Thanks for starting this conversation. I also agree that it is properly better to start with these policies now, and discuss and agree on them calmly, then get into an actual problem and then create policies in the heat of the moment.
Three things come to my mind when I think of Wikidata and community behavior:
1) Although I hear the community being called out a lot for how awesome we are - and I think the community is pretty awesome - I am repeatedly negatively surprised by the tone in the Project Chat and some discussions. Often these are specific contributors in their discussions to each other that have a harsher tone. I am not particularly fond of this harshness, and would prefer if we all get friendlier - but at the same time I certainly don't want to see a tone police being set up. I try to be a positive example in my interactions.
2) In general, because Wikidata is more about structured data than about prose text, a lot of nuance is lost in Wikidata in the first place, which reduces the area to attach controversies to quite a bit. The discussions, when they get heated, are rarely on the level of single items, but rather on a higher level, such as the usage of a specific dataset, or what a property means. It might be that due to the loss of nuance we need, in general, less rules per contributor than we need on most Wikipedias, and maybe less bureaucracy in comparison to other wiki projects. I think that is a good thing, and that tends to put me in the "fluid enforcement" of rules instead of the "strict and organized" camp. This also because I don't want to burden anyone with such a role.
3) Also, our edit in the content namespaces to edits in the other namespaces ratio skews much more towards the content namespaces than it does in many other wiki projects. Which I find is great. I call this the do / meta ratio. This also leads to less conflicts, which is neat. I think this also comes from the fact that we have people collaborating across languages without understanding each other - and without the need to do so. This works a lot of the times.
So, I think that kinda tells us why we got so far and so big without having behavioral norms being instituted already. But we will need some rules, and I am super thankful for having this conversation started.
Regarding where to take the rules form or if we should implement it ourselves: as discussed above, our do / meta ratio skews heavily towards do. This means we don't have that many resources that we are spending on meta and creating such rules. Which is why I pragmatically think that stealin/d/d/d/d/d-adapting rules form other communities is the way to go. Just because we are lazy and we really don't want to discuss and put the effort into creating our own rules. But if anyone wants to actually commit to the effort of creating our own rules, sure, go ahead!
So here are a few behaviors I think should be covered, and are partially already covered:
- no stalking, outing, doxxing, attacking, harassing, threatening, etc. of other contributors.
- remember that contributors are common humans. Remember that we are all doing this in order to benefit humanity. We don't want to burn our fellow contributors, we don't want to intentionally anger them, distress them, annoy them.
- remember that our contributors might be very diverse. A project like Wikidata may find people who are very different from each other being interested in contributing. These people, us, might have very diverse backgrounds, with regards to where we come from, who we are, what we believe, how we think, what interests us, what motivates us, and what discourages us in the many forms of discouragement, from demotivation to anger. Let us not assume that all of us are like ourselves, and let us not build barriers towards collaboration for those that are different. Abusing these differences for attacks or harassment should not be allowed.
- if bots and humans get into a 'fight', humans should have an advantage. Using a bot to inconsiderately overwrite human contributions, particularly repeated human statements, should be bad style and discouraged.
- no intentional misuse of Wikidata in order to proof a point or to deliberately hurt other sites or projects using our data, even though, narrowly seen, the action is legitimate within Wikidata (this, in particular, is meant to protect the Wikipedias from suddenly displaying misleading data just because their queries might be off or something - go and help fix their queries first, discuss the situation with them, and only then make the changes). We should be aware of the great power we yield on other projects, and should regard this as a token of trust from those projects, and should not deliberately abuse this power.
- no deliberate use of bad sources and then not marking the statement accordingly. Seriously.
There are probably a few ground basic truths that I missed and that we missed so far. I would like if there is some experienced, e.g. collected by the WMF team, about which points should be covered, and then we can start walking around these. This can be ongoing in parallel to the collection of use cases below by Ijon and Spinster for positive and negative behavior.