Thursday, December 25, 2008
NN Misconceptions
[Buried lede alert: Peer to peer is edge caching at its finest, and it's NOT a violation of NN.]
Boy, howdy! Network Neutrality news blankets the bloomin' blogosphere. Steve Schultze has compiled 20-some blog articles triggered by the WSJ's clumsy-to-clueless article.
I have been in smoldering slow burn mode on a couple points, and Brendan Ballou's piece in JZ's blog fanned the smoke to light . . . so here goes:
Making bits go from one location to another faster is not, of itself, a violation of ANY REASONABLE DEFINITION of Network Neutrality. I am flipping tired of hearing about treating all bits (or packets) exactly the same. Nobody does this. Nobody can do this. There is no reasonable definition of NN in terms of exactly the same treatment of bits or packets.
If my server has a 10 gigabit per second connection to the Internet, and yours only has a 256 kilobit connection, chances are good that bits from my server will arrive faster than bits from yours. This is not a violation of NN.
Of course a 10 gigabit connection costs more than a 256 kilobit connection. I need to be rich enough to afford a 10 gigabit connection. There are some advantages that come because you're big and rich that have nothing to do with NN. If I'm a big and rich Internet applications company, I can afford a great cafeteria for my employees, et cetera. By the same token, maybe I can afford to access multiple paths through the Internet so I can figure out which paths are faster, and send my data preferentially over these paths, and if so, that's not a violation of NN.
This doesn't mean there aren't reasons to fear big, rich Internet applications companies, or to limit their power. But these reasons might well have nothing to do with NN.
Edge caching, as Google's Derek Slater defines it in an earlier comment in this blog, is not inside the Internet. It is at the edge of the Internet. Content arrives at an edge cache via anybody's stupid pipes. It leaves via anybody's stupid pipes. An edge cache, defined this way, is not a means for violating NN.
Note that Peer to Peer is a form of edge caching. It might be THE ULTIMATE form of edge caching. When I use Skype or BitTorrent, some of the relevant data may be cached on my machine, precisely because it is closer (in some statistical Internet topology sense) to the next potential user of that data.
Do the hackles on your neck go up when somebody suggests that the Internet should be used for what it was originally designed to do? Mine sure as flack do! As if it were designed to do Web surfing and email but not video or meter reading! Hey, we're not done inventing all the things we can do on the Internet, and we sure don't need the telcos and cablecos, who didn't invent any of it, telling us what we can and can't do. When the telcos and cablecos tell us we can do this, but not that, or that we can do this activity or access that information but it will cost us more, that's where the Internet starts to lose its groove.
Boy, howdy! Network Neutrality news blankets the bloomin' blogosphere. Steve Schultze has compiled 20-some blog articles triggered by the WSJ's clumsy-to-clueless article.
I have been in smoldering slow burn mode on a couple points, and Brendan Ballou's piece in JZ's blog fanned the smoke to light . . . so here goes:
Making bits go from one location to another faster is not, of itself, a violation of ANY REASONABLE DEFINITION of Network Neutrality. I am flipping tired of hearing about treating all bits (or packets) exactly the same. Nobody does this. Nobody can do this. There is no reasonable definition of NN in terms of exactly the same treatment of bits or packets.
If my server has a 10 gigabit per second connection to the Internet, and yours only has a 256 kilobit connection, chances are good that bits from my server will arrive faster than bits from yours. This is not a violation of NN.
Of course a 10 gigabit connection costs more than a 256 kilobit connection. I need to be rich enough to afford a 10 gigabit connection. There are some advantages that come because you're big and rich that have nothing to do with NN. If I'm a big and rich Internet applications company, I can afford a great cafeteria for my employees, et cetera. By the same token, maybe I can afford to access multiple paths through the Internet so I can figure out which paths are faster, and send my data preferentially over these paths, and if so, that's not a violation of NN.
This doesn't mean there aren't reasons to fear big, rich Internet applications companies, or to limit their power. But these reasons might well have nothing to do with NN.
Edge caching, as Google's Derek Slater defines it in an earlier comment in this blog, is not inside the Internet. It is at the edge of the Internet. Content arrives at an edge cache via anybody's stupid pipes. It leaves via anybody's stupid pipes. An edge cache, defined this way, is not a means for violating NN.
Note that Peer to Peer is a form of edge caching. It might be THE ULTIMATE form of edge caching. When I use Skype or BitTorrent, some of the relevant data may be cached on my machine, precisely because it is closer (in some statistical Internet topology sense) to the next potential user of that data.
Do the hackles on your neck go up when somebody suggests that the Internet should be used for what it was originally designed to do? Mine sure as flack do! As if it were designed to do Web surfing and email but not video or meter reading! Hey, we're not done inventing all the things we can do on the Internet, and we sure don't need the telcos and cablecos, who didn't invent any of it, telling us what we can and can't do. When the telcos and cablecos tell us we can do this, but not that, or that we can do this activity or access that information but it will cost us more, that's where the Internet starts to lose its groove.
Technorati Tags: Content-Conduit, End-to-End, NetworkNeutrality, WSJ
Comments:
The text above (Who wrote it? The attribution is unclear) misses the point.
Network neutrality advocates — especially the most extreme ones, such as Free Press — claim that ISPs, who are primarily in the infrastructure business but sometimes provide content and services, will use their infrastructure to privilege those content and services. But Google, which started in the content and services business, is now building out infrastructure which is exclusively devoted to delivering their content. (They say that others can deploy such infrastructure, but this is akin to saying that freedom of the press is available to anyone who can afford to own one.) So, Google — though it is coming from the opposite direction — winds up in the same place: as an infrastructure owner whose infrastructure can discriminate in favor of its own content. But there's a big difference. The claim that the carriers might discriminate is speculation and fearmongering. However, Google has announced, point blank, that it fully intends to discriminate in a way that puts competitors at a disadvantage.
The result is that advocates of “network neutrality” now face three conundrums. The first is that “network neutrality” is such an ill defined bundle of issues — with so many bags hung on the side by various interest groups — that it is impossible to discuss it sensibly as one issue. To date, many people who have opposed sweeping and onerous “network neutrality” regulation — including myself — have stated that they would be willing to support a narrow definition that doesn’t include all of these “bags on the side.” But this issue threatens to add yet another item to the bundle. Even for many supporters of “network neutrality,” it’s beginning to look a little too bulky.
What’s more, some lobbyists, especially those who are strongly aligned with Google (e.g. Free Press and Public Knowledge), have sought to dismiss “edge caching” as unrelated, because the issue is inconvenient and adding it would threaten their alliance with Google. But deep down, they know that it is actually more relevant than many other things which they’ve already opted to add to the bundle.
Finally, the advocates of “network neutrality” — in particular, Free Press, Public Knowledge, and Media Access Project — have always claimed to be populist but now find themselves to be very much, and very obviously, in bed with corporate interests. If they break with those interests, they will likely lose the generous financial support of the corporations, such as Google, which have driven the “network neutrality” agenda from the start. But if they don’t break with those interests, they will have so obviously sold out that they will forever have compromised their integrity and public images.
For all of these reasons, I would assert that it is time to dump the term "net neutrality" entirely and start afresh for the new year, hammering out a workable broadband policy unburdened by the baggage of the past. Let's start by defining goals (such as greater broadband deployment, good quality of service, and allowing innovation both by ISPs and by content and service providers) and by addressing issues related to those goals individually. We should start with the issues on which most people agree — for example, that anticompetitive conduct should be prohibited — and act only on those issues where there is a reasonable consensus. To instead rush headlong into legislation that attempts to treat the issues as an inseparable bundle — such as last year’s Dorgan-Snowe or Markey bills — would be a very bad move and would be detrimental to the country.
Network neutrality advocates — especially the most extreme ones, such as Free Press — claim that ISPs, who are primarily in the infrastructure business but sometimes provide content and services, will use their infrastructure to privilege those content and services. But Google, which started in the content and services business, is now building out infrastructure which is exclusively devoted to delivering their content. (They say that others can deploy such infrastructure, but this is akin to saying that freedom of the press is available to anyone who can afford to own one.) So, Google — though it is coming from the opposite direction — winds up in the same place: as an infrastructure owner whose infrastructure can discriminate in favor of its own content. But there's a big difference. The claim that the carriers might discriminate is speculation and fearmongering. However, Google has announced, point blank, that it fully intends to discriminate in a way that puts competitors at a disadvantage.
The result is that advocates of “network neutrality” now face three conundrums. The first is that “network neutrality” is such an ill defined bundle of issues — with so many bags hung on the side by various interest groups — that it is impossible to discuss it sensibly as one issue. To date, many people who have opposed sweeping and onerous “network neutrality” regulation — including myself — have stated that they would be willing to support a narrow definition that doesn’t include all of these “bags on the side.” But this issue threatens to add yet another item to the bundle. Even for many supporters of “network neutrality,” it’s beginning to look a little too bulky.
What’s more, some lobbyists, especially those who are strongly aligned with Google (e.g. Free Press and Public Knowledge), have sought to dismiss “edge caching” as unrelated, because the issue is inconvenient and adding it would threaten their alliance with Google. But deep down, they know that it is actually more relevant than many other things which they’ve already opted to add to the bundle.
Finally, the advocates of “network neutrality” — in particular, Free Press, Public Knowledge, and Media Access Project — have always claimed to be populist but now find themselves to be very much, and very obviously, in bed with corporate interests. If they break with those interests, they will likely lose the generous financial support of the corporations, such as Google, which have driven the “network neutrality” agenda from the start. But if they don’t break with those interests, they will have so obviously sold out that they will forever have compromised their integrity and public images.
For all of these reasons, I would assert that it is time to dump the term "net neutrality" entirely and start afresh for the new year, hammering out a workable broadband policy unburdened by the baggage of the past. Let's start by defining goals (such as greater broadband deployment, good quality of service, and allowing innovation both by ISPs and by content and service providers) and by addressing issues related to those goals individually. We should start with the issues on which most people agree — for example, that anticompetitive conduct should be prohibited — and act only on those issues where there is a reasonable consensus. To instead rush headlong into legislation that attempts to treat the issues as an inseparable bundle — such as last year’s Dorgan-Snowe or Markey bills — would be a very bad move and would be detrimental to the country.
P.S. -- One more point. Peer-to-peer isn't "edge caching" at all, because it does not create a single common cache that serves all of the users of a particular portion of the "edge" near it. What's more, it is extremely wasteful of resources, because it consumes valuable and expensive edge bandwidth instead of less expensive bandwidth closer to the center of the network.
Post a Comment
# posted by 
: 12/28/2008 3:51 PM
: 12/28/2008 3:51 PM
