PSF voting methods

PSF
1

The Steering Council election recently concluded, using a new voting method and voting service. It’s fair to say it was a success by any measure. Here are the results:

An informal poll after the election showed that nobody preferred older voting services, although half were indifferent.

More to the point, almost everyone preferred the new voting method. Two had no preference, and only one preferred Approval (which most PSF-related elections still use).

The newer method is called STAR, Score Then Automatic Runoff. Each ballot gives each candidate from 0 through 5 stars. The two candidates with the highest total number of stars proceed to a runoff phase: whichever of the two is preferred on more ballots is the winner. A is preferred to B on a ballot just means the ballot gave more stars to A than to B. The magnitude of the difference doesn’t matter. 2 to 1 is just as much “a preference” as 5 to 0.

For a multi-winner election, put the winner aside, and repeat the process with the candidates remaining. You don’t need to vote again. You cast one ballot at the start and you’re done.

There’s nothing wrong with Approval. It’s a fine method! And it will usually deliver the same winners as STAR. In large-scale voting simulations, STAR does somewhat better, but that’s not the primary point here.

The point is that people like being able to express more of what they believe, and particularly to express that they like A “a lot more” than B, or “just a little more”, or that they have no preference between them (if so, give them the same number of stars). The happier people are with their voting experience, the more likely they are to vote, and that matters too.

STAR is more involved than Approval, but not much so. Approval is basically a “0 or 1 star” method Every part of STAR is simple, and easy to understand.

STAR is a highly regarded (by voting theory experts) method, and - a story for another day - encourages honest ballots in several ways. Games people play to try to “outwit” purely scored (“cardinal”) or purely preference (“ordinal”) methods are as likely as not to backfire under STAR, because it incorporates aspects of both cardinal and ordinal methods. Each tends to counteract the weaknesses of the other.

STAR’s leading advocacy group is “The Equal Vote Coalition”, which has a world of info about the method:

It’s refreshingly non-hysterical for an advocacy site, though. They play fair, and - yup! - they like Approval too. They just like STAR better.

I’d like to see the PSF move to STAR for Board elections too. How about you?

Hoppy to answer any questions I can. But tagging @ArendPeter, who is an expert on both the method and the voting service. Arend is the best source for definitive answers.

1 Like
2

Once upon a time, a nonprofit I’m involved with wanted to switch from a proportional voting model to a ranked choice model for its Board of Director elections. This was about a decade ago and there was a preference to use Condorcet because elections for the technical leadership of the project community represented by that same nonprofit used it.

Ultimately, legal counsel strongly recommended against the change, because the state in the USA where the organization was incorporated did not have any other known organizations using that method. Instead they suggested IRV because it had precedent in the bylaws of other nonprofits incorporated in the same state.

It’s not an argument for or against any particular methods (I happen to quite like Condorcet, STAR, and even IRV a bit), I just wanted to point out that there can be legal concerns when it comes to the voting methods used for member-elected boards of incorporated organizations that aren’t a problem for community-elected technical leadership bodies.

I’m not a lawyer, but past experience suggests it might help to begin by looking into what other organizations incorporated in Delaware are already relying on STAR for director elections.

1 Like
3

Yes, legalities are important too. I’m not a lawyer and can’t speak to them. Perhaps @ArendPeter can? Not as a lawyer, but because the Equal Vote Coalition has experience with the legalities here.

The PSF bylaws say

The directors shall be elected by a plurality of the votes of the members present in person or represented by proxy at the meeting and entitled to vote on the election of directors.

which appears to have approximately nothing in common with how directors have been elected for years already. That is how it was done during at least most of my years on the Board, and it’s possible during all of them (memory fades),

Before its election, the Steering Council’s voters changed its governing docs (PEP 13) to explicitly require STAR.

Where there’s a will, there’s a way :wink:

4

I used to prefer Approval voting. It has the advantages of being about as simple as possible while still giving good results. I think simple is important because ordinary people want to understand and trust how elections are run. That’s why I prefer pencil and paper voting for civil elections.

However, this “express what they believe” issue is perhaps fatal to Approval in terms of getting wide adoption. People don’t like to be forced into the binary choices and have a hard time believing that it likely doesn’t matter in an election with many voters. So, I think the little extra complication STAR requires is okay. Something like Condorcet is disqualified on the complexity aspect, IMHO.

I think the PSF should change as well. It will likely make the voters a bit more happy. It will also popularize the STAR method, which will hopefully one day replace the horrible methods being used in other elections (e.g. plurality and IRV, for example).

5 Likes
5

Given the importance of bylaws, could it be that “by a plurality of the votes” can be interpreted as supporting a variety of voting methods? Otherwise, shouldn’t the bylaws be changed to align with the actual voting method used (ideally this should have been done before the change but better late than never).

Maybe Section 14.8 of the bylaws is what authorizes the current method as it seems to offer an explicit alternative to section Section 3.9: “Any vote of … the members … shall have the same effect as acton taken by written consent; provided that such voting mechanism meets the criteria set forth in this section 14.8.”

In full:

Section 14.8. Electronic Voting.

Any vote of the Board of Directors, any committee, or the members may be conducted through electronic means and shall have the same effect as action taken by written consent; provided that such voting mechanism meets the criteria set forth in this Section 14.8. Any vote conducted through electronic means must be done through a mechanism by which both the identity of each voter and the date that such vote is made can be verified. No vote conducted pursuant to this Section 14.8 may remain open for more than sixty (60) days after the commencement of the applicable voting period. Each such vote shall have a specific approval requirement identified prior to the commencement of such vote, which requirement shall not be less than the requirements set forth in the corporation's Certificate of Incorporation, these Bylaws or the General Corporation Law of the State of Delaware. The effective date of any vote conducted through electronic means shall be the first date upon which the requisite threshold for approval of such action has been obtained.`
1 Like
6

I think you can pay a lawyer to read anything you want into words :wink:

For context,

  • The text about electronic voting was added during my years on the Board, and reviewed by Van Lindberg (former PSF chair and actual real-life lawyer).

  • One reason so many US corporations (like the PSF) incorporate in Delaware is that the state doesn’t hassle businesses much. You could certainly pay a lawyer to raise fear about not following the bylaws here, but … why bother? Delaware probably won’t.

  • During my years on the Board, there was an annual members meeting, and directors were elected at that meeting. Which the bylaws still seem to insist must be the case.

  • The annual meeting was held at the primary US PyCon that year. I’ve related elsewhere why that came to be disliked. Short course: it became clear that candidates not physically at the meeting were severely disadvantaged by “out of sight, out of mind”, giving undue advantage to US candidates. Divorcing Board elections from US PyCons was, and remains, an excellent idea. Ditto the move to electronic voting.

  • “Any vote conducted through electronic means must be done through a mechanism by which both the identity of each voter and the date that such vote is made can be verified:”

    I’m not sure that’s ever been met, with respect to Board elections. Due to privacy concerns, I’m not aware of any voting service that allows the public to learn when anyone voted - or even of who voted, apart from possibly revealing a list of all qualified voters (entitled to vote, not whether they voted).

    An election admin can usually find out such stuff, but not the public. So we can all pay lawyers now to read the meaning of “verified” in the way we like :wink:

  • If a lawyer said STAR voting is incompatible with our bylaws, I’d be very surprised if they could say, with a straight face, that Approval voting is compatible. Approval very intentionally isn’t “plurality voting” either.

2 Likes