On Sat, Sep 25, 2021 at 10:02 AM <ned+ietf@xxxxxxxxxxxxxxxxx> wrote:
I'm afraid I have to agree with Bron on this.
> On Sat, Sep 25, 2021, at 12:56, Brian E Carpenter wrote:
> > 3. I do object strongly to classifying "Last call announcements for I-Ds"
> > as non-important. They are such a fundamental part of the IETF process that they really must go to everybody, and specifically to everybody who is
> > *not* in the WG concerned. In fact, this would amount to an end-run around
> > RFC2026, for standards track and BCP drafts.
This aspect of the IETF process was in its death throes back in 2000-2004 when
I was on the IESG and should have been pronouced dead a few years after that.
Not only was any sort of truly comprehensive cross-area review becoming
increasingly difficult due to growth in technical complexity, comments when
made were starting to be met with increasing hostility. (And while some of the
latter was pure NIH, a lot of it was reluctance to spending the time to try and
explain things requiring a deep understanding of both the technology as well as
how we got to this point to people who, expertise in their own areas
notwithstanding, had no real familiarity with the subject matter.)
<snip>
Ned leaves out one of the main causes of hostility which was the game of trying to force group X to do the job of creating a reason to make use of the work product of group Y.
There are still people in the DNS world who believe that creating new resource records is the way to solve problems. And the people with real world experience of getting records deployed can assure them it is not. It took me over a decade to get CAA adopted and that and SRV are the only two widely supported extensions to the DNS canon since IPv6.
Cross area review is hard because most of the areas have already become far more complex than most people can understand on their own. There are very few people who are comfortable with complexity and it is very common to see people demanding simplification of the requirements as a gating condition to start work. Which then leads to people pointing out that a show stopper requirement has been ignored when the drafts reach last call.
In the days when people met in person, there was the idea that merely having all the areas meet in the same time and place would cause expertise to cross over through some sort of osmotic process. I don't think it did.
There is also the issue of gatekeeping and ring kissing. Very few areas are interested in collaboration except on very specific terms.
If we are going to get anywhere with cross area, we are going to need to approach the problem in a different way. I have not considered routing since my undergraduate thesis and that was message routing within a MIMD computer system. So there is information I really would like to know in order to be able to understand BGP security. Specifically, by what processes are bad routes detected and eliminated?
Nobody seems to want to tell me which suggests I might be correct in suspecting the answer is that they aren't and we have open loop control. The routing layer seems to be assuming that this is handled at the transport layer but if you look at how TCP works, that obviously can't happen.
The current system isn't working. If we want to have cross area collaboration, we have to address that as an end in itself. The IAB technical plenaries once played this role but that was probably the wrong forum.
What would probably work better would be for the IAB to ask experts in the various areas to produce presentations that tell people in other areas what they need to know. Since we are working remote these days and likely through 2023, podcasts are probably the ideal medium. This could be followed by a discussion session,
Here is my contribution to the effort, a complete intro to the current canon of cryptography:
That is probably at a more basic level than most folk in IETF would be interested in. The successor course covering threshold cryptography is probably more interesting. I will be releasing that fairly soon.
