Re: [tsvwg] Requesting TSVWG adoption of SCE draft-morton-tsvwg-sce

[Kyle Rose <[e-mail]>]

On Mon, Nov 18, 2019 at 11:03 AM G Fairhurst <[e-mail]> wrote:

> The phrase: "is not an automatic barrier to WG adoption" - is true, but
> I would expect there to be a barrier to *immediate* adoption.
>
> Thw WG has a chartered work item, following a BOF and adoption
> procedure. The Chairs work with the AD to manage the chartered work
> items, so I expect there will be no immediate change to the milestones
> (list of adopted drafts).
>
> To clarify my own Chair position, I would not expect a WG to consider
> lightly starting a second deployment experiment in parallel using the
> same set of codepoints.
>

I'll offer a counterpoint as a former co-chair of tcpinc, the early history
of which this work area increasingly resembles.

When there are two competing proposals for the same deliverable (in this
case, high-fidelity congestion signaling) such that rough consensus does
not seem obviously forthcoming, the right approach may very well be to
adopt both and see which pans out.

If multiple experimental uses of the ECT(1) codepoint are not deployable
simultaneously, this might suggest withholding publication of any
experimental RFC making use of that codepoint until rough consensus is
achieved, which would suggest a longer period of evaluation and comparison
in synthetic environments.

OTOH, if the two experiments are deployable simultaneously (because the
consequences of conflict are not dire), perhaps an experimental RFC
outlining success criteria for evaluation of the two approaches would be
helpful in breaking the logjam.

What I don't want to see are variants of the sunk cost fallacy or
preference given to one experiment over the other for reasons other than
consensus based on demonstrated technical superiority. I've witnessed too
much of that on this mailing list over the past several months.

Kyle