There Is No Success Like Failure

By: Scott Bradner

Back in the 1970's TCP/IP was offered to ISO as a technology contribution but was turned down. The world of telecommunications might have been rather different if that effort had succeeded rather than failed.

This train of thought was suggested by Tony Rutkowski in a note the other day. He made the above observation and wondered how important the basic nature of the IETF has been in producing the dynamic Internet we have today. Clearly, as Tony also pointed out, any speculation on alternate futures is specious at best, but it also can be fun.

There is frequent criticism of the IETF by proponents of the traditional standards bodies. The IETF is seen by them as out of control, frequented by technonerds who, if given a phone system to run, wouldn't know how to operate their way out of a paper bag, who don't plan before they do, and who don't have specific voting during the standards evaluation process. In addition they feel that the IETF does not have any way to establish a specific membership, and also is not fair because national standards bodies are not doing the evaluation. All in all, it sounds like rather a bad place; you wouldn't want your kid to go there.

The traditional standards bodies are seen as more rule-based, less likely to be subjected to the rantings of an individual, tied into the governmental standards acceptance mechanisms, who plan before doing, and in general, somehow more refined. The kind of place someone might feel the need to wear a tie.

I expect that all of the above are right.

The ISO, for example, is a prodigious producer of standards. I've been told that on an average business day there are 6 ISO working groups meeting somewhere. They produce standards on topics ranging from bolts to networks. They have specific procedures in place to ensure fairness of process and technical competence of the resulting standards.

The IETF has about 70 working groups, most of which deal with defining specific network-based technologies for various applications. They also have specific procedures in place to ensure fairness of process and technical competence of the resulting standards.

But, what is the real difference? To quote from RFC 1726, " A major contributor to the Internet's success is the fact that there is no single, centralized, point of control or promulgator of policy for the entire network. This allows individual constituents of the network to tailor their own networks, environments, and policies to suit their own needs. The individual constituents must cooperate only to the degree necessary to ensure that they interoperate." The authors call this "Co-operative Anarchy," and this is the world that the IETF assumes.

The basic networking standards promulgated by most traditional standards bodies assume that there is structure to the management of the underlying network. X.400 email and X.500 directory standards are examples of this. The network providers provide for much more than just transport of data; they also operate services, without which the network would not function.

It is not clear what the effect of top-down management (for that is what standards like these assume) would have done to the Internet but I do find it hard to believe we would have the explosive growth in usage and applications if we had relied on the innovative skills of the telephone companies to provide us "what we need." I expect that is what would have happened if the offer had succeeded. I think I'm glad it did not.

PS - One trivia point to those who can identify the quote in the title.

disclaimer: Failure is not taught at Harvard so any implications of the benefits of failure must be mine alone.