I have tried to represent everyone's opinions, including my own, accurately; but, my primary focus is on the competition of ideas, rather than the competition of people, and I have sacrificed detail in the latter for clarity in the former when necessary. No doubt I have also many times failed to recognize when one message prompted another, but I hope that I have written most of the obvious and consequential ones. Order of events is confused by the fact that many people reply to a discussion after it has concluded and a new one has taken its place; I have tried to balance clarity and accuracy when dealing with these cases. I add notes to moves of hours or days, but generally not for those of shorter intervals of time.
I have omitted:
(Info here is from jodizzle's log.)
Fusl and JAA agree with each other about how bad EFNet is, the primary item being Fusl's complaint of having to have multiple connections in order to be in a large number of channels at once. JAA tells the story of "how our presence on hackint came to be":
a number of people were annoyed by EFnet's issues and started to brainstorm about other networks. Various suggestions were made, and hackint came out on top due to things mentioned earlier elsewhere (CCC as operators, policies) and generally being a modern IRC network (TLS with valid certs and whatnot). So we created an #archiveteam [break in messages] test channel just to play around a bit. As people seemed to like it that far, we created some new project channels over there as well to see how it holds up in real-world usage; Your Shot (#gotshot) was the first one I believe. It grew from there as people seemed to like hackint
(Everything after this is from my (OrIdow6) own logs.)
OrIdow6 asks whether or not the Wiki's (i.e. SketchCow's) prescription against the process being "[s]et up for not being some majority vote of god-knows-who weighing in and deciding things" is against votes generally, or just votes of everybody in the channel. atphoenix replies that "one interpretation is there is a seniority + engagement/participation factor to quorum. Brand new faces or even older lurkers who have basically never engaged probably should not carry much weight", and proposes a minimum of "30 days + significant involvement with 1 or more projects". OrIdow6 replies that the "[t]hings to consider" in determining the "longevity criterion" should be "(a) Experience (b) Dedication / commitment (c) Possibility of ulterior motive", proposes a 6-month threshold, and criticizes the ability to define "activity" objectively. atphoenix replies that they would be fine with 6 months, and that they just means 1 month as a reasonable minimum.
Fusl rejoins the chat after "[getting] booted out by efnet's crap servers again", and asks that someone "spread some ops again pls". As a result, astrid ops nearly? everyone in the channel who is not already opped (bitbit, wp494, kiska, Datechnom, atphoenix, Frogging, jodizzle, mtntmnky, PurpleSym, VADemon, kpcyrd, Hecatz, Ryz, Deewiant, Kenshin, marked1, pew, Craigle, and Fusl).
Kaz opines that AT is on EFNet "purely for legacy reasons". JAA responds that "[t]his channel is not really about the IRC question but more about how we want to make decisions like that as a collective". Frogging "*feels* like the network's stability has gotten worse over the years", and asks for confirmation; JAA responds with a table of netsplits per month since November of 2014.
SketchCow encourages people to edit the Wiki page. Fusl, JAA, and hook54321 talk about EFNet's webchat. atphoenix asks whether the question of who gets access to the ArchiveTeam e-mail inbox is a quorum topic; SketchCow replies in the negative: "[t]hat's just a process discussion". OrIdow6 puts two links (https://studiousguy.com/max-webers-three-types-of-authority/ and https://fivethirtyeight.com/features/when-political-parties-splinter/) in chat without context. Fusl and JAA discuss the EFnet webchat again. marked1 proposes that a logbot be invited in the face of netsplits and other instability.
hook54321 says that "we need an ice breaker here". JAA proposes that the chat "brainstorm about what such a [quorum] process could look like", and gives a "broad structure" of "proposal, discussion, vote". marked1 says that it would be bad to use Robert's Rules, as they are intened for verbal discussions, and, in order to prevent people from talking over each other etc., they are cumbersomely long. OrIdow6 says that AT "is smaller than many other groups and... has less in the way of centrally-controlled resources", which also make Robert's Rules impractical for it. A short discussion of the Wikimedia Foundations's meetings happens. They have apparently done some voice meetings; everyone agrees that AT should keep to text. JAA says that MediaWiki talk pages are "a big and awful hack" and a bad forum for discussion.
hook54321 asks, "how do we want to limit who has the ability to vote?" JAA proposes that "[p]erhaps we don't really need to restrict it", as "[i]deally, the vote result would be clear enough", even if newcomers vote contrary to the majority opinion; "[i]f it [the result] isn't [clear], then possibly the available options are not the best ones". hook54321 quotes the Wiki. marked1, on the contrary, thinks that there should be voting restrictions; they "[have] had a few IRL clubs with open membership where new people show up once to the vote meeting and aren't around for the consequences, let alone the context needed to make an informed judgment".
(Simultaneously, but separately.) atphoenix says that regardless of who votes, everybody should be able to be a part of the discussion. JAA agrees.
[I leave the channel; break in my logs of about 8 minutes.]
hook54321 says that presence on IRC does not indicate participation. JAA criticizes the ability to define "participation" objectively, giving two examples: the ambiguity of "participated in a grab" [in response to marked1], and the problem of someone participating in a few projects but not others, "[a]nd so on". Ryz proposes their own activity (heavily concentrated in ArchiveBot) as an example of the second; JAA says that Ryz was the example they were thinking of. marked1 says that activity-based participation criteria could work if broad enough, e.g. "[position] on the [warrior] leaderboard". hook54321, JAA, and atphoenix are against this, and don't think that running a warrior should be enough to get voting rights, although atphoenix thinks that "having warrior leaderboard positions could be a contributing factor". hook54321 proposes "just requiring that people have been around and participated for x amount of time" as "objective enough". atphoenix thinks that, regardless of specifics, "the most active people will kind of know who would/wouldnot [sic] be eligible".
Ryz and atphoenix both give fairly elaborate systems for determining eligibility: Ryz proposes that "the criteria for voting could be people who have been voiced or opped on major IRC channels and/or having activity [in those channels]". atphoenix proposes that "you get +1 for running a warrior. +1 for being a regular contributor to a project. +1 per month of participation, etc. 6 months activity on a single project = 6x +1 = +6. Maybe voting requires a +5?". OrIdow6 calls these "complicated" and "potentially arbitrary". atphoenix replies that "metrics take away the arbitraryness", but JAA in turn replies that they whole process is arbitrary from the start "[a]s long as the weighting and the threshold are arbitrary", something atphoenix agrees with.
JAA says that "the only non-arbitrary ways would be BDfL or completely open voting, and both have obvious other drawbacks". Ryz proposes practical problems with any system of voting restrictions: "there's a bunch of people to interact with and keep track of, that it's sometimes hard to keep track of who does what and when... (unless kept track with notes)". JAA would "like to think that it shouldn't be a problem, but then again, it might". atphoenix proposes "a hypothetical 'Application to Vote'" with "checkboxes" for whatever criteria are decided on; "enough boxes get checked... [ellipsis sic] right is granted". hook54321 thinks that "most of the time it would be somewhat obvious whether or not someone would qualify", and atphoenix agrees that this would be the case, at least among "some of the senior members".
(marked1 tries to lead into other issues: "there might be some side effects of depending on the final human count, how many people would be needed for a quorom and how long a debate period has to be left open" - but, no one bites.)
OrIdow6 asks "the purpose... of excluding people, besides filtering out sockpuppets and the like". JAA replies that it is "[k]eeping control within AT, i.e. not having random people show up and outweigh the people actually running the show", which leads atphoenix to think that "JAA is trying to prevent a hostile takeover or a coup"; OrIdow6 summarizes this as "determin[ing] who is an insider vs. an outsider", which JAA agrees with. OrIdow6 says that "a vote based on the 'I know them when I see them' approach will give the same result as a more precise system" unless ArchiveTeam either splits on opinion or fragments into multiple social groups, and that the only situation where the details of the voting eligibility rules will matter is in one of these situations. atphoenix proposes BDfL as a "tiebreaker" "or edge-case arbiter". JAA objects to BDfL entirely.
(Simultaneously, but separately.) atphoenix talks about alternative systems for voting, namely ranked-choice and double majority systems.)
hook54321 thinks that the process should prevent the situation of a close vote entirely; "[i]f there's a close split there could be more discussion and then re-voting." JAA proposes that a supermajority—specifically, "two thirds or maybe even 75%"—be required in any case; "if a vote is close to 50/50, chances are it won't go well no matter which side of the vote narrowly 'wins'". hook54321 thinks that 75% may be fine "where it's either action or no action", but thinks it's too high in situations "when action is required either way"; asked by JAA for an example, they give the hypothetical of "if our current network were to close and we had to decide on a new one".
JAA says that even many-choice like in hook54321's scenario should be made into a yes-or-no questions, after "a discussion about possible choices" - e.g. "'Should we go to Network X? Yes/No' based on which network seems to have the most support in the discussion", and says that "[t]his is... how popular initiatives work in countries where that exists (e.g. Switzerland).". JAA explains some of the procedure for the Swiss process. marked1 says that "we can be more efficient than the swiss system": "[s]omeone points out problem/open question, people agree it is one, people propose solutions, and choose best option", an example of a series of questions here being "1) Is EFnet instability harming AT productivity? 2) should AT move new channels to another network? 3) Should AT move old channels to another network? 4) if 2 or 3 carries, which network should we move to?", essentially endorsing hook54321's proposal for multiple-choice voting. [I cannot tell whether marked1 is giving these as examples of voted-on questions, or only discussion questions.] JAA criticizes this, on the grounds that having more than two options makes options "weaker", something that alternative systems of voting can only partially "counteract", and on the grounds that this makes a supermajority impractical. marked1 says that "instant-runoff is a really terrible system. (complexity while disenfranchising minorities) condorcet or approval voting are decent systems for multiple options".
kiska adds their logbot to the channel. kiska and JAA discuss specifics of the IRC network question. [A small amount of discussion on voting systems occurs here, which I have moved to the previous section.]
Nemo_bis says that, before any other discussion takes place, it is necessary to "know... the order of magnitude of the persons involved", and finds a total of 3233 nicks in their #archiveteam logs. They propose eligibility criteria based on number of messages, e.g. "10 or 20", and say that determining eligibility is an important problem. marked1 agrees with this last point; it "would help contextualize processes decisions". JAA agrees that "not everyone in #archiveteam should be able to vote", but that "most people in there probably also wouldn't vote", anyway; regardless, they are "not entirely sure whether we need to formally restrict the voting", "[a]s discussed last night". hook54321 objects to this, and suggests a criterion based on "involvement"; JAA replies that it is hard to define "involved".
(Simultaneously, but separately.) atphoenix, replying to marked1's message from earlier, is "in favor of voting systems that allowing for communicating preferences, which condorcet does".
marked1, making an analogy to governmental politics, suggests that voting, but not discussion, be based on "sending representatives" from each channel. JAA says that this wouldn't work, "because the same people are active in different channels". atphoenix says that there are a small enough amount of people in AT that this is unnecessary. marked1 also suggests that there be components analogous "citizenship/language tests" to "confirm people are integrated into society", and "[the] House of Lords, appointed after historical allegiance to the monarchy", neither of which get any responses.
hook54321 says that "OpenNIC has open-ish voting and it seems to work fine for them", although they don't know what, if anything, OpenNIC does to prevent sockpuppets. Nemo_bis affirms that sockpuppet-prevention is important.
hook54321 says that, in cases where someone's eligibility is in question, "there could be a short vote". JAA suggests a process whereby "existing voters accept new people as voters", as they "sort of... suggested last night". [I cannot find any previous mention of this specific idea; it might refer to protection from outsiders generally.] Somewhat to the contrary, marked1 suggests that "we should choose some notion of active member so the vote doesn't appear arbitrary, popular oriented, or cruel"; hook54321 replies that this isn't "really measurable", which JAA agrees with. marked1 suggests some specific criteria: "anyone who ran a target, anyone who has code on github, anyone in top 10 of a grab, alumni/emeritus". hook54321 opposes "a precise definition of what's considered an active member", although they are in favor of "something to an extent". marked1 suggests that "as we try to hold the first vote", it will become "clearer" what degree of precision is needed.
[The various participants seem to be talking about different things here; some discuss the quorum process, some specifically discuss the future IRC quorum, some discuss this present conversation itself, and some discuss this present conversation as an instance of a quorum, all while assuming the others are thinking the same way. Even if their intentions are different, their discussion is unified, and I have treated it as such.]
marked1 counts that the number of users in the quorum discussion channel (#archiveteamquorum) is 16% of #archiveteam-bs. Frogging points out that the only way for anyone to know about it is from the initial announcement on IRC 5 days ago, and suggests that there be additional "discovery points" put into place. JAA and Nemo_bis discuss naming conventions for Wiki pages on discussions. marked1 suggests "a minimum announcement period/number", "like make an announcement 3 different days over 1 week. record/measure attendance, start accepting motions". hook54321 suggests that the current discussion will "likely" be moved "to some other network". marked1 asks which will come first: discussion of "procedure for voting", or discussion of "IRC". JAA replies that "procedure for voting" must come first. marked1 suggests that "we tackle multiple topics simultaneously", lest the discussion be dragged out, but hook54321 replies that "we can't vote before there's a voting procedure". Frogging suggests that a long discussion is good: "the idea of long periods was to ensure everyone has time to notice and act", though they then suggest that they "misunderstood" marked1. JAA opines that "parallel" discussion is possible, "[b]ut only once we have a voting mechanism".
marked1 says that "it's harder to manage multiple topics is because IRC is not threaded nor everyone present at once... so the logs will look like someone pops in and adds a viewpoint to N topics, then pops out". hook54321 criticizes these people who "pop" in and out; they "shouldn't expect instantaneous responses all the time". JAA goes back on their previous opinion, and thinks that "wiki talk pages might be the best way unless we want to set up something new for this". Taking the course of "set[ting] up something new", hook54321 suggests that the discussion use something like Discourse. JAA responds that "Discourse doesn't have full threading either", and promotes something with "nested comments", like Reddit, "wiki talk pages" (again), or a mailing list. On the topic of mailing lists, hook54321 says that "[y]ou can "reply" to a post in a thread and it'll show up under "x Replies", but I don't think there's anything beyond that". [I can't tell what "I don't think there's anything beyond that" means.] OrIdow6 says that "a tree of discussion" has better organization, but leads to "shallower discussion". marked1 suggests that "[we can] tell people to put a hashtag on their AT wiki talk page and aggregate them into a single debate document".
hook54321 asks whether we would self-host the mailing list, or use a service like groups.io [the effective successor of Yahoo Groups, whose shutdown spurred a major recent project]. JAA is in favor of self-hosting for everything, "to be independent of sites shutting down". Frogging warns that "new forum software is extra stuff to maintain". atphoenix says that groups.io "has an easy to use full-data-export option, at [least] for group owners", and says that there is a dormant Google Group. JAA says that "the problem" is not whether there is an export feature, as AT would have the ability to do this anyway; "it still sucks". atphoenix replies that "AT doesn't self host all of its tools...IRC included" and that all services will close down, eventually, whether self-hosted or not. JAA in turn replies that AT is moving towards complete self-hosting; "[o]ur own IRC network has been suggested", and "we have a Gitea instance now", to which JAA has been migrating their own projects. hook54321 says that due to IRC networks' general unprofitably (precluding a motive to sell them), they may be less likely to shut down than than other types of services.
marked1 suggests that the current ArchiveTeam Reddit page may be "useful", and gives opposition to all mailing lists as leading to "lower participation". JAA replies that it "is mostly used by non-AT people". Frogging, atphoenix, and JAA oppose Reddit, the first and last because it if for-profit and lacks "long-term viability". Frogging gives the URL of a self-hosted Reddit alternative, but opines against the Reddit mode of discussion.
marked1 says that Slack is increasing in popularity because mailing lists "were inefficient"; Frogging replies that they serve separate purposes. atphoenix says that "slack/teams/hipchat/discord are IRC competitors" with more features; Frogging says that "history/logging" is badly-done on these services. hook54321 says that AT would "get kicked off" from any of these services, due to its activities breaking their terms of service; JAA agrees. hook54321 says that if AT were to ever "move off of IRC", which they "don't think we should right now", Matrix would be the most likely candidate in such an event. JAA asks that specifics of "moving away from EFNet" not be discussed until "a discussion and voting process" is in place. marked1 suggests that "we stick to Wiki and IRC for now"; "[f]inalized propositions should be up on the wiki. People can edit a summary Pro and Con page", "and also can put longer personal Op-eds on their Wiki talk page", while "lively discussion" takes place on IRC. JAA is against discussions on IRC in general, "[a]s we're actually seeing in this very channel", primarily due to the torture of reading logs from discussions you didn't take place in [which this summary is meant to counteract]. atphoenix supports moving to Matrix from experience, and suggests that marked1 advance their Wiki proposal with "some content/style examples". marked1 asks how "propositions" be named - "date or number"?
A discussion of IRC alternatives moves to #archiveteam-ot.