[llvm-dev] RFC: Introducing an LLVM Community Code of Conduct

[Joachim Durchholz via llvm-dev]

Am 13.10.2015 um 17:12 schrieb Robinson, Paul via llvm-dev:
>> Am 13.10.2015 um 14:02 schrieb Renato Golin:
>>> On 13 October 2015 at 12:37, Joachim Durchholz via llvm-dev
>>>> Just curious: Do people ever do it the other way around?
>>>
>>> What do you mean?
>>
>> Report first, then try to resolve directly.
>> My experience is that people try to resolve things directly and ask for
>> help only when they find it doesn't work, but YMMV.
>
> Yes, people sometimes report first. And some might *not* try to resolve
> directly, for various reasons. (Unable to deal with the event calmly;
> feeling generally uncomfortable with confrontation; etc.)

Actually that would be valid reasons to report first.

Oh. New idea.

Avoid the term "report"; it implies a firm stance that the other side is 
misbehaving. Which means a kind of "invoking authority to triumph over 
my opponent", a mentality that should not be encouraged at all.

Instead, use the term "arbiter". It implies "we're in disagreement, 
please help us resolve this". It doesn't change anything in the process 
itself, but it frames it in an entirely different context. (Framing 
strongly influences decisionmaking, see 
https://en.wikipedia.org/wiki/Framing_effect_%28psychology%29 .)

 > I sympathize
> as I feel uncomfortable with it myself. Going through an intermediary
> does run the risk of leaving the offender clueless as to what the actual
> problem is, though (I've been there too).

I guess the "call in an arbiter" approach would have prevented that 
because it would have triggered the committee member into asking all 
sides and looking for a consensus, instead of simply looking whether 
some interaction was "too much".

HTH
Jo