Code of Conduct Next Steps - Community feedback needed

LLVM Community,

The LLVM Code of Conduct has been in draft mode for several years now. In order to finalize the Code of Conduct, there are 3 steps left to complete:

  1. Draft an Incident Response Guide.

This guide is intended for someone who is considering reporting a potential code of conduct violation. You can view and comment on the proposed guide here (or by email if you prefer): https://docs.google.com/document/d/10LClyw1x1e4OIiKFFRqklbZ2D_xwZz85HZ7kWwZ_yys/edit?usp=sharing

  1. Draft a Response Guide.

This guide is intended for members of the code of conduct committee or organizers of an event. You can view and comment on the proposed guide here (or by email if you prefer):
https://docs.google.com/document/d/1dEvF9NwZdaDLQNPVBudipdFiZ0E553Lz3qOqeCsfpNk/edit?usp=sharing

  1. Form code of conduct committee.

This committee will be responsible for responding to code of conduct reports as described in the response guide. The LLVM Foundation board will propose the initial members of this committee and provide a period of time to collect feedback from members of the community.

In order to keep this process running smoothly and moving forward, we would appreciate getting any community feedback within the next two weeks…

Some previous Code of Conduct discussions have been difficult - please remember to treat others with respect and kindness in your response.

Thanks,
Tanya

Hi Tanya,

Is there a reason for hosting things for review on Google Docs? We currently have both Phabricator and GitHub that work for review of any text-based format. When I click on a Google Docs link, I am asked to agree to a privacy policy that is very vague and I am somewhat uncomfortable agreeing to it.

Thank you,

David

The LLVM Code of Conduct has been in draft mode for several years now. In order to finalize the Code of Conduct, there are 3 steps left to complete:

Hi Tanya,

I've added my comments to the documents, but I agree with David, we
should use the standard review tools we have for text, like we did
before. Google Docs not only needs agreements but it can also be
confusing to review if you're not used to.

The LLVM Foundation board will propose the initial members of this committee and provide a period of time to collect feedback from members of the community.

It's very important that this committee is diverse and inclusive. We
should be looking for inclusion in all areas: gender, ethnicity,
corporate/academia/hobbyist, sub-projects, disability, geographical
location, etc.

The CoC is not about code, but about behaviour and interpretation, so
the committee needs to be more than just heavy coders. But it can
directly affect coders, and subsequently, the code.

Therefore, it also cannot be random people from the Internet, just
because they have done something in other projects. They need to be
directly engaged into LLVM long term, either with code, research,
infrastructure, documentation, conferences, etc.

I'd also be more comfortable with a large number of potential people,
where sub-committees get selected as a small sub-set. If the same
people make the decisions for all cases, the bias would be incredible.
We need to make sure that we have enough people to be able to create
sub-committees with enough members and still have a balanced
(incomplete) block design, to ensure fairness.

Finally, my own personal issue is with mental health and disability.
How will you make sure that someone on the committee understands
(either through experience or clinical knowledge) the intricate
details of the common causes of misinterpreted behaviour in
non-neurotypical or with temporary/permanent poor mental health state?
Not many people identify themselves as clearly as I do, but they
suffer the same nonetheless. Without that covered, we may be
alienating an important part of our community.

cheers,
--renato

Hi Tanya,

Is there a reason for hosting things for review on Google Docs? We currently have both Phabricator and GitHub that work for review of any text-based format. When I click on a Google Docs link, I am asked to agree to a privacy policy that is very vague and I am somewhat uncomfortable agreeing to it.

David,

I understand not everyone wants to use Google docs which was why I gave an option to reply via email as well. My intention on using Google docs is really just based on my experience using it for docs that span different groups of people and it is what I primarily use for text documents.

I would like everyone to feel like they have a way to respond and provide feedback so please let me know if the options I have given do not work.

Thanks,
Tanya

I won’t be actively participating in this, but I will be read-only following it. I like github for this as well since it could be easier to track the precise questions/objections and track that flow to resolution. Compared to google docs which is like what a secretary might send to the boss, but I’ve never seen used to handle a technical issue like this. To me this is “code” and I’d love to see how questions are asked and how that discussion evolves until resolution. Like a github issue… Maybe google docs has this and I’ve just been missing it. There’s also the question of how transparent you want this process to be and if you want it to be accessible to every llvm developer. Telling David to just email you or making him agree to some Google privacy thing isn’t a very warm, friendly or productive start.

David - I’d propose to get a copy of things in their current state and just fork it to github. Then invite others to open issues against it.

I find this response quite offense. I am not a secretary sending an email to my boss. My intent is not malicious. I asked him if the methods I proposed were not sufficient as my goal is to include all input. Google docs is not an invalid way to collaborate on a text document. Is one way among many. It is not code in the purest form of the word which we can debate. But that is not the point. No format is perfect. Some people may be against github or Phabricator as the don’t want want an account there or don’t like it for text documents.

I would really appreciate some thinking that I am not just trying to be disagreeable or not trying to be flexible. I am not an evil person. I am actually trying to do a good thing and meet the needs of as many individuals as possible.

-Tanya

Hello everyone,

Tanya, thank you for putting the documents together, highly appreciated!

Both documents mention meetup organizers, so I think it makes sense to include them into the conversation explicitly, as not everyone may follow this discussion.
I’m cc’ing Arnaud as I think he is the right person to share the message.

Another point: both documents mention that meetup organizers should be contacted first, which is, imho, not the best option.
The meetup platform is a poor choice when it comes to communication (I missed some messages couple of times). Additionally, not every user group is hosted on meetup.com.
This implies that we should store contacts of organizers somewhere in a public space, which may not fit every organizer. Also, the contacts should be maintained and updated in a timely manner (e.g. an organizer stepped down, another organizer joined, etc.).
With this in mind, I’d suggest that reports are always sent to the conduct@llvm.org, and from there distributed to the organizers if needed.

What do you think?

Cheers,
Alex.

Hi Tanya,

I understand not everyone wants to use Google docs which was why I gave an option to reply via email as well. My intention on using Google docs is really just based on my experience using it for docs that span different groups of people and it is what I primarily use for text documents.

As I understand it from the pop-ups on the page, I cannot *read* a Google doc without agreeing to a somewhat invasive privacy policy. As such, an option to reply via email does not help me if my goal is to avoid agreeing to this policy.

Irrespective of your intent (and I strongly believe that this is not your intention) or the value of the tools, the message that you are sending is being received as 'you must agree to Google's privacy policy to participate in discussions that are core to the LLVM community'. Your replies in this thread are being perceived as the LLVM Foundation does not care and does not value the privacy of its constituency.

The CoC draft currently on the LLVM web site states:

> Be considerate. Your work will be used by other people, and you in turn will depend on the work of others. Any decision you take will affect users and colleagues, and you should take those consequences into account

During the switch to GitHub, there were very long discussions about moving to a tool that is blocked in some countries, is under the control of a single company, and requires signing up to a fairly complex set of T&Cs to use. The consensus (though not unanimous) view was that the benefits significantly outweighed the costs. This deliberation process took several months and the community heard many arguments on both sides.

A similar discussion took place with regards to Discord and the privacy policy there was seen as a reason that a sufficiently large proportion of the community would not wish to participate that it was a problem. There has been no such deliberation on the merits of Google Docs (or Office 365, or any other proprietary document hosting service).

If placing this on Google Docs was your personal decision and not any form of official LLVM Foundation policy, then I would ask that you reflect on what I have said and place it in a more inclusive location, such as a Phabricator diff to the existing CoC.

David

+1 to routing *everything* through a single entry point. The three
documents are already confusing enough with dates and processes, etc.

Related question: is this email going to a person, a group or a
ticketing system?

I'd strongly suggest we have a ticketing system. This way we can track
everything, update cc lists, create subgroups to handle requests, hide
access from CoC members, if they're involved, etc. These systems also
make it very easy to follow up, add resolutions, documents, links,
comments from committee members, closure type, etc. This makes
creating a report at the end of the year much easier.

It's really hard to do any that with groups (CC all-minus-person, have
some dropbox for evidence, track email body for keywords), and the
information will be lost, or at least, stored in a very confusing and
decentralised way. Worse still, when people leave the committee (or
the project), all that information, potentially including confidential
personal data, will be laying around in their mail servers.

I'd very strongly suggest this doesn't fall into a single person, for
obvious reasons. Even though there could be usually a single person
handling the requests, there should always be other people copied on
every new request.

cheers,
--renato

Hi, Tanya.

At this stage, I’m concerned that some time periods are not yet defined (e.g., “X amount of time”).

Also, I’m afraid that the process focuses too much on the reporter and the reportee, possibly bowling down to one’s word against the other’s. I’d be more assured if witnesses were encouraged when the situation allows for it.

Additionally, methinks that the process should proceed protecting the privacy of the parties until the final decision, including any appeal.

One thing that I wish were part of and a goal of the process would be to promote reconciliation between the parties, in addition to or in lieu of punitive measures, as long as both parties agree.

I have questions on the make up of the CoC as well, besides what Renato pointed out, but also about the rotation of its members, election and term duration. Is this spelled out elsewhere?

Thank you,

Hi Tanya,

I understand not everyone wants to use Google docs which was why I gave an option to reply via email as well. My intention on using Google docs is really just based on my experience using it for docs that span different groups of people and it is what I primarily use for text documents.

As I understand it from the pop-ups on the page, I cannot *read* a Google doc without agreeing to a somewhat invasive privacy policy. As such, an option to reply via email does not help me if my goal is to avoid agreeing to this policy.

Irrespective of your intent (and I strongly believe that this is not your intention) or the value of the tools, the message that you are sending is being received as 'you must agree to Google's privacy policy to participate in discussions that are core to the LLVM community'. Your replies in this thread are being perceived as the LLVM Foundation does not care and does not value the privacy of its constituency.

The CoC draft currently on the LLVM web site states:

Be considerate. Your work will be used by other people, and you in turn will depend on the work of others. Any decision you take will affect users and colleagues, and you should take those consequences into account

During the switch to GitHub, there were very long discussions about moving to a tool that is blocked in some countries, is under the control of a single company, and requires signing up to a fairly complex set of T&Cs to use. The consensus (though not unanimous) view was that the benefits significantly outweighed the costs. This deliberation process took several months and the community heard many arguments on both sides.

A similar discussion took place with regards to Discord and the privacy policy there was seen as a reason that a sufficiently large proportion of the community would not wish to participate that it was a problem. There has been no such deliberation on the merits of Google Docs (or Office 365, or any other proprietary document hosting service).

If placing this on Google Docs was your personal decision and not any form of official LLVM Foundation policy, then I would ask that you reflect on what I have said and place it in a more inclusive location, such as a Phabricator diff to the existing CoC.

To be clear, we (LLVM Foundation Board), reviewed the email and the use of Google
docs, this was not Tanya's personal decision.

Our intention was not to exclude anyone. We have heard your feedback and will
work on making the document accessible in some other way.

-Tom

Hi Tanya,

I understand not everyone wants to use Google docs which was why I gave an option to reply via email as well. My intention on using Google docs is really just based on my experience using it for docs that span different groups of people and it is what I primarily use for text documents.

As I understand it from the pop-ups on the page, I cannot read a
Google doc without agreeing to a somewhat invasive privacy policy. As
such, an option to reply via email does not help me if my goal is to
avoid agreeing to this policy.

I wonder what kind of popup you got? I opened the link in a “private window” of my browser, I am not logged into any Google service, and the doc just opens directly without any popup. I can also comment “anonymously” on the document apparently. I would expect that the experience I relate here is “privacy friendly”? (I’m curious why you didn’t get the same, maybe connecting from Europe imposes the popup?).

I wonder what kind of popup you got? I opened the link in a “private window” of my browser, I am not logged into any Google service, and the doc just opens directly without any popup. I can also comment “anonymously” on the document apparently. I would expect that the experience I relate here is “privacy friendly”? (I’m curious why you didn’t get the same, maybe connecting from Europe imposes the popup?).

+1. No popups, I can view and comment on the doc from a private session from Berlin, Germany.

I think his point is that he hasn’t agreed to any goog anything. You’re probably signed in with your gmail and or maybe you previously signed in with your gmail.

I had a similar experience with Google Docs. I (or rather, the people I
support) was sent a document and we just couldn't access it full stop.
As soon as we logged in with a google account, the document appeared.

Cheers,
Wol

Plus one. Take a look at the recent CoC thread on lilypond-devel

https://lists.gnu.org/archive/html/lilypond-devel/2020-02/msg00387.html

(sorry, there are several thread-breaking email clients there ... :frowning:

You'll see the issue there is that (a) the lead developer - an NNT - saw
it (the punishment stuff) as a personal attack, and (b) that like *most*
NNTs, his natural reaction to conflict is to walk away. The result would
likely be serious collateral damage to the project.

My personal take is that if you punish someone for WHAT they are, it's
discrimination. End of. Including NNTs. If someone *chooses* to be an
asshole then ban them. But if an NNT doesn't realise it then you need to
be a mentor and fix it - again, NNTs usually want to put things right.

And given that NNTs tend to be the most productive, most caring, most
conscientious workers then yes you will be alienating the most important
part of the community.

Cheers,
Wol
(self-diagnosed Aspergers)

I think his point is that he hasn’t agreed to any goog anything. You’re probably signed in with your gmail and or maybe you previously signed in with your gmail.

Not really, it’s a private session (“private window”), I’m not logged in.

I had a similar experience with Google Docs. I (or rather, the people I
support) was sent a document and we just couldn’t access it full stop.
As soon as we logged in with a google account, the document appeared.

This could definitely be the case depending on the access rights. The documents Tanya shared are accessible for anonymous, not-logged in users.

The LLVM Code of Conduct has been in draft mode for several years now. In order to finalize the Code of Conduct, there are 3 steps left to complete:

Hi Tanya,

I've added my comments to the documents, but I agree with David, we
should use the standard review tools we have for text, like we did
before. Google Docs not only needs agreements but it can also be
confusing to review if you're not used to.

The LLVM Foundation board will propose the initial members of this committee and provide a period of time to collect feedback from members of the community.

It's very important that this committee is diverse and inclusive. We
should be looking for inclusion in all areas: gender, ethnicity,
corporate/academia/hobbyist, sub-projects, disability, geographical
location, etc.

The CoC is not about code, but about behaviour and interpretation, so
the committee needs to be more than just heavy coders. But it can
directly affect coders, and subsequently, the code.

Therefore, it also cannot be random people from the Internet, just
because they have done something in other projects. They need to be
directly engaged into LLVM long term, either with code, research,
infrastructure, documentation, conferences, etc.

I'd also be more comfortable with a large number of potential people,
where sub-committees get selected as a small sub-set. If the same
people make the decisions for all cases, the bias would be incredible.
We need to make sure that we have enough people to be able to create
sub-committees with enough members and still have a balanced
(incomplete) block design, to ensure fairness.

Finally, my own personal issue is with mental health and disability.
How will you make sure that someone on the committee understands
(either through experience or clinical knowledge) the intricate
details of the common causes of misinterpreted behaviour in
non-neurotypical or with temporary/permanent poor mental health state?
Not many people identify themselves as clearly as I do, but they
suffer the same nonetheless. Without that covered, we may be
alienating an important part of our community.

I appreciate the feedback on forming the committee. I have not written down any of my thoughts yet for review, but I share many that you brought up here. I’ll try hard to be transparent as possible with this process and share more details when I have them written down.

Thanks,
Tanya