>>>>> "Scott" == Scott Kitterman <deb...@kitterman.com> writes:
Scott> On Monday, January 07, 2019 07:06:28 PM Russ Allbery wrote: >> Miles Fidelman <mfidel...@meetinghouse.net> writes: > On the >> other hand, the IETF seems to do just fine - with a much larger > >> base of participants, and a lot more room for discussion and >> debate on > contentious issues. Global infrastructure, with >> distributed ownership, > lots of stakeholders, all held together >> by agreements, with the decision > processes open to pretty much >> anybody who shows up. The process puts > pretty much everyone >> else to shame - with lots to be learned from it. >> >> Speaking as someone who is a listed author on three published >> RFCs and chaired one IETF working group, I will take Debian >> process over IETF process any day, and find your description of >> the IETF pretty entertaining. :) >> >> Also, please note that many IETF participants are paid as part of >> their job to participate in the IETF. (We keep coming back to >> that.) That's true of some Debian contributors as well, of >> course, but I strongly suspect the percentage is lower. Scott> Similarly here (also three RFCs, but never chaired a working Scott> group). Scott> The IETF rough consensus model is very useful in many Scott> circumstances. I've used it successfully in multiple Scott> settings outside the IETF to great success in both moving Scott> technical work forward or driving decision making in a closed Scott> group to closure. It's not relevant to the problem a group Scott> like the Debian tech ctte has, however. Scott> Groups like the tech ctte have a different problem than an Scott> IETF working group. They have to make final decisions on Scott> things that affect the project as a whole, many of which are Scott> Scott> I'll also remind you that the IETF process as a whole is not Scott> whoever shows up. IETF working groups and IETF last call are Scott> open processes. IESG decision making is not. You can have Scott> all the working group consensus you want, if there are Scott> uncleared discusses against your draft, it's not moving Scott> forward. If you want a comparison, the tech ctte is a lot Scott> more like the IESG than an IETF working group. I've served on the Debian TC, I've served as an IETF working group chair, and I've served on the IESG. I think I have a fairly good handle on the differences between the IETF and Debian processes. The IETF process is good for developing a consensus where you want to focus on technical quality and where you have the right stakeholders as motivated participants. It requires a certain familiarity with consensus building to avoid a number of pitfalls. IT IS NOT TIME BOUNDED. It's great for situations where you are more concerned with the right decision than concerned with a decision within a particular time line. There are ways that you can try to control the time the IETF process takes, and it's even possible to do that if you can get a consensus on what the timeline is and on the technical tradeoffs that are in scope to achieve that timeline. Debian did not meet those conditions for the init system decision by the time it came to the TC. Debian had already done a lot of consensus building. We understood the scope of the problem, we understood some of the complexities surrounding having multiple init systems. TC members did do some additional excellent work curating and summarizing that knowledge (I was not on the TC at the time but was following the discussion). A consensus process would not have achieved the goal shared by most of the project of having a decision in time for the jessie release. I am unlikely to contribute to this thread again; like Ian I think init systems are off topic. --Sam