rough consensus (was Re: "trouble maker")

James Seng jseng at
Thu Jun 26 04:35:04 CEST 2003

While we dont have the concept of "stakeholder", we have a concept of 
"running code". This is another swing factor in the consideration of 
"rough consensus".

Someone also reminded me that sometimes you need as high as 95% for 
rough consensus and sometimes as low as 65%, varies according to the 
topic and the situation.

So really, there is no any fixed methodology in determine rough 
consensus and the current process leave it to the wg chairs to determine 
"rough consensus". To balance this power, the appeal process was there 
to catch the exception case.

Since everyone has their own interpretion of what is "rough consensus", 
I can understand the resentment for those who lost out. And the wg 
chairs is in this unfortunate position of been the center of this 

But please spare a thought for the wg chairs too. They are in this "damn 
if you do, damn if you dont" position...

-James Seng

Hallam-Baker, Phillip wrote:
> I think it also depends on what the stake the parties have.
> If we have 20 people vote on an issue (lets call things by their names eh?
> it is still a vote even if the supreme court decides not to count it).
> Say people object because they want the spec to require implementations to
> use some proprietary IP they own. Pretty easy call, could be 5 objectors or
> 15 there is something going on here that is an IETF-wide issue.
> Say on the other hand you have an issue that is hard to implement. In that
> case 5 objections from non-developers might carry much less weight than 5
> from developers.
> Or say again that you have a case where cost is involved. Objections from
> five people who are unaffected by the cost issue might carry much less
> weight than five objections from people directly affected.
> But of course, nobody represents anything, nobody can speak from authority
> except those the establishment choose to recognize. So we are back to raw
> numbers.
> Going back to my case study (yes facts are sometimes unplesant things) the
> objections to the spec did not come from people with a stake in the outcome
> - except for those who had stated their objections were explicitly political
> and wanted to use DNSSEC to force some sort of change in the ICANN
> contracts.
> So what does the consensus process do except create inbuilt vetos in favor
> of the status quo?
> The fact that a spec has been undeployable for ten years is in my view and
> that of most software vendors a comment on its quality.
> As I said to Keith in private email, what you are doing here is projecting
> on a massive scale. You are affraid that the big corporations will dominate
> the process so you have rigged the rules so that they have zero influence -
> even though you depend on their market influence to propagate standards. Now
> you are surprised that we are abandoning the IETF in droves?
> Compare the decline in IETF attendance with the rise in OASIS participation.
> If the reason was the post dotcom boom recession then OASIS should have been
> affected at least as baddly as IETF.
> Contrary to the paranoid assumptions that many in this group appear to act
> on corporations do not in general want to exclude the academic and
> individual contributors from the standards process. But the way the IETF
> rules are being applied are leaving little choice in the matter.
> 		Phill
>>-----Original Message-----
>>From: James Seng [mailto:jseng at]
>>Sent: Wednesday, June 25, 2003 2:11 PM
>>To: Hallam-Baker, Phillip
>>Cc: 'Thomas Narten'; problem-statement at
>>Subject: rough consensus (was Re: "trouble maker")
>>Maybe lets put this particular case aside and look at the 
>>more general 
>>"What is rough consensus?" How rough is "rough"?
>>Unfortunately, this was not particular well-defined even tho 
>>it is one 
>>of our core principle. Could we tag this as a problem in the 
>>I remember I asked this questions before when I am a newbie 
>>in IETF. The 
>>general answer of "rough consensus" are super-super majority.
>>The general figure is >75%-80% but bear in mind that we dont "vote", 
>>absolute figures isnt the way to go. We also need to take 
>>those who are 
>>"neutral" into considering as "neutral but wont object" and 
>>"no but wont 
>>object strongly" are important swing factor.
>>If we can get a rough consensus on what is "rough consensus", then 
>>perhaps we could avoid a lot of misunderstanding.
>>-James Seng
>>Hallam-Baker, Phillip wrote:
>>>Count the votes as 12:2 or 12:9 there was a clear majority 
>>for OPT in and
>>>the 7 who said no but were willing to stand asside for the 
>>sake of forward
>>>However you count them a clear majority of the group 
>>believes that the
>>>DNSSEC specs as currently defined are broken and are in 
>>need of fixing. Why
>>>then are ANY drafts being forwarded to the IESG?

More information about the Problem-statement mailing list