+1 i.e., maybe N=0 in this case changing the basic genetics of the IETF is not warranted in this case (imo) in the past (e.g., multicast routing) we formed two WGs and the problem worded itself out (one WG produced a result) Scott On Nov 28, 2013, at 10:35 AM, Sam Hartman <hartmans-ietf@xxxxxxx> wrote: >>>>>> "Gonzalo" == Gonzalo Camarillo <Gonzalo.Camarillo@xxxxxxxxxxxx> writes: > > > Hi. I've thought about the general issue of alternative decision making > a lot, and it did come up a number of times while I was an AD, although > none with such significant stakeholder interest as this. > > The issue I'm most concerned about is confirming that there is rough > consensus to make a decision. I think any variance from rough consensus > for decision making must itself meet a fairly high consensus bar. > > "But we can't make a decision; we have to make a decision, so of course > we're going to do something alternative," you might say. > > You don't have to make a decision though. You can decide the IETF is > not the right forum for your decision. You can decide not to > standardize. You can decide to run experiments, get data, try to > implement. You can give up. > You can redefine the problem. You can for example decide to have two > mandatory-to-implement options, or require receivers to implement both > and not mandate which senders must pick. > > In many cases it does turn out to be easier to get consensus that a > decision must be made than it is to get consensus on a decision. > However without chairs being able to clearly show that consensus to > decide has been reached I'd expect an appeal and expect that appeal to > be successful. > > Next, it's strongly desirable to get consensus on your process for > making the decision. > If you can get that, or lacking that, get consensus on someone who will > decide the process (chairs or AD), I think you are coming along > reasonably. > > Obviously concerns of fairness are important. A process that inherently > advantaged one company or viewpoint would be suspect. > > I'd personally favor coin-flips, external arbitors or similar over > voting. We don't want to encourage voting because we could get into > situations where people block the consensus process in order to force a > vote. Absent things like well-defined membership and procedures to > avoid one company stuffing the ballot box, going there seems > problematic. > > Coin flips are nice. They really create pressure among anyone who > actually cares about the outcome to compromise, to explore whether a > consensus can be built. However for the frequent situations where a > decision is critical but it really doesn't matter (even if some people > think it does), they get the job done.