One of the Internet’s chief virtues is inefficiency.
“Best effort” packet routing – as Jonathan Zittrain describes it, the “bucket brigade” where each link in the network tries to pass packets to the next hop, but without guarantees – is less efficient than a protocol that seeks to guarantee transmission and thereby minimizes bandwidth used to communicate. Stateless protocols, such as HTTP, can be less efficient: the server doesn’t keep information about my client or its state and so, by default, each request is a new session. For those of us with a penchant for law and economics as an analytical tool, this state of affairs seems initially sub-optimal.
Yet inefficiency means that data is redundant – cached, preserved, more readily accessible. Years ago, in my life as a sysadmin, I managed to delete my department’s primary database while “cleaning up” our servers. Fortunately, the Lotus Notes replication model means that databases are typically “replicated” (copied and synchronized) widely within a network; I managed to find a nearly-up-to-date replica on a server in Singapore. Inefficiency removed a single point of failure. Many arguments favoring network neutrality emphasize inefficiency’s benefits: rather than tune transmission for high-priority or low-latency applications, “stupid networks” preserve flexibility.
Arguably, there are technical zones where the Internet should be less efficient. One of the characteristics of e-mail that makes spam cheap, and potent, is that a sender can transfer a single copy of a message intended for many recipients in a domain. The receiving mail server will helpfully copy that single copy into each recipient’s mailbox. Require a less efficient model – for example, one message per recipient – and spam’s economics shift dramatically.
My tentative conclusion here is that we should not set efficiency as a goal for legal or technical regulation of Internet-based information; in fact, we should be prepared, even eager, to embrace inefficiency. Hidden virtues are virtues nonetheless. But I’d love to get your thoughts (including, possibly, whether I’ve simply mangled the definition of efficiency!).