PEP 8015: Organization of the Python community

governance

(Victor Stinner) #1

PEP: 8015
Title: Organization of the Python community
Author: Victor Stinner
Status: Active
Type: Informational
Content-Type: text/x-rst
Created: 2018-10-04

Abstract

This PEP formalizes the current organization of the Python community and
proposes 3 main changes:

  • Formalize the existing concept of “Python teams”;
  • Give more autonomy to Python teams;
  • Replace the BDFL (Guido van Rossum) with a new “Python Steering
    Committee” of 5 members which has limited roles: basically decide how
    decisions are taken, but don’t take decisions.

PEPs are approved by a PEP delegate or by a vote (reserved to core
developers, need >= 2/3 majority).

Rationale

This PEP describes the organization of the whole Python development
community, from Python users to the Python Steering Committee.
Describing all groups and all roles in the same document helps to make
the organization more consistent.

The number of governance changes is minimized to get a smooth transition
from the old BDFL organization to the new Steering Committee
organization.

One key design of the organization is to avoid decision bottlenecks.
Discussions and decisions are distributed into Python teams where
experts in each topic can be found. The expectation is smoother
discussions on PEPs: fewer people with better knowledge of the topic.

Previously, most decisions have been taken by the Benevolent
Dictator For Life (BDFL), Guido van Rossum. The growing popularity of
Python increased the pressure on a single person. The proposed
organization distributes decisions and responsibilities to reduce the
pressure and avoid wearing any individual down.

To keep most of the decision power within the hands of the community,
the Python Steering Committee has very limited roles. The idea is to reduce the risk
that a group of people or companies “takes over” the Python project
through just a couple individuals. The project must remain autonomous
and open to everybody.

The most sensitives PEPs are decided by democracy: a vote reserved to
core developers, see the PEP process_ section below for the voting
method.

Common Guidelines

  • The Python community is open to everyone.
  • Members must respect the Python Community Code of Conduct <https://www.python.org/psf/codeofconduct/>_ which ensures that
    discussions remain constructive and that everybody feels welcomed.
  • Python is and will remain an autonomous project.
  • People with decisions power should reflect the diversity of its users
    and contributors.

Community Organization

Right now, there are different group of people involved in the Python
project. The more involved you are, the more decisions power you get. It
is important that the people acceding to the deepest group are the most
trusted ones.

This PEP formalizes the following groups:

  • Python Users
  • Python Contributors
  • Python Teams Members
  • Python Core Developers
  • Python Steering Committee Members
  • PSF Code of Conduct Workgroup

Python Users

This is the largest group: anyone who uses Python.

Python Contributors

Once a Python user sends an email to a Python mailing list, comments on
the Python bug tracker, proposes or reviews a Python change, they become
a Python contributor.

Python Teams

Python became too big to work as an unique team anymore, people
naturally have grouped themselves as teams to work more closely on
specific topics, sometimes called “Special Interest Group” (SIG).

When enough developers are interested by a specific topic, they can
create a new team. Usually, the main action is to ask the Python
postmaster to create a new “SIG” mailing list, but the team can choose
to use a different communication channel.

Team members are Python contributors and Python core developers. The
team is self-organized and is responsible to select who can join the
team and how.

Team members can get the bug triage permission on the team bug tracker
component. The more involved in a team you are, the more decisions power
and responsibilities you get.

A team might become allowed to decide on their own PEPs, but only the
Python Steering Committee can allow that (and it has the power to revoke
it as well). Such a case is exceptional, currently a single team has
such permission: the Packaging Team.

See Annex: Examples of Python Teams_.

Python Core Developers

One restricted definition of a core developer is the ability to merge a
change (anywhere in the code) and have the bug triage permission
(on all bug tracker components).

Core developers are developers who are proven to have the required skills to
decide if a change can be approved or must be rejected, but also (and
this is more important) what changes should not be made. Python has a
long history, big constraints on backward compatibility, high quality
standards (ex: changes require new tests). For these reasons, becoming
a core can take several months or longer.

Becoming a core developer means more responsibilities. For example, if a
developer merges a change, they become responsible for regressions and
for the maintenance of that modified code.

Core developers are expected to be exemplary when it comes to the Code
of Conduct. They are encouraged to mentor contributors.

Promote a contributor as core developer

Once an existing core developer considers that a contributor is ready to
join the core group, to become a core developer, that core developer
asks the contributor if they would like to become a core developer. If
the contributor is interested in such new responsibilities, a vote is
organized.

The vote is reserved to core developers, is public, and is open for 1
week. Usually the core developer who proposes the promotion has to
describe the work and skills of the candidate in the description of the
vote. A contributor is only promoted if two thirds (>= 2/3) of
votes approve ("+1") the promotion. Only “+1” and “-1” votes are
accounted; other votes (ex: null, “-0”, “+0.5”) are ignored.

If the candidate is promoted, usually they get a mentor for 1 month to
help them to handle new responsibilities.

If the candidate is not promoted, a new vote can be organized later,
when the candidate gets the missing skills, for example 6 months later.

Python Steering Committee

The Python Steering Committee is made of the most trusted core
developers since it has the most decision power. The roles of this group
are strictly limited to ensure that Python keeps its autonomy and
remains open.

The Python Steering Committee is composed of 5 members. They are elected
for 3 years and 1/3 is replaced every year (first year: 1, second year:
2, third year: 2). This way, a member will stay for one full Python
release and the committee composition will be updated frequently. A
committee member can be a candidate for the seat they are leaving.
There are no term limits.

Committee members must be Python core developers. It is important that
the members of the committee reflect the diversity of Python’ users and
contributors. A small step to ensure that is to enforce that only 2
members (strictly less than 50% of the 5 members) can work for the same
employer (same company or subsidiaries of the same company).

The size of 5 members has been chosen for the members diversity and to
ensure that the committee can continue to work even if a member becomes
unavailable for an unknown duration.

Python Steering Committee Roles

Python Steering Committee roles:

  • Decide how a PEP is approved (or rejected or deferred).
  • Grant or revoke permissions to a Python team. For example, allow
    a team to give the bug triage permission (on the team component) to a
    contributor.

To decide how a PEP is approved (or rejected or deferred), there are two
options:

  • The committee elects a PEP delegate (previously known as “BDFL-delegate”):
    a core developer who will take the final decision for the specific
    PEP. The committee select the PEP delegate who can be proposed by the
    Python team where the PEP is discussed.
  • The committee can organize a vote on on the PEP, see PEP process_
    for the vote organization. The committee decides when the vote is
    organized. A vote is preferred for changes affecting all Python users,
    like language changes.

The committee keeps the “vision” and consistency of Python. It also makes
sure that important features reach completion. Their ability to pick PEP
delegates is meant to help them to achieve that goal.

Election of Python Steering Committee Members

The vote is organized by the Steering Committee. It is announced 3 weeks
in advance: candidates have to apply during this period. The vote is
reserved to core developers and is open for 1 week. To avoid
self-censorship, the vote uses secret ballots: avoid the risk of
hostility from someone who may get more power (if they get elected).

The vote uses the Schulze/Beatpath/CSSD variant <https://en.wikipedia.org/wiki/Schulze_method>_ of the Condorcet method <https://en.wikipedia.org/wiki/Condorcet_method>_ using an
online service like Condorcet Internet Voting Service (CIVS) <https://civs.cs.cornell.edu/>_. This voting method reduces the risk of
tie. It also produces a ranking of all candidates, needed for the
creation of the committee.

In case of tie, a new vote is organized immediately between candidates
involved in the tie using the same voting method and also during 1 week.
If the second vote leads to a tie again, the current Steering Committee
is responsible to select the elected member(s).

If a committee member steps down, a new vote is organized to replace
them.

If the situation of a committee member changes in a way that no longer
satisfies the committee constraint (ex: they move to the same company as
two other committee members), they have to resign. If the employer of a
member is acquired by the employer of two other members, the member with
the mandate ending earlier has to resign once the acquisition completes.

Election Creating the Python Steering Committee Members

To bootstrap the process, 5 members are elected at the committee
creation. The vote follows the same rules than regular committee votes,
except that the election needs 5 members, and the vote is organized by
the PSF Board.

In a council election, if 3 of the top 5 vote-getters work for the
same employer, then whichever of them ranked lowest is disqualified
and the 6th-ranking candidate moves up into 5th place; this is
repeated until a valid council is formed.

In case of tie, a second vote is organized immediately between
candidates involved in the tie and following candidates to fill the
remaining seats. The vote follows the same rules as the regular
committee vote. If the second vote still result in a tie, the PSF Board
is responsible to elect members and decide their position in the vote
result.

The order in the vote result must be unique for elected members: #1 and
#2 are elected for 3 years, #2 and #3 for 2 years, and #5 for 1 year.

Example of vote result with a tie:

  • A
  • B
  • C
  • D
  • E, F
  • G

The first 4 candidates (A, B, C and D) are elected immediately. If E
works for the same employer than two other elected member, F is also
elected. Otherwise, a second vote is organized for the 5th seat between
E and F.

Special Case: Steering Committee Members And PEPs

A committee member can be a PEP delegate.

A committee member can propose a PEP, but cannot be the PEP delegate of
their own PEP.

When the committee decides that a PEP must be voted, committee members
can vote as they are also core developers, but they don’t have more
power than other core developer.

PSF Code of Conduct Workgroup

Charter

The workgroup’s purpose is to foster a diverse and inclusive Python
community by enforcing the PSF code of conduct, along with providing
guidance and recommendations to the Python community on codes of
conduct, that supports the PSF mission of “ongoing development of
Python-related technology and educational resources”.

We work toward this common goal in three ways:

  • Review, revise, and advise on policies relating to the PSF code of
    conducts and other communities that the PSF supports. This includes
    any #python chat community & python.org email list under PSF
    jurisdiction.
  • Create a standard set of codes of conduct and supporting documents for
    multiple channels of interaction such as, but not limited to,
    conferences, mailing lists, slack/IRC, code repositories, and more.
  • Develop training materials and other processes to support Python
    community organizers in implementing and enforcing the code of
    conduct.

The organization of this workgroup is defined by the
ConductWG Charter <https://wiki.python.org/psf/ConductWG/Charter>_.

Special Case: Ban a core developer

As any other member of the Python community, the PSF Code of Conduct
Workgroup can ban a core developer for a limited amount of time. In this
case, the core developer immediately loses their core developer status.
Core developers are expected to be exemplary when it comes to the Code
of Conduct.

In general, a ban is only the last resort action when all other options
have been exhausted.

At the end of the ban, the developer is allowed to contribute again as a
regular contributor.

If the developer changes their behavior, another core developer can
organize a new vote to propose the developer for promotion to core
developer. The vote follows the same process than for any other Python
contributor.

PEP process

There are 2 main roles on PEPs:

  • PEP Authors
  • PEP Delegate

PEP Authors do their best to write high quality PEP.

The PEP delegate is responsible to help the authors to enhance their PEP
and is the one taking the final decision (accept, reject or defer the
PEP). They can also help to guide the discussion.

If no decision is taken, the authors can propose again the PEP later
(ex: one year later), if possible with new data to motivate the change. A
PEP Delegate can also choose to mark a PEP as “Deferred” to not reject
the PEP and encourage to reopen the discussion later.

PEPs specific to a Python team are discussed on the team mailing list.
PEPs impacting all Python developers (like language changes) must be
discussed on the python-dev mailing list.

Vote on a PEP

When the Python Steering Committee decides that a PEP needs a wider
approval, a vote is organized.

The vote is reserved to core developers, is public, is announced 1 week
in advance, and is open for 1 week. The PEP can still be updated during
the 1 week notice, but must not be modified during the vote. Such vote
happens on
the mailing list where the PEP has been discussed. The committee decides
when the vote is organized. The PEP must have been discussed for a
reasonable amount of time before it is put to vote.

A PEP is only approved if two thirds (>= 2/3) of votes approve
("+1") the PEP. Only “+1” and “-1” votes are accounted; other votes
(ex: null, “-0”, “+0.5”) are ignored.

A PEP can only be approved or rejected by a vote, not be deferred.

Lack of Decision

If a discussion fails to reach a consensus, if the Python Steering
Committee fail to choose a PEP delegate, or if a PEP delegate fails to
take a decision, the obvious risk is that Python fails to evolve.

That’s fine. Sometimes, doing nothing is the wisest choice.

Change this PEP

The first version of this PEP has been written after Guido van Rossum
decided to resign from his role of BDFL in July 2018. Before this PEP,
the roles of Python community members have never been formalized. It is
difficult to design a perfect organization at the first attempt. This
PEP can be updated in the future to adjust the organization, specify how
to handle corner cases and fix mistakes.

Any change to this PEP must be validated by a vote. The vote is
announced 3 weeks in advance, is reserved to core developers, happens in
public on the python-committers mailing list, and is open for 1 week.
The proposed PEP change can still be updated during the 3 weeks notice,
but must not be modified during the vote.

The change is only approved if four fifths (>= 4/5) of votes approve
("+1") the change. Only “+1” and “-1” votes are accounted; other votes
(ex: null, “-0”, “+0.5”) are ignored.

Annex: Summary on votes

====================== ======= ====== ======= =================================
Vote Notice Open Ballot Method
====================== ======= ====== ======= =================================
Promote contributor none 1 week public >= 2/3 majority
PEP 1 week 1 week public >= 2/3 majority
Change this PEP 3 weeks 1 week public >= 4/5 majority
Steering Committee 3 weeks 1 week private Condorcet (Schulze/Beatpath/CSSD)
====================== ======= ====== ======= =================================

All these votes are reserved to core developers.

Annex: Examples of Python Teams

Below are examples of some Python teams (the list will not be kept up to
date in this PEP).

Packaging Team

The packaging team runs its own PEP category and can approve (or reject)
their own PEPs.

  • Website: packaging.python.org <https://packaging.python.org/>_
  • Mailing list: distutils-sig <https://mail.python.org/mm3/mailman3/lists/distutils-sig.python.org/>_
  • Bug tracker component: Distutils
  • Example of members: Paul Moore, Nick Coghlan, Donald Stuff
  • Stdlib module: distutils
  • Current PEP delegate: Paul Moore

IDLE Team

IDLE is a special case in the Python standard library: it’s a whole
application, not just a module. For this reason, it has been decided
that the code will be the same in all Python stable branches (whereas
the stdlib diverges in newer stable branches).

  • Bug tracker component: IDLE
  • Example of members: Terry Reedy, Cheryl Sabella, Serhiy Storchaka
  • Stdlib module: idlelib

Mentorship Team

Becoming a core developer is long and slow process. Mentorship is an
efficient way to train contributors as future core developers and build
a trust relationship.

Note: The group is not responsible to promote core developers.

Documentation Team

  • Mailing list: doc-sig <https://mail.python.org/mailman/listinfo/doc-sig>_
  • Bug tracker component: Documentation
  • GitHub tag: type-doc
  • Example of members: Julien Palard, INADA Naoki, Raymond Hettinger.

The team also manages documentation translations.

See also the Mentorship team which maintains the “Devguide”.

Security Team

  • Website: https://www.python.org/news/security/

  • Mailing lists:

    • security@python.org (to report vulnerabilities)
    • security-sig <https://mail.python.org/mm3/mailman3/lists/security-sig.python.org/>_
      (public list)
  • Stdlib modules: hashlib, secrets and ssl

  • Example of members: Christian Heimes, Benjamin Peterson

The security@python.org mailing list is invite-only: only members of
the “Python Security Response Team” (PSRT) can read emails and reply;
whereas security-sig is public.

Note: This team rarely proposed PEPs.

Performance Team

Usually PEPs involving performance impact everybody and so are discussed
on the python-dev mailing list, rather than the speed mailing list.

Asynchronous Programming Team

  • Website: https://docs.python.org/dev/library/asyncio.html
  • Mailing list: async-sig <https://mail.python.org/mailman/listinfo/async-sig>_
  • Bug tracker component: asyncio
  • GitHub label: expert-asyncio
  • Stdlib modules: asyncio and contextvars
  • Example of members: Andrew Sveltov, Yury Selivanov

PEP only modifying asyncio and contextvars can be discussed on
the async-sig mailing list, whereas changes impacting the Python
language must be discussed on python-dev.

Type Hints Team

  • Website: http://mypy-lang.org/
  • Repository: https://github.com/python/typing
  • GitHub label for mypy project: topic-pep-484 <https://github.com/python/mypy/labels/topic-pep-484>_
  • Stdlib modules: typing
  • Example of members: Guido van Rossum, Ivan Levkivskyi,
    Jukka Lehtosalo, Łukasz Langa, Mark Shannon.

Note: There is a backport for Python 3.6 and older, see
typing on PyPI <https://pypi.org/project/typing/>_.

Version History

History of this PEP:

  • Version 7: Adjust the Steering Committee

    • The Steering Committee is now made of 5 people instead of 3.
    • There are no term limits (instead of a limit of 2 mandates:
      6 years in total).
    • A committee member can now be a PEP delegate.
  • Version 6: Adjust votes

    • Specify the Condorcet method: use Schulze/Beatpath/CSSD variant to
      elect Python Steering Committee members. Specify how to deal with
      tie and the constraint on the employers.
    • Vote on promoting a contributor and on PEPs now requires >= 2/3
      rather than 50%+1.
    • Vote on changing this PEP now requires >= 4/5 rather than
      50%+1.
    • Explain how to deal with a company acquisition.
  • Version 5: Election of Python Steering Committee Members uses secret
    ballots

  • Version 4:

    • Adjust votes: open for 1 week instead of 1 month, and announced
      in advance.
    • Rename the “Python Core Board” to the “Python Steering Committee”;
    • Clarify that this committee doesn’t approve PEPs and that committee
      members cannot cumulate more than 2 mandates;
    • Add the “Type Hints” team to the annex.
  • Version 3: Add “Special Case: Ban a core developer” and “How to update
    this PEP” sections.

  • Version 2: Rename the “Python board” to the “Python Core Board”,
    to avoid confusion with the PSF Board.

  • Version 1: First version posted to python-committers and
    discuss.python.org.

Copyright

This document has been placed in the public domain.


(Doug Hellmann) #2

How do we handle the case where it is time for someone to rotate off of the board (they have served their 2 terms) and there is no one available with the time or inclination to take their place?


(Victor Stinner) #3

I’m sure that we will find candidates :slight_smile:


(Carol Willing) #4

I would like to suggest a more formal approach that will not preclude the above but will make the process more transparent and actionable. Perhaps the mentorship team or workgroup prepares once or twice a year (set timetable) a list of contributors, in whatever capacity, who may be ready for promotion. The workgroup would propose a slate of contributors to be recommended for promotion. Voting on the slate would be all or none for the slate and not individual voting.


(Carol Willing) #5

Victor, I’ve left you a PR on the peps repo. It renames Python board to Python Core Board to minimize confusion with PSF Board.


(Victor Stinner) #6

I’m trying to work incrementally with small changes. My PEP only formalizes the current process to promote a contributor, but it also makes some things stricter, especially the vote. For example, previously, it was common that the vote ended once we have more than 5 “+1” votes. Now the vote is open for 1 month to make sure that all core developers have the opportunity to speak up.

Sure, the Mentorship Team can help to detect ready candidates. But I’m not sure about the all-or-nothing vote. Today, core developers take this vote very seriously and have very strong requirements before someone is promoted. Each profile is examined in depth. Wait, I’m not giving my opinion here, again, just describing the current process. Moreover, having to vote for multiple candidates at once can more stressful and the risk is to get more “-1” just because voters don’t have time to examine all profiles.

I like the idea of being able to promote a candidate as soon as I consider that they is ready. Why do you want a strict timetable? If we have a timetable, I would prefer to have a maximum of one month. Otherwise, there is a risk that the candidate becomes bored and moves away from Python.

More generally, my intent is focus strongly on teams rather than on the “core developer” superpower which seems to be a bottleneck (it’s currently too hard to become a core developer). Joining a team should be way simpler since only the few members of the team decide who is ready or not, not the 100+ core developers. I would like to distribute the workflow to better scale horizontally.


(Antoine Pitrou) #7

A relatively minor comment, but I think 3 people is too small a number for the Board of a large project like CPython, especially since those will be volunteers and not available at all times.

I would go for 5, 7 or 9 (an even number would also work, but I know some people prefer odd numbers because it makes it easier to break ties in inconclusive discussions).


(Victor Stinner) #8

I have no strong opinion on the exact size of the Python Core Board, but let me elaborate my rationale on the size.

I hope that the maximum response time will be 1 month. People usually take holiday for 1-3 weeks, not longer. If a member is away longer, others can decide to take decision without them. If everybody is away, well, you will have to wait. And I’m not sure that it’s a major blocker issue is the pause is not too long. With a BDFL, it was common to have no decisions for months. My hope is that my proposed organization doesn’t make things worse. We replace 1 person with a board, many teams and many PEP delegates to take decisions.

My fear is that it’s way more difficult to take a decision with 5 people, and worse with more people. With 7 people, we start to want to know how board members take a decision. Should we require a vote inside the board!? I like the idea of a small group to have a smoother process.


(Victor Stinner) #9

General note: my hope is that the governance PEP will only be a first milestone, but that it will remain to adjust the organization later with small changes to optimize the decision process. Since nothing was formalized previously, we cannot design the perfect organization at once.

For me, the main question is the level of autonomy that we want to give to Python teams. I decided to not put it in my PEP (maybe I’m wrong and everything should be in the PEP), but an important question will be if a team will be able to give the commit bit on the files handled by the team. For example, can the board allow the Documentation team to give the commit bit to someone on all files inside the Doc/ subdirectory? Technically, contributors who are not core developers will need to use the automerge bot which would know the Python teams and fine permissions.

In the meanwhile (with my current PEP), contributors have to wait for a core developer to press the [Merge] button. But I hope that contributors and core developers in a team will be able to build a trust relationship so core devs rely on contributors of the team to review changes.


(Carol Willing) #10

Thanks Victor. I agree that the details do not need to be in the PEP. My suggestion of an all or nothing vote on a slate of ready candidates identified by the Mentorship Team is to give the Mentorship Team the trust to identify “ready” core contributors for push privileges.

The slate encourages a more mindful way to identify those who are ready rather than rely on one core dev to recognize and put forward a particular candidate. I would encourage both mechanisms so that we don’t lose “ready” core devs through oversight or unconscious bias.

While it’s not necessary to have in a PEP for governance, it’s a concept that would be willing to consider if a Mentorship workgroup or team materializes post-governance selection. :smile:


(Antoine Pitrou) #11

Do note that not all core developers, even recent, went through an official mentorship process before becoming core developers.


(Carol Willing) #12

Agreed :smile: I’m advocating both individual recommendations as we do now and an additional opportunity for identifying consistent contributors.


(Victor Stinner) #13

Mariatta has been mentored by Guido, Pablo has been mentored by me, Lisa has been mentored by Raymond, Emily has been mentored by Guido and Eric. Mentoring is effective :slight_smile: I’m not sure of what you mean by “official mentorship”.


(Antoine Pitrou) #14

You misunderstood me. I said “not all core developers”, not “no core developers”. Yes, Mariatta et al were mentored. Other people were not mentored before becoming core developers.


(Victor Stinner) #15

I like the idea of trusting people more easily, but I don’t think that we are ready to push too many changes in the organization at once. Right now, we have an identity crisis and a lack of decision process. My PEP tries to formalize a lot of things that have never been formalized before. A friend who reviewed my PEP asked me “what is the current definition of a core developer?”… Well, there is no official definition :slight_smile: That’s just one example.

Again, I’m in favor of making incremental changes. We cannot design a new perfect governance at once. Well, maybe the PEP should explain that. I mean that once we have a strong decision process, we can more easily enhance the governance.

Well, that’s my opinion. Maybe others are in favor of putting more changes in the governance PEP :slight_smile:


(Carol Willing) #16

Incremental changes after the governance is decided works for me. For the best interests of the Python language and to better reflect the diverse communities (science, data science, web, etc.) that the language serves, it’s reasonable to revisit this at a later date. Thanks Victor.


(Victor Stinner) #17

I proposed an update to my PEP: https://github.com/python/peps/pull/810/files

Explain the special case of the PSF Code of Conduct Workgroup banning a core developer, and explain how this PEP can be updated. As I wrote, I would like to that the PEP can be updated later to adjust it.

For the ban, see the current discussion on python-committers: https://mail.python.org/pipermail/python-committers/2018-October/006287.html


(Donald Stufft) #18

In this PEP, you say that new board members are elected using the “Condorcet method”, however there really isn’t a singular “Condorcet method”, the closest thing is just running all of the pairwise elections, but that runs into issues with cycles. So I think that it is generally important that we define the specific Condorcet method that is being used (Shulze is pretty popular, Ranked Pairs is as well and is maybe slightly better but either is fine), and also what happens in ties.

Another problem is that in your initial election, you need to elect multiple winners, not just a single winner (and presumably this case can also happen if someone steps down around the time we need to elect a new person as well), but not all Condorcet allow you to see who is second/third/etc and only allow you to see the actual winner. Some of the Condorcet methods allow you to get a complete ranking (if one exists), which would be a useful property in cases where you need to elect multiple people (Shulze and Ranked Pair both do this). Alternatively it could be specified that you’d just re-run the election with the first winner removed to select the second winner.

Something that’s not clear to me here, can the Python Core Board itself decide on a PEP, or can it only delegate or open it up for a vote?


(Victor Stinner) #19

I’m not an expert on voting system. My requirements are that the vote will “sort” candidates to be able to see who are the 3 winners, and limit the risk of having two candidates at the same rank. What is your favorite voting method to implement that? What I had in mind is to require to rank all candidates and disallow to put two candidates at the same ranking.

If the vote fails to give an unique rank for two winners, would it be crazy to organize a short vote to select on of these two candidates? In France, we vote for one candidate for the president. A candidate only wins at the first round if they have more than 50%. It didn’t happen (at least recently?). So we always have a second vote between two (or sometimes three) candidates. The candidate with most votes wins. It seems fair to me. I’m not proposing this method, I’m just saying that I’m fine with having a second vote if needed.


(Victor Stinner) #20

Oh, I expected this point to be very explicit. The Board does never approve a PEP. They are only involved in the decision process if they chose a vote open to all core developers, since they are also core developers. But in such case, they only have one vote, as any other core devs. They don’t have power than others.

The Board only decides how a PEP is approved, and they cannot be PEP delegate: * Special Case: Board Members And PEPs.

Do you have a suggestion to rephrase the PEP to make this point more explicit?