Also another question Question About Steering Committee Future
Is this election for 5 seats or 3 seats? Wasn’t made clear exactly.
2 Likes
It’s unclear whether I got the Github notification because I’m in the org or because I’m eligible to vote (I have no idea whether I am or not this time around, and the algorithm to find out for myself seems a tad complex). Looks like the process could use some clarification.
1 Like
I think it was because we can vote? It used the voter team and I see myself on this list:
1 Like
Thank you to everyone who has put their hat in the race for doing this important work!
1 Like
Just as a reality check, there were 450 votes last year, 20% of which is 90, and as of right now there are 118 e-mails registered to vote, two days after the election was announced and one month remaining in the registration window.
6 Likes
i would see some room to further clarify this on granularity, e.g. as to whether this might cover:
- same company? (this one may be most obvious)
- same industry (e.g. Nix consulting)?
- same job title (e.g. two entrepreneurs)?
i’m probably missing things, but yeah, i do maybe still see some room for interpretation here.
Good question, thanks!
Constitution says «e.g. employees of the same company or otherwise same payer of Nix work», and tasks EC with deciding the interpretation and corner cases. Same industry / same job title are too far from the central examples.
Note that even for same-payer EC has latitude as «subsidiaries, grants, part-time work, etc.» are listed as cases which might be decided not to be CoI if the circumstances warrant this, and again the list ends with etc.
4 Likes
EDIT: Sorry for the double-reply, I missed the one above while drafting this one.
As defined in the Nix governance constitution, this seems to reduce in it’s most basic form to
conflict of interest (e.g. employees of the same company or otherwise the same payer for Nix work), as decided by the EC.
My interpretation of that would be summarized as “up to the EC, but here are two examples that clearly present CoI.”
There is more in the doc that speaks to various aspects of how CoI is currently intended to be handled from a procedural standpoint. Non-exhaustively, there are sections about how the Conflict of Interest balance is supposed to work, the current mandates for disclosure of CoI, and ways for an incumbent SC “to prevent a nominee from becoming a candidate by supermajority (among the currently serving members of the SC) in case their public image or conduct would not be compatible with the position in the SC”
2 Likes
Yet another thing I have messed up by not saying: if you are added to voters-2025 team of NixOS GitHub organisation, you are in the eligible.csv; GitHub CSV preview might show a part of the file by default.
3 Likes
Very nice data you guys set up there
@7c6f434c thank you for organizing the election!
1 Like
We are not yet there; and it’s not me doing most of the work.
1 Like
Good point. I should have clarified this from the get-go, sorry.
The current situation is an election for 5 seats, 3× 2-year terms to succeed 1-year terms from the initial election, and 2× 1-year terms to serve out the rest of the terms of the resigning-at-handover 2-year-term members.
4 Likes
Is there a tool for checking if I’m automatically eligible?
Look for your GitHub username in the file eligible.csv. If you are there, you are automatically eligible.
3 Likes
Or, check if you are in voters-2025 team (this is downstream of eligible.csv)
I’m not confident the proposed election method facilitates conditions on the selected committee (in our case: preventing same CoI) to help select the committee satisfying such criterion most proportionally representing voter preferences.
methods scoring committees (e.g. Chamberlin-Courant) should address that, raising questions on OpaVote support - tho fortunately, without high-ranking candidates with similar CoI this concern should be moot.
Chamberlin-Courant
Is it that one method that implies that the resulting committee has to use weighted voting?
And Monroe is without that issue but still gets hard to compute for large committee sizes?
Last time we tried a committee-oriented method, I later had to report a mismatch between what it tries to achieve and where optimisation algorithm as designed might get stuck. On the other hand, given the ballots, there is a meaningful sense in which this problem that I reported didn’t matter in the actual election. That time. But even without CoI issues there are possible catches.
CoI-forced recounts are a mess, although not too bad of a mess. Nobody wants them, so people with same CoI seem to have talked in advance last time. I am grateful to them for that. I hope we have the same this time.
Meek-STV is reasonable-ish at proportionality, provably not too bad at proportionality, and converges to what it says on the tin. I think that for choosing anything even fancier to be a good idea, one needs to run a true working group over a few months, to check the actual detailed promises per-method, plus bugginess of available implementations.
If someone ran that working group, published a survey of methods and implementations, and applied for EC position with that survey linked in «why me», probably they would be in a position to do things better than what will be used this year.
2 Likes
While we are voting-theory-nerding out, I don’t know how much it is worth it to make waves about this now, but I do think that a Condorcet method like the one we used last year is superior to a non-Condorcet method for this kind of choice. It does make a difference in outcome; I’ve implemented and run the Meek STV method on last year’s ballots, and it would have elected a different SC (by one member). 184 ballots express a preference for the ‘true’ SC over the one Meek STV would have chosen, while only 153 would have preferred the reverse (with the remaining 113 ballots ambivalent between the two).
I’d go so far as to say that I’d like to have the voting method (not the implementation) codified in the constitution at some point. When favoring one method over another favors one outcome over another, one really doesn’t want it to be selected anew each year, for fear of trying (or the appearance of trying) to derive a particular outcome undemocratically.
6 Likes
184 ballots express a preference for the ‘true’ SC over the one Meek STV would have chosen, while only 153 would have preferred the reverse
This statement as written has form indistinguishable from «AAAAA committee is preferred to AAABB given 60:40 A:B community split». The method used is better than that, but has enough nuances that writing it into the constitution was arguably unfortunate (although for reasons that would be hard to figure out quickly when we were writing).
DM me or mention me in a fresh thread if you want to discuss in detail the actual properties of the methods I have looked at, including obviously proportional-ish Condorcet, and ballot data crunching.