by Tim Lee in a post he’s made about the time Comcast was caught breaking BT.

    Let’s all recall exactly what Comcast said IN THEIR OWN FAQS:

    From the Comcast website:

    Do you block access to peer-to-peer applications like Bit Torrent?

    No. We do not block access to any web site or applications, including BitTorrent. Our customers use the internet for downloading and uploading files, watching movies and videos…”

    http://enigmafoundry.files.wordpress.com/2007/1

  • http://bennett.com/blog Richard Bennett

    The fact remains that millions of people run millions of BitTorrent transactions on Comcast each day, without incident. I’m one of them, and I’ve reported on my test results periodically. The reports of BitTorrent “blocking” on Comcast are greatly exaggerated; the real scenario is rate-controlled seeding after the user’s download finishes.

    EF, don’t keep making the same false claims over and over, it’s boring to have to slap you down several times a day.

  • http://bennett.com/blog Richard Bennett

    The fact remains that millions of people run millions of BitTorrent transactions on Comcast each day, without incident. I’m one of them, and I’ve reported on my test results periodically. The reports of BitTorrent “blocking” on Comcast are greatly exaggerated; the real scenario is rate-controlled seeding after the user’s download finishes.

    EF, don’t keep making the same false claims over and over, it’s boring to have to slap you down several times a day.

  • http://enigmafoundry.wordpress.com eee_eff

    The fact remains that millions of people run millions of BitTorrent transactions on Comcast each day, without incident. I’m one of them,

    Yes, Richard AFTER Comcast got caught they no longer block BT…

    But they did.

  • http://enigmafoundry.wordpress.com eee_eff

    The fact remains that millions of people run millions of BitTorrent transactions on Comcast each day, without incident. I’m one of them,

    Yes, Richard AFTER Comcast got caught they no longer block BT…

    But they did.

  • Pingback: Cigarette electronique

  • Pingback: 1300 numbers australia

  • Pingback: football fantasy

  • Pingback: prix de l'immobilier

Previous post:

Next post: