The IETF's problems

[Keith Moore]

> > because, left to themselves, WGs will happily create things that
> > don't satisfy various requirements (such as security), or which
> > interfere with other parties' interests (like zeroconf or nat WGs
> > trying to fundamentally change the way IP works), and then the WGs will
> > get very annoyed when they only learn *after* they think they're done
> > why IESG (for quite valid reasons) refuses to approve their documents
> > and their work isn't even fixable.
> 
> I don't think the annoyance can be avoided in cases like this. All this 
> work just to move the moment when this happens around doesn't seem like 
> a very effective use of time and energy.

so if you're trying to go somewhere, you'd rather just start out travelling in
a random direction, and wait until you think you should be there to find out
that your course is 90 degress off?

> And does the current way of doing this really work?

not as well as we'd like.  that's one of our biggest problems.

> And what's the problem with zeroconf anyway?

it breaks widely-held assumptions about the scope and stability of IPv4
addresses.

> > what I don't get is why anyone could think that WGs should be allowed 
> > to make standards without some broad-based oversight.
> 
> The current way of doing this doesn't work very well. If someone can 
> come up with something better, that would be great. If not... then we 
> need to be more radical.

so if banging your head against a foam pillow doesn't work, try banging it
against a brick wall?