>>>>> "Eliot" == Eliot Lear <lear@xxxxxxxxx> writes:
Eliot> Sam,
Eliot> On 3/4/13 6:34 PM, Sam Hartman wrote:
Eliot> We're here because of the extremely specialized nature of
Eliot> transport. PhDs who specialize in it have gotten it wrong.
Eliot> One such person drove Van Jacobson into the field, as I
Eliot> recall. There are very few people who get it right. And yet
Eliot> it's so close to the waist of the hour glass that it's
Eliot> critical to get right. Security has a lot of visibility and
Eliot> so it will never have this very same problem.
I absolutely agree that there are few people who can design certain
aspects of transport protocols.
(I'll note that security has this problem too: designing crypto is
really hard; I wouldn't be too quick to be sure that transport is so
much more difficult than the hardest problems of other areas.)
Fortunately, an AD need not do all the work in their area; they only
need to review it.
The entire IETF is founded on the idea of consensus. Central to that is
the idea that we can get together as a group and by doing so we'll come
up with better specifications. Not every person will be able to design
the inputs to that process: new proposals and discoveries of problems in
existing proposals. Some aspects of that really do require expert
knowledge.
my claim is that the AD skill set should be focused around evaluating
these inputs, coming up with an opinion, and explaining that opinion to
others. I don't believe that reviewing internet-drafts in transport,
reviewing reviews of thoes drafts, evaluating whether enough review has
happened, making an informed opinion about issues that were raised and
explaining that opinion to the community requires the same level of
expertise in transport as designing TCP. It does require significant
experience, both technical and management.
I stand behind my original comments.