MAJOR ISSUE: "Concentration of power"
narten at us.ibm.com
Fri Jun 20 17:57:05 CEST 2003
A few thoughts about trust...
One thing IMO that complicates the discussion about trust is that some
of the postings could be read to imply "trust in WG chairs" is an all
or nothing thing. I.e., "we don't trust chairs" (by implication for
For me, trust is earned. That goes for everyone, not just for "some"
folk, or for WG chairs. Even fellow IESG members have (or have had to)
earn my trust. And even then, one trusts certain things, not
necessarily everything. I.e., I trust person A in area X, but not
necessarily in area Y. And this isn't necessarily bad or an indication
of flaws. People have different strengths and weaknesses, and one
places trust accordingly.
> - My understanding is that in the IESG, a shepherding AD does a
> write-up of each document that is to come up for ballot. Some of the
> other ADs often just read the write-up, trusting it to be a fair
> evaluation of the document. ADs trust each other's opinions of the
> technical quality of the document, often voting "no objection" to a
> document that's out of their area of expertise because they trust
> that the shepherding AD or other ADs who are more familiar with the
> document. Assuming my perception is correct of how things work on the
> IESG, I think that's all good stuff. However, working group chairs
> are *not* routinely expected to do something like an AD write-up of a
I've never asked or expected my WG chairs to do the writeup. I guess
partly this is for historical reasons (it wasn't done that way when I
started). Also, I don't really see this as a place where offloading
the chairs gets much bang for the buck.
> In fact, I've been directly told by some folks who've been
> on the IESG that the idea that a working group chair could do
> something like a good technical write-up that could be trusted by the
> IESG is absurd.
I think one would need to look at this on a case-by-case basis. And
it's not just about "trust" in some good/bad sense, it also has to do
with experience. Better writeups come with experience, not on the
first try. And whether this is something that needs to be pushed down
> (I have been told by some IESG folk that a chair
> write-up is a really good idea.) That means to me that the IESG does
> not, at least as a general rule, trust even working group chairs, not
> just random newbies. If the view that chairs cannot be trusted is
> widespread, that's a really bad thing.
To me, it's not really about "trust". I'm just not sure why the WG
chairs should be doing this (except maybe under the "offload as much
work as possible" theory). In practice, doing the writeup doesn't take
that much time and I suspect that if I asked the chairs to do it, I'd
end up spending as much time in the end just in iterating on the text
and going back and forth. Often (and this is _really_ an important
point), there is no time for that, because I'm trying to get the
writeup finished in time for a deadline in order to get the document
on the agenda for the next telechat. In such cases, one may not have a
few hours or days to iterate. For me, this just isn't a place where it
seems to make sense to try and off load to the chairs that
much. Others may feel differently.
> - I've been in multiple working groups in which I've heard things
> from the chair like, "That's never going to get by the IESG." (When I
> was still young and impressionable, I've heard such words come out of
> my mouth.)
I've heard that said too. And I've also too often heard things said
that are a misintepretation or oversimplification of what the IESG (or
an IESG member) has actually said. And I've also heard random people
_assert_ things in the name of the IESG that are just plain
rubbish. Questioning is good. ADs need to be able to explain their
reasoning to everyone. When that is not happening, and folks do things
because "the AD just says so", the system is broken and we all
lose. Note though, that this point is also touching on the same point
discussed earlier about cryptic one liners vs. explaining points in
detail so everyone can understand.
But it's even more complicated sometimes. In one WG document recently,
I raised issues on a document, and the author ended up responding by
saying the equivalent of "I'm going to take this out, because that's
what it will take to get it through the IESG". This is despite my
raising what IMO were real technical issues, *and* a willingness to
talk about them and work through them. But there was simply (from my
perspective) an unwillingness to discuss the actual issues. I really
hate that when it happens (and I pushed on the WG to remove the text
because it was not needed rather than just because I was asking for
it). But I'm sure that for at least some people, the message that will
be relayed will be "removed to satisfy AD" rather than "because it was
the right thing to do". I don't know what to do about that.
> Often, I don't think these words are used just to pass off
> blame onto the IESG for a technical issue that the chair thinks is a
> dog (although that's clearly one of the uses). I think the chair
> often means, "No matter how important we think this feature is, no
> matter how technically sound our argument is to have our protocol do
> this, the IESG (or some particular AD) is going to balk at this and
> it will take us forever to make them happy, so let's do it a
> different way that they won't complain about." That is, working group
> chairs do not feel like they and their working groups are trusted by
> the IESG to make good technical judgements, and they equally feel
> that they don't trust the IESG to listen open-mindedly to them.
I'm not going to deny the above happens. But there are two sides to
the coin. Sometimes, it's just the easiest way (unfortunately) to deal
with an issue. As a (real) example. One of the too-common themes I've
heard before in WGs is "TCP is too much overhead, we need to invent
our own reliable datagram/UDP transport protocol that is more
efficient than TCP". The above comment is just the simplest way to
deal with issue (in one sentence) when this topic comes up after a
while. The real answer is longer and complex, and it was (in cases I
can think of) extremely hard to get buy-in from those advocating using
something other than TCP.
So, when sound-bite "IESG" reasons are given for not doing something,
we also need to look at the specifics of the issue, because sometimes
what appears to be a "because the AD says so" is actually a bit more
When in doubt, question further.
More information about the Problem-statement