June 24, 2017

When Is a "Network" Not a Network?

Last week, in response to the MPAA lawsuits against BitTorrent trackers, I wrote that it’s impossible to sue BitTorrent itself, because it is nothing but a communications protocol. Michael Madison was skeptical, which was a fair response given what little I had written on the subject. Let me say a bit more, to clarify.

Opponents of P2P technologies often make the rhetorical move of calling the thing they oppose a “network.” The word carries connotations – especially for nonexperts – of a physical contrivance that is operated by some organization. Think of the old phone system, or the electrical power grid. Somebody has to build and manage all that equipment. The implication is that there is somebody in charge who can supervise the use of the network. Read the plaintiffs’ briefs in the Grokster case and you’ll see many references to a “network” that is “operated” by the defendants.

Computer scientists sometimes use the word “network” to refer to something more virtual. Others are now using “network” in this sense, as when people talk about the social network of friendships among the residents of a small town. Nobody owns and operates the social network. There is nobody you can sue to shut it down, because it’s not a network in the same sense the power grid is.

A communications protocol is an agreement or convention about how computer systems can cooperate to accomplish some task. It isn’t owned or operated by anybody. (People might own copyrights or patents relating to a protocol, but let’s set aside that possibility for now.) There’s a sense in which English or any other human language is a kind of protocol that people use to cooperate with each other. Again: nobody owns, operates or controls the English language, and there is nobody you can sue to shut it down. This isn’t to say that you can’t punish misuses of English, such as fraud or criminal conspiracies that use the language; but punishing misuse is not the same as attacking the language itself.

Given a lawsuit about a particular technology, how can we tell whether that network is more like the power grid or more like a social network? Here I think the Grokster courts have gotten it right. Rather than arguing over what is a “network,” or what “network” means anyway, they looked at the nature of the technology and the defendant’s control or influence over it. That is, as lawyers say, a fact-intensive inquiry.

The MPAA, in suing the operators of BitTorrent trackers rather than trying to attack the BitTorrent protocol itself, seems to be recognizing this distinction. That in itself good news.

Comments

  1. BitTorret and the MPAA

    Ed Felten at Freedom to
    Tinker has a couple of interesting
    articles and links
    regarding BitTorrent
    and the MPAA’s recent lawsuits.

  2. Great text, thank you!
    I if care about a bulgarian translation you can find it at:
    http://blog.peio.org/index.php?p=238

  3. Banning Networks

    Ed Felten claims that it is impossible to sue BitTorrent itself, because it “is nothing but a communications protocol”. … He also mentions “English or any other human language” as another “BitTorrent-like” network… .
    It should be noted, however, …

  4. “English” as an example of a network is a good analogy. It should be noted, however, that “the use of language X” *has* in fact been frequently legislated against, and one might expect that “the use of the BitTorrent protocol” might also seem like a good thing for our legislators (and their lobbyists) to ban. It might even be a wise ‘whack-a-mole’ strategy. I expand on this idea on my own blog, since my original comment grew far too large. =)

  5. Actually, a BitTorrent tracker is the wrong thing to sue. A tracker is just a meeting point for computers that want to talk about the same hash value.

    The tracker doesn’t know what file is being shared (the file has the hash value supplied by the peers, but one of the features of a hash is that you can’t easily go backward from a hash to the object that it represents). Suing a tracker would be like suing a router. Both don’t care about the semantics of the data they transfer; they will transfer any kind of data for anybody, as long as it’s formatted in a particular way. There isn’t even a way for the tracker to ASK the peers what file they are sharing. Nor could the tracker become a peer and examine the file contents themselves; the peers have a .torrent file that the tracker doesn’t have, and which is required to participate in the protocol.

    A web server operator whose site contains lots of “Download this .torrent file to see Leave it to Beaver from July 5, 1964” links is a more appropriate target — particularly if the server operator had any involvement in placing those links on their server (as opposed to merely being a forum for user-contributed information).

    [Enhanced trackers that have been manually given a copy of the .torrent file can recognize which peers are using which .torrent file, and report statistics to a web interface. But this is not part of the protocol; it’s just a convenience. If doing so becomes illegal, people will run trackers that don’t have those features.]