On Sat, 7 May 2005, Tom Lord wrote: > >> it's not that cut-and-dried. it can be very costly to users to > >> let the market decide. sometimes the market doesn't decide, it > >> just fragments. > > > > So? > > so "let the market decide" is a lousy rule. there's no justification > for it. it's just the sort of thing that someone says when he fears > competition from a better product. I agree with Keith. But how about: "So... We all work very hard to create standards using a vendor-neutral, open process that work for many parties, not just a single vendor" > The original focus of IETF was to create and firm-up the Internet. > That war was won. The internet evolves. New standards are created. Old standards are modified. That process doesn't stop. But if you aren't interested, why are you here? What's your interest? I don't understand your point. Are you here to convince the rest of us that the IETF is irrelevant? The IETF will "end" when people lose interest in its works. > You're complaining that some application-layer stuff like IM isn't as orderly > as you'd like. Disorder isn't good for the users, either. Its not just a personal view of orderliness. And it isn't good for the market to have such unnecessary and gratuitous disorder. That's why standards of any form exist. > I don't see the connection between your complaint and the original > focus. > > Now, refining a few core protocols -- that'd be great. Trying to be > the government of all protocols -- huh? > The SRFI process, in the world of Scheme programming, seems to me the > more utilitarian approach to working on higher-level protocols: there's > nearly nothing to fight over in that process. I suspect that the architecture of Scheme and Lisp has a lot to do with this. You have a few core language constructs and everything else is built on top of that. Try to take away CAR or CDR and you'd have big problems with consensus, I suspect. Better examples is the Common Lisp/Scheme schism. There can easily be many languages, but its harder to say there will be multiple BGP or TCP variants. Some order, beyond the "you're welcome to create a code fork" is necessary when you have different pieces of hardware that have to interoperate. A program only needs its particular runtime, and we can easilly have many runtimes for different languages. If you were making scheme/lisp hardware, there would be more concern about the compatibility of language primitives. (Didn't we already have this battle with LMI and Symbolics?) So I don't think the Scheme programming analogy works. But I agree that the "consensus" is a vague term. Most of the people who don't like it are the ones where the consensus didn't go their way. In any specific case, its hard to tell whether they have a valid complaint or not. I agree that's a problem. And partly because the definition and determination of "consensus" is so vague, there is sometimes genuine cause for suspicions about motives, politics, and such. However, putting together a simple voting process won't work either. Like democracy, its just about the worst thing there is, except for the alternatives. So I think collective judgement by the WG chairs and the IAB is the only way. I think trustworthy and honest WG chairs and IAB members are critically important, and a fair complaint resolution process is also important. -- Av8 Internet Prepared to pay a premium for better service? www.av8.net faster, more reliable, better service 617 344 9000 _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf