Re: [irtf-discuss] Is there a home for Semantic Routing in the IRTF?

[Colin Perkins <[e-mail]>]

Adrian,

You sent me a proposed research group charter, along with an extensive motivation document, to review in November 2021. I discussed this with relevant members of the IRSG, and we had a call – at your request – where I gave you feedback. That feedback was that it was not appropriate to charter a semantic routing group because, to the extent that there was research being proposed, it seemed to overlap with existing IRTF groups, primarily COINRG. Much of the work also seemed to be outside the scope of IRTF or better suited to other venues.

After further discussion with members of the IRSG, and with others in the community, during IETF 112, I repeated that feedback by email on 15 November, in response to discussion you initiated on the COINRG list.

Colin





> On 10 Mar 2022, at 15:57, Adrian Farrel <[e-mail]> wrote:
> 
> Hi,
> 
> On a recent thread that I was cc'ed on, you said:
> 
>> The IRSG reviewed the proposal for a semantic routing research 
>> group, and concluded that it did not meet the bar for chartering
>> as a new IRTF group.
> 
> This came as a real surprise to me as one of the main proponents of the
> Semantic Routing work. No one ever made a formal request or proposal to you,
> and the furthest we got was an early draft of a charter to scope what we
> wanted to work on and to gather feedback from friends and researchers.
> 
> Indeed, the IRSG minutes (by the way, very many thanks for making IRSG
> meeting minutes public) don't show any evidence of detailed discussion, no
> consideration of the draft charter, and no discussion with any of the
> proponents. I did have a chat with you in early November to give you a heads
> up that we were thinking about this, and you were pretty certain at that
> point that there was a considerable overlap with COIN.
> 
> From the minutes of the 2021-10-26, I see:
> 
> | Lars Eggert mentioned a group wanting to meet. 
> |
> | Colin mentioned he thought it was one specific company asking for a 
> | research group, but unless they can bring a verified community he 
> | wasn't going to approve a group.
> 
> It's not clear from the minutes whether this was Semantic Routing or not.
> There was, I think, a lot of debate about research into Satellite Networks
> at around the same time. At the time of this IRSG meeting, you and I hadn't
> talked about the proposal, so I suppose this was an early view without any
> deep dive.
> 
> From the minutes of the 2021-11-30 meeting I see:
> 
> | Colin also spoke about the IETF 112 side meetings, one on 
> | semantic routing that proposed a research group charter that was 
> | narrower than expected. He noted he would likely not charter a 
> | new group at this time, as it looked like the work would overlap 
> | with both COINRG and also possibly PANRG. The contractual 
> | routing, semantic routing, and internet addressing that seem 
> | closely related. He suggested the proponents start with COINRG 
> | to develop the ideas further.
> 
> I like that you were surprised by the narrowness of the proposed charter
> since it was not substantially changed from what we talked about when we met
> on 2021-11-05.
> 
> I'm not worried that you recommended us to air the ideas in COIN, and this
> has helped us a lot to formulate the concepts and to attract a number of
> people to discuss their work. That it turns out that "routing is out of
> scope of COIN" (to quote all three of the co-chairs in separate emails) is
> not a surprise, although it leaves us with not a lot left to discuss in
> COIN. Perhaps the mechanisms used to programme the forwarding plane in
> support of Semantic Routing can be discussed there, and perhaps there will
> be room for work on the applicability of Semantic Routing in support of
> specific compute in the network requirements if someone is interested in
> that work. Of course, this has used up some time and energy, and arguably
> clogged up the otherwise empty COIN mailing list, but so long as we are
> gaining understanding, that is not a problem.
> 
> But I would note that you were not approached with a formal request for a
> new RG. Rather, we put together a talking point for discussion to see what
> reaction we would get and what would be needed to take it further. This
> seems consistent with the level of debate recorded in the IRSG minutes, and
> it doesn't seem that there was discussion about a proposal to form an RG for
> Semantic Routing, an opinion expressed by the IRSG, or any exploration with
> the proponents of what they wanted to do. But it does all seems a bit
> pre-emptive and closed.
> 
> So, just as a matter of getting the record straight, can you clarify a few
> points:
> 
> - Was the discussion from the 2021-10-26 meeting quoted above about Semantic
> Routing?
> - If the 2021-10-26 discussion was about Semantic Routing, what specific
> company were you implying I was associated with?
> - Did you at any time suggest to any of the proponents of Semantic Routing
> that unless they can bring a verified community you were not going to
> approve a group?
> - What discussion of the Semantic Routing proposal did the IRSG have, what
> input did they seek from the proponents to clarify the proposal, and how was
> the conclusion that the bar had not been met communicated to the proponents?
> 
> Lastly, now that it is clear that a substantial part of Semantic Routing is
> related to routing, and reiterating that we are looking at packet routing
> within "limited domains" as well as for wider connectivity, do you have any
> further advice about where our work sits within the IRTF/IETF?
> 
> Many thanks,
> Adrian