Peer-to-Peer Makes the Internet Interesting Again -- Andy Oram reports from the O'Reilly Peer-to-Peer summit, where people from groups as diverse as IBM and Freenet talked about the technology's potential and limitations.
Yesterday, at the first meeting of their proposed Peer-to-Peer working group (attended by over 300 people) Intel proposed an organization for the group that was shocking in its distrust of the very decentralization that peer-to-peer is supposed to enable.
The proposed structure is best described as "pay to play", an industry consortium model in which the amount of money you contribute to the organization determines the level of your influence. A seven member steering committee, defined not by technical excellence or vision, but by the $25,000 admission fee, will be the final arbiter of standards. For $5000, you earn the right to chair a technical committee and vote on technical issues. For $500, you may participate in a committee by invitation, attend group events, and review (but not vote on) draft standards. In a sop to participation, two of the steering group members would be elected from the $5000 members.
I stood up in the meeting yesterday and asked for the sense of the group whether this proposal was going in the wrong direction, and was met with thunderous applause. Intel agreed to go back and think some more about the structure, and came back after lunch with a new idea: that they'd accept submissions about alternate structures, and would decide which of the proposals might be best.
This response still showed Intel's top-down approach at its worst. The proposals would be sent in to Intel; it wasn't at all clear who would be making the decision, or how open it would be. Given the general tenor of the meeting, which consisted of presentations (mostly by Intel), with very little time for dialogue, I'm not optimistic that Intel has the instincts that are required for defining open standards.
I'm not interested in sending in a proposal to some invisible committee who will decide on its merits and whether or not anyone else will see it. I want to make my proposal in the open, where everyone can see it.
First off, let me clarify my objections. I have no problem with the idea of asking sponsors to fund the organization, at the level that Intel proposed. I have no problem with a steering committee, committee chairs, or other structural elements. The problem I have is the tight linkage between the two. Funding should be decoupled from technical decision making.
The IETF, which is a fully participatory organization that should be used as a model for the P2Pwg, has a steering committee, working group chairs, and all the kinds of structures that Intel says are needed to ensure that discussions are driven forward rather than devolving into anarchy. However, the leaders emerge through participation rather than by appointment.
For a good description of the IETF standards process, see Scott Bradner's article, The Internet Engineering Task Force, which was published as part of the book Open Sources.
Here are a few excerpts from Scott's piece:
IETF standards are developed in an open, all-inclusive process in which any interested individual can participate. All IETF documents are freely available over the Internet and can be reproduced at will....
The IETF can be described as a membership organization without a defined membership. There are no specific criteria for membership other than to note that people and not organizations or companies are members of the IETF. Any individual who participates in an IETF mailing list or attends an IETF meeting can be said to be an IETF member.
At this writing there are 115 officially chartered working groups in the IETF. These working groups are organized into eight areas: Applications, General, Internet, Operations and Management, Routing, Security, Transport, and User Services. Each of the areas is managed by one or two volunteer Area Directors. The Area Directors sitting as a group, along with the chair of the IETF, form the Internet Engineering Steering Group (IESG). The IESG is the standards approval board for the IETF. In addition there is a 12-member Internet Architecture Board (IAB), which provides advice to the IESG on working group formation and the architectural implications of IETF working group efforts. The members of the IAB and the Area Directors are selected for their two year terms by a nominations committee randomly selected each year from among volunteers who have attended at least two out of the previous three IETF meetings....
One of the principal differences between the IETF and many other standards organizations is that the IETF is very much a bottom-up organization. It is quite rare for the IESG or the IAB to create a working group on their own to work on some problem that is felt to be an important one. Almost all working groups are formed when a small group of interested individuals get together on their own and then propose a working group to an Area Director.
Anarchy, you might say. Bob Knighten of Intel said as much when I suggested the IETF model for the P2Pwg. But keep this in mind: the IETF has been one of the most successful standards groups ever created, and is responsible for most of the core standards on which the Internet depends. Again, let me quote Scott Bradner:
The Internet is growing faster than any single technology in history, far faster than the railroad, electric light, telephone, or television, and it is only getting started. All of this has been accomplished with voluntary standards. No government requires the use of IETF standards. Competing standards, some mandated by governments around the world, have come and gone and the IETF standards flourish. But not all IETF standards succeed. It is only the standards that meet specific real-world requirements and do well that become true standards in fact as well as in name.
The IETF has significant participation by large companies, but is not controlled by them. It is a technical meritocracy focused on interoperability.
In my opinion, Intel squandered the opportunity to kick off a real working group process. What should have happened yesterday was a proposal for a series of committees (what the IETF calls working groups). Some possible committees might have been suggested--for example, security, interoperability between distributed computation applications, search and resource discovery, and so on. There should have been time for the group to propose additional or alternate groups. Those that seemed most important would be selected by acclamation--or simply by the BOF test, in which a space is set aside for each meeting, and the ones that get attendees are clearly the important ones.
When those individual working groups got together, they should have been provided a few simple rules:
The number of subgroups would determine the formation of the steering committee. If the number is manageably small, the chairpersons of the various subgroups would meet collectively as the steering committee. If the number is large, the chairpersons could elect a steering committee, or, as in the IETF, elect a nominating committee, who would propose members to be elected by by the full body.
There is a rich history at the IETF, and many documents that describe its standards process. (RFC2026 is a bit dry but lays out the process in some detail.) While the IETF has its warts, including its "old boys network" of key contributors who've become enshrined over time, it is still the standards body that provides the best model for the kind of participatory standards making we need in the peer to peer space.
Though I have argued here that Intel put the wrong foot forward, I am convinced that they have their heart in the right place. This was not a misstep conceived in malice or an attempt to corner the market. Intel has played a magnificent role in bringing visibility to peer to peer, and in calling for early work on standards for interoperability between peer to peer applications. They now need to take the next step in helping to shape a forum that itself has many of the characteristics of peer to peer, in which structure and standards emerge rather than being imposed. Intel's role as the convener of the working group is a great one. Now, what they have to do is to let go, and let the working group itself decide on its organization.
I'd suggest very strongly that the next working group meeting, proposed for January, be organized along the lines I've outlined here. Because we can't wait till then to take the next step, the process of deciding on technical committees will have to be done on a mailing list.
I'd suggest that Intel's p2pwg mailing list be used as the initial forum, that they solicit proposals not for the structure of the group, but the structure of the problem. What are the areas that need working group committees? Where would this emergent cluster of technologies benefit more from interoperability and common standards rather than competition?
If Intel would prefer not to take this road, those of us who are interested in working participatively on the development of p2p interoperability standards will need to find another forum.
Tim O'Reilly is the founder and CEO of O'Reilly Media, Inc., thought by many to be the best computer book publisher in the world. In addition to Foo Camps ("Friends of O'Reilly" Camps, which gave rise to the "un-conference" movement), O'Reilly Media also hosts conferences on technology topics, including the Web 2.0 Summit, the Web 2.0 Expo, the O'Reilly Open Source Convention, the Gov 2.0 Summit, and the Gov 2.0 Expo. Tim's blog, the O'Reilly Radar, "watches the alpha geeks" to determine emerging technology trends, and serves as a platform for advocacy about issues of importance to the technical community. Tim's long-term vision for his company is to change the world by spreading the knowledge of innovators. In addition to O'Reilly Media, Tim is a founder of Safari Books Online, a pioneering subscription service for accessing books online, and O'Reilly AlphaTech Ventures, an early-stage venture firm.
Return to the P2P DevCenter.
Copyright © 2009 O'Reilly Media, Inc.