[Accountability-tf] Notes from Accountability BoF at RIPE 76
Antony Gollan agollan at ripe.net
Fri Jun 1 13:31:45 CEST 2018
Dear task force members, Here are the notes from our BoF at RIPE 76. Cheers Antony +++ RIPE Accountability Task Force BoF - RIPE 76 14 May 2018 1. What are the RIPE community's values? Randy Bush said the reason he participated within RIPE was for two things: stewardship of the address space, and agreement among operators who were competitive in business but who still needed to cooperate to run the network. Athina gave some background of the need for accountability in the wake of the IANA transition. The RIPE NCC and other RIRs had showed that they were accountable by publishing an overview of their corporate structures and transparency of processes early into the ICANN transition and it had resulted in a good outcome. After the IANA transition, the focus of the discussion had shifted to the communities, such as RIPE. The question being asked was how accountable were these forums/communities? The task force had looked into this and found that it was accountable. But a lot of faith was placed in individuals – what happened when they started to leave and new people came in to the community? She said it was about values – by developing something that they could agree to as a community. Daniel said they had inherited ICANN from the government crowd and the people who wanted to influence communities by looking for levers. So they decided to own the process. That was the short description. Regarding values – their goal was to be a community where operators came together to coordinate. These people were often competitors. So they wanted to bring these people together so they could keep talking with one another. Motherhood and apple pie (open Internet, etc.) was never part of their values - at least not directly. They had consciously tried not to make these kinds of statements. The values that came to mind – openness, equal access, transparency. These seemed to cover most of it. From that followed how they structured themselves. There had been an evolution from the early days when it was very loose, with no rules – where they’d said “we'll just talk.” But then they gradually formalised over time. When he thought about values, he thought the ones he’d described were all they needed. David Farmer said he’d heard a couple of things - and “accountable to whom?” came to mind. He thought they were pretty accountable to themselves, but not accountable to the general public and maybe they could improve here. He said some of this stuff came from the fact that they were competitors – they needed to avoid creating suspicion. Randy said he thought they were part of an ecosystem and they should let protocol design mostly lie in the IETF, with the community taking a baseball bat to them when they did things that weren’t useful or needed. He thought the same could be said for the role of ISOC – from Vint Cerf’s “the Internet is for everyone”. They were about the responsible operation of the Internet – not the socialisation of it or the promotion of it. It wasn’t that these were not useful things. They were the administrative network information centre - the RIPE NCC as an RIR was the most expansionist. But the directions that it had expanded were operationally useful. And they were about operating and administering the Internet. The job of operating it was hard enough. Daniel supported this statement. Salam Yamout said from her perspective coming from Lebanon – she liked open and transparent. This was hard to understand at first, but she liked it very much. The fact that the Internet was operated by its users was quite good. She liked the idea of doing some kind of “who am I, what am I here for?” But maybe not pushing it too far. It was also about legitimacy as well as accountability. By them having a minimum of rules, maybe the whole exercise was about writing them down. They needed to be more cohesive, because the day of the questions would come and they would need to prove they were legitimate as a community. Shane Kerr said RIPE historically had a role in the Internet. People who weren’t operators and participated in the community also obviously got something out of it and contributed to it. One of the issues was that RIPE was whatever it wanted to be – with Open Source and IoT opening recently as WGs for example. There was only so far you could go in terms of defining RIPE’s values. And he said he’d be interested in seeing a scope of RIPE. Alexander Isavnin said he would like to remind everyone that RIPE was created to support global IP connectivity. And all it's actions and WGs and meetings and even the RIPE NCC were actually there to support this. They needed to explain this to the broader world, especially because the Internet had become a significant part of modern life. Also, they needed to explain that RIPE was the best at doing this IP connectivity support, to counter anyone else who wanted to do the same job at a local/UN/country level. Daniel said the RIPE Terms of Reference (ripe-001: https://www.ripe.net/publications/docs/ripe-001) had been developed years ago and he thought this was still as good as anything they could come up with today. He said they had carefully considered the wording in this document. These were the current terms of reference, so if they wanted to change these, they would need to get a community consensus. 2. How do we define consensus? Randy Bush said there was an excellent article a few years back about how consensus was subject to both conscious and unconscious attack (such as DoS attacks): -Article is here: https://www.jacobinmag.com/2015/05/consensus-occupy-wall-street-general-assembly/ Anna Wilson said she wanted to go back to first principles. They used consensus in RIPE not because they were hippies – but because this was how the Internet worked. The legitimacy of ICANN came from the fact that they all agreed to use the same Root. If a bunch of people don’t agree, it just won’t happen. It always came back to what could you persuade people to change. It bounded the question nicely – because whatever mechanism you had needed to go through that check. Peter Koch said he was trying to respond a bit to Anna. They could have week-long discussions on consensus. Looking at the development of policies was already available. While Anna had said people “voted with their feet” in terms of the Root zone, that was the nuclear option. They were trying to describe what took place at the meeting – which could easily devolve into mob rule and social media-type decision making. Consensus was about a process, it was not something that took place at a single moment. And some things couldn’t be decided by consensus and thinking in terms of +1s and -1s could just turn to voting. He agreed with first principles – but explaining how the decision-making of the RIR community around policies was legitimate was a tougher problem. Lars-Johan Liman said this consensus thing carried some weight. But if you were going to try to defend processes that you had defined through consensus-mechanisms, you needed to show that there was some basis for these decisions and that they were made through consensus. From working on the IANA transition, he was concerned about how governments saw these kinds of processes – because they liked having power and were used to having responsibility. In order to make that work on both sides, there needed to be channels between both sides to convince the other that the approach works. There were people outside that wanted other mechanisms other than open/transparent and bottom-up and had power to change things. Nurani said it was useful for newcomers to have these kinds of discussions, but they should be careful not to fool themselves that if they came up with the perfect discussion they could sit back and be done. She liked Anna’s comments around what made good operational sense. The reason she always cringed when she saw other communities do variants of voting/temperature taking was that it ultimately didn’t result in good policy. If 50% of the room agreed and 50% didn’t – that didn’t guide her in any way. So, to her, consensus was what made good policy development and good operational sense. Randy said they shouldn’t confuse how they negotiate with how they decided. Having a legitimate decision-making process was important, but then again they could legitimize anything – even emperors if they wanted. Ruediger Volk took Nurani’s remarks to mean that quality counted. Quality was easily lost by people who only looked at processes. He appreciated the work that went into the RFC that described the IETF’s rough consensus process. But for external accountability, having well-defined processes was important. Filiz agreed that consensus could be defined as a process and decision making could be described separately. But a third component that was necessary was values. You had to get that all described well. 3. What is the role of chairs Hans Petter Holen said he wanted to apologise for the idea of taking a bottom-up approach for replacing WG Chairs. He thought maybe it wasn’t worth having different approaches for each WG. William noted that he himself been selected as Database WG Chair out of a hat, because of the process in that WG. Hans Petter said you wanted a process so you could explain how something happened. “because no one objected’ worked well when you were 20 people – but it didn't scale to a community of 800 quite as well. An audience member said that not having a replacement procedure for chairs was a problem if the RIPE Chair was captured. He thought they had good people in these positions, but it would be good to have something in place that made it a bit more robust and current for the times so people could have faith that it would be the right person performing this role and doing well for the community. 4. Capture Nurani said they shouldn’t take themselves too seriously. She wasn’t too worried about someone taking over the role of IoT chair. They would survive and it would be worse for the person involved. At ICANN they had put these powerful people at the top – and then spent a year working out how to remove them because they had too much power. They should avoid that as a community by not giving people not too much power. And because no one had too much power she wasn’t worried about it. Lars said he was worried about becoming irrelevant. They should maybe include something explaining why these positions were relevant/why the community was relevant. Randy said the kind of capture he was worried about was by process geeks who wanted to document everything and have it all ruled, and the space-cased hippies who just wanted someone else to do it. And as individuals in the community each of them was somewhere in-between. Filiz said capture sounded quite scary – when it really wasn’t. She said there wasn’t such a separation of powers as William had alluded to – someone could be in a WG announcing consensus calls, and could also be on the RIPE NCC Board at the same time. At the end of the day, it was about having things somewhat documented and having one process around things like how a WG named its chair. 5. How does community retain trust over time Shane said he’d been working at ARIN many years ago when it became separate from Network Solutions. At the time they’d been struggling to build a community. One of the things that ARIN looked towards RIPE for was that they wanted the sense of community that RIPE had – and John Crane had said the best way to get people to trust you was to be trustworthy. He thought they could do the right thing and people would recognise that. It was easy to lose that, but he didn’t know what else they could do. He thought they could be friendly and open, and as long as there were many newcomers, they would be sharing these values with them. Daniel agreed with Shane. He thought they could integrate new people. He thought the charter of the task force was to produce something that explained this for newcomers. Ruedieger asked whether divergence of the RIPE NCC community and membership could come into play. He thought right now the community and the membership were about an order of magnitude apart. He guessed that a lot of the members were members not because they wanted to join the community – so you couldn’t expect them to subscribe to the values of the community. The trustworthiness of the community would ultimately be about what the organisation had in its executive branch. Randy agreed with Daniel’s point that they should document the way they did things. But he would like to subtext that with the idea that it could be more like a history or an informal explanation like the Tao of the Internet – to give people the feel of that.