
Facing an investigation by the FCC for their
traffic shaping practices, Comcast last week quietly
changed their terms of service. While Comcast insisted to us the changes were an effort to make traffic shaping more clear to consumers, it's clear to us that the wording was changed to more directly mirror the FCC's
policy statement on network neutrality, which allows all but a blockade of competing traffic as long as an ISP lawyer can justify the practice as
"reasonable network management."
The word "reasonable" makes repeated appearances in the new TOS, which lacks any specifics on exactly how much monthly bandwidth consumption is too much. Comcast's use of Sandvine hardware to send forged TCP packets -- which throttles upstream p2p traffic for all Comcast users regardless of excessive use -- also isn't clearly illuminated in the new TOS, which justifies more than it explains.
Comcast yesterday filed a
statement (pdf) with the FCC vigorously defending their traffic shaping practices. That defense makes the claim that
"no valid conclusions about the effects of Comcast's network management practices could be drawn" from tests of Comcast's traffic shaping (including those by
our users, the
Associated Press and
the EFF) because the tests
"did not replicate how p2p protocols operate in the real world."
The EFF clearly showed in their
recent report that Comcast could have chosen less draconian traffic shaping methods that incurred less collateral damage. Still, Comcast urges the FCC to determine that Comcast's particular brand of network management is reasonable under the definition in the FCC's policy statement on network neutrality.
Critical decisions should not be based on the demands of the vocal minority who make the most noise in public forums
-Comcast
The company goes on to ask the FCC
"to make it clear that it will not be drawn into second guessing the reasonable network management decisions that engineers and service providers must make on a daily -- and sometimes hourly -- basis to respond to a dynamic and ever-changing Internet." According to the company, such
"critical decisions should not be based on the demands of the vocal minority who make the most noise in public forums."
All told, Comcast uses the word "reasonable"
more than forty times in their statement to the FCC. Unlike the company's TOS, they do inform the FCC that the company uses "reset packets" to throttle upstream p2p traffic, though they don't mention specific hardware, and the company takes issue with use of the word "forged" to explain these packets:It is not accurate to describe these reset packets as "forged," and Free Press's attempted analogy to a telephone operator impersonating the called and calling parties to a phone conversation is inflammatory hyperbole, not fact. A "reset" is nothing more than a bit in the TCP packet header that is used to signal that there is an error condition within the network, and that a new connection needs to be established...
it is much like what occurs when a fax machine receives a busy signal and the machine automatically redials until the facsimile goes through, except that in the case of P2P, the downloading computer may have hundreds or thousands of other computers to look to for the desired file.
Indeed the defense is jam packed with vibrant new metaphors to justify Comcast's particular choice of network management.
"One would not claim that the car is "blocked" or 'prevented' from entering the freeway," insists Comcast in their 80-page defense.
"Rather, it is briefly delayed, then permitted onto the freeway in its turn while all other traffic is kept moving as expeditiously as possible."
We'll let you dissect Comcast's semantics here, lest we get long winded. Do you think the FCC, who has a recent history of picking on cable operators, will buy Comcast's defense? Do you think ISPs should be clear in informing potential customers as to precisely what kind of traffic shaping they're using to manage their networks? Does repeated use of a single word actually bend space/time and change reality? Voice your opinion in the comment section below.
read comment(s)