« As the Cartel Turns |
Main
| Get Your War On, Cartoonists »
February 20, 2008
Could BitTorrent Be Disabled Automatically?
Posted by Alan Wexelblat
This is being painted in the context of net neutrality and copyright enforcement; I see it as a way to automate attacks on any particular users of any information. There's no reason this technique couldn't be used by, say, the Chinese government to disable access to Web sites it finds objectionable. Or paint your own picture.
The story starts with an announcement by AT&T that it's going to police all traffic it carries for copyright violations. That's both stupid (legal liability anyone?) and practically impossible. Packet volumes and encryption render this a nonstarter. The volume of lawsuits alone would be staggering even at AT&T's size.
But let's apply a little intelligence to the problem. Assume you don't need to examine every packet - just the highly visible and highly accessed sources of copied material. Torrents are the prime example of this. Nicholas Weaver wrote in his blog last month of a hypothetical method that would permit AT&T, in cooperation with a copyright-holding entity like the MPAA, to disable torrent downloading.
The plan involves examining a torrent to see if it has material the MPAA doesn't want sent around, then selectively disabling pairwise communication between providers of the torrent and would-be consumers. The torrent identifies participants, so they can be blocked and Weaver describes a fairly clever scheme that disables pairwise communication without harming general network communication. The system has significant advantages to its users, not least of which are that it's completely automated and scalable. It also means AT&T gets out of the content-examination business and avoids the associated liability. The copyright holder (MPAA or other) is examining the content and assuming liability if legitimate content is blocked. This is the same situation we have now with DMCA 'takedown' notices.
The system isn't perfect - I can imagine counter-strategies - but it would certainly disable general P2P networks as they presently operate.
Comments (2)
+ TrackBacks (0) | Category: Big Thoughts
POST A COMMENT
- RELATED ENTRIES
- › MSFT to Region-Lock Xbox One on Launch
- › Myriad Genetics, One Opinion
- › Analyzing Netflix's Economics Misses Netflix's Long Game
- › Oh, Yeah, DOJ is Still Suing Apple over E-Books
- › Why CNN Makes Lots of Money Despite Sucking
- › Microsoft Turns Xbox One into DRM Nanny
- › NFB Settles with Free Library of Phila
- › The White House vs Patent Trolls - So What?
1. Anon. on February 21, 2008 11:23 PM writes...
The system is a joke and would fail instantly.
Why? There is no way to "examine a torrent to see if it has material the MPAA doesn't want sent around" except by having a human look at it individually and personally. Material simply can't be compared with MPAA material automatically. This guarantees failure, at least until massive breakthroughs in computerized pattern-matching are made. The MPAA is simply not going to employ enough people to analyze all the BitTorrent traffic; the moment one copy went down, a new one which was just different enough to not be detected by the same automatic detector would go up.
Second, blocking pairwise communication would have to be based on IP addresses. Which are easy to change, and would promptly be changed.
Meanwhile, the 'false positives' would generate a firestorm of lawsuits and bad publicity for the ISPs. Blocking traffic between two arbitrary computers constitutes a pretty clear violation of the ISP's contract with its customer.
So, no, this wouldn't do a damn thing to the P2P networks; it would be completely routed around within months, if that. It's another impractical scheme with more bad side effects than results.
Permalink to Comment2. drwex on February 22, 2008 9:26 AM writes...
I don't think the number of torrents exceeds the number of songs on P2P nets, and those are examined. Scale and automated examination aren't that difficult.
If you read the proposal you'd see that the block is a temporary one. Essentially just long enough to cause the bittorrent client to have to go elsewhere. Yes, you can change IP address and try again, but as noted in the proposal it can update in near-real time. Since downloading torrents is a long process the system doesn't have to respond instantly, nor does it have to block every single downloader. Even if it only gets 95% that's still a huge hit.
I doubt anyone is going to make a stink about a false positive. For one thing, how would you know the difference between "my ISP is blocking this torrent provider" and "for some reason I'm unable to reach this site for 10 minutes"? The author provides at least one way to gather prima facie evidence that blocks are being used, but as long as they're transient and short-lived who's going to know? Your ISP doesn't provide that kind of service-level guarantee. Or at least none I know of provide it.
Even so, why should the Cartel care? The current DMCA takedown regime generates lots of bad publicity and nobody much notices or cares. The system continues to chug along.
Finally it's not clear to me how to route around this kind of thing. I can see some improved fallback and retry algorithms that would work but it would mean much slower access to torrents.
Permalink to Comment