Log in

No account? Create an account

The IETF and Non-violent Social Change

The usual storm is brewing about process reform in the IETF. I keep getting stuck in the middle because I'm too emotionally involved to not care about the process issues and because I try to do a good job of things I care about. I had an intuitive leap Monday night around 4 AM. A few months ago I was comparing IETF process issues to comments from a class Shava was teaching about non-violent social change. I noticed similarities. Monday night I actually managed to understand how that fit the IETF at an intuitive level. This entry is me brainstorming about what is needed and how to tell people. Your comments are welcome, but I'm putting the rest behind a cut to save those who don't care. So, here are the symptoms that the process reform effort is having trouble
  1. There is consensus that change is needed. However there doesn't seem to be consensus around any particular change or on how to get consensus. Some argue that since we cannot get consensus we should go make changes anyway because we never will get consensus (or consensus against).
  2. People seem to want to make radical changes. It's kind of funny. If we proposed making radical changes like replacing TCP with some new transport with a flag day, it would be a complete non-starter. However there are a number of proposals on the table that are fairly radical changes to the IETF. There's significant concern though about breaking the IETF. We do approve around 300 documents a year; people do depend on our output. It is possible to make things worse. However we need to make things better, and it may be that significant change is required to accomplish that.
  3. It feels like people are attacking each other not working on problems. Yes, comments are phrased in terms of problems—things like "The IESG is not being constructive," or "the WG documents aren't of sufficient quality when they come to the IESG." Still if you're the IESG and you've spent hours trying to balance concerns, the blanket statement that you are not being constructive stings. Similarly working hard on a document and getting the general impression that the quality isn't good enough makes you wonder whether you are being held to reasonable standards.
  4. People are not building up shared understandings.
So, what do I want to happen? Well, the first thing is I think the IETF needs to be treated as a full standard: it is a working system and we are going to have a high bar for any radical changes to it. As a practical result that means we're going to need to run experiments and to propose evolutionary changes rather than redoing things from scratch. If you can propose a reform as an optional experiment run by the people who are interested in that experiment rather than a change everyone has to deal with, it will be easier to get consensus. Now, you need to be careful that success of your experiment isn't used to claim things your experiment does not support. For example, an experiment with a friendly group of dedicated testers showing that you can approve documents faster doesn't show that you will not reduce document quality in the general case. People need to be able to trust that your successful experiment will not be used as a way to force a global change without eventually addressing their concerns. Experiments should be a way to collect information and build experience not a way to subvert the need to build consensus.

Next, we need to build a culture of working together to build mutual understanding of concerns and goals. We need to show each other respect. Convincing everyone of this is impossible. I hope that convincing a core group of people to work this way in a very visible manner is doable though. I believe that building mutual understandings will be more effective than what we're doing now. So I hope that the core group of people exhibiting what I will call positive behavior will be more effective and that the community will notice they are more effective. I then hope that this will encourage people to adopt these behaviors.

The final question is what to do to bring about these desired outcomes. I'm still mulling this over. I'm thinking of trying to start by writing a message comparing the IETF itself to one of our more mature protocols and discussing what change strategies work when changing a mature system. I'm not entirely sure how to bootstrap a core group of people that is actually trying to build mutual understanding.



If you could get people at IETF to behave, I bet my mom would be eternally grateful :) (And in fact, along that front, you might want to brainstorm with her about how to proceed.)