[SciPy-Dev] establishing a Code of Conduct for SciPy

[Matthew Brett]

Hi,

Back to the code of conduct discussion, Nathaniel has raised a
pertinent theme over at the Scipy PR - main comment at:

https://github.com/scipy/scipy/pull/7963#discussion_r145580285

Nathaniel's basic point, as I understand it, is that one common type
of behavior that we should be able to deal with, is flagrant and
aggressive abuse, likely from people otherwise not involved in Scipy.
He gives this example "Last night someone logged into the #scipy
channel on Freenode and started pasting racial slurs in giant
letters.".

Nathaniel then goes on to argue that the language and procedures in
the CoC as stands don't apply to that case.

I think that's reasonable, but I think we have to be careful to distinguish:

1) obvious flagrant abuse, likely from someone who does not
contribute, possibly from someone who does contribute who is having a
breakdown of some sort and
2) discussions that started in good faith and have gone out of control.

It's true that the current code of conduct is aimed more or less
squarely at the second.

I don't personally think we're going to have too much trouble
distinguishing these two cases, so I'm going to suggest that, instead
of switching the doc to aiming at case 1 rather than case 2, we have a
safely-valve mechanism for case 1.   This would go something like:

"""
As a special case, we know that it is painfully common for internet
communication to start at or devolve into obvious and flagrant abuse
including violent, racist and sexist language.   In the specific case
of violent, racist or sexist language, these {named moderators} will
use the following procedure:

* immediately disconnect the person from all Scipy communication channels;
* if the originator appears to be a previous contributor, the
moderator may try to contact the contributor by some other means to
check whether their account has been hacked.
* if the originator is in fact a previous contributor, and the
contributor wants to be reconnected to the Scipy channels, then
{consider some cooling off period, an agreement not to repeat the
behavior, and email moderation.  A previous contributor also has the
right to an appeal to the code of conduct committee}.
* in every case, the moderator should make a reasonable effort to
contact the originator, and tell them specifically how their language
qualifies as "violent, racist or sexist language", and they should
copy this explanation to the code of conduct committee.  The code of
conduct committee should formally review and sign off on these cases
every year to make sure this mechanism is not being used to control
ordinary heated disagreement.
"""

I've argued before [1] that the best way to think about these
documents, is in terms of specific use-cases.  In Nathaniel's case
above, I think it's fairly obvious how the mechanism above would work.
  Next we consider the famous case of the sexist joke on the Ubuntu
mailing list [2].  I think that would also qualify for the mechanism
above, but where we would expect the resolution to be that the
originator would have to agree not to post sexist material to that
list, and be moderated for a while,   Last we consider the SpacedGirl
Software Carpentry Case [1], where this procedure could not reasonably
be invoked, and the rest of the current code of conduct would apply.

Cheers,

Matthew

[1] https://github.com/jupyter/governance/pull/23#issuecomment-269244281
[2] http://geekfeminism.wikia.com/wiki/Ubuntu_Code_of_Conduct_incident