Shared publicly  - 
Wikipedia is a great example of the difficulties in maintaining a positive culture over time in the face of growth. What marks them out especially is the way they recognise the problem, quantify it, and are trying to fix it.
Dave Neary's profile photoSimon Phipps's profile photoElin Waring's profile photo
You think they do? I see them trying to train newcomers to survive the negative culture rather than taking steps to correct it.
I see a cross between both - "there's a problem - let's measure to confirm", followed by "why is this happening?", and now there are two camps - as you say, help people understand why things are the way they are (doomed to failure IMHO) and try to figure out how to change the culture for the better.
Key finding imo (not from that paper)

How has communication to new editors (on their user talk pages) changed over time?
Key conclusions
Since 2004, there has been a significant drop in messages including praise and thanks, corresponding with an increase in the overlap of teaching/instructional communication with criticism.
Currently, about 80% of messages to new Wikipedians come from bots or semi-automated editing tools like Twinkle and Huggle. (see charts)
Currently, about 65% of the communications to new Wikipedians are warning templates on their talk pages. (see charts)
Honestly I think that's a distraction. The problem I and so many others have found is not with bots; it is with established members of the community who make engagement by outsiders a nightmare. Wikipedia has a stable community of people who mutually support each other's toxic pedantry. They use community rules to sanction and control outsiders, and on the rare occasions those outsiders fight back gather community support to exclude the outsiders. This feedback loop of sociopathy is the problem, not user education or out-of-control bots.
I think it's fascinating that they have done so much analysis of aspects of this, yet you're right that the focus is on obvious, easy to fix things like what is the message that bots send, rather than harder to fix things like how to get past insider/outsider issues. Those are basically inevitable, but what doesn't have to be inevitable is whether you work consciously to name and counter them.
Add a comment...