On 1/13/19 7:13 AM, Sharan Foga wrote:
Hi All

A while ago I mentioned that I was doing a paper based on the ASF and the 
Apache Way culture

https://s.apache.org/XLXi

Well it is done and for those of you not following the Kibble mailing list 
(since I used Kibble as my main research tool :-), here is a link to my paper 
on the Kibble wiki

https://s.apache.org/VTAy

It’s over 30 pages long so for those of you who don’t want to read that many 
pages, I’ve also created a wiki page without all the nice Kibble graphs :-)  
that summarises the main points from the paper here.

https://s.apache.org/ESEh

Huge thanks to Shane for agreeing to be one of my supervisors, and a shout out 
goes out to the Kibble community for creating a tool that provides such a 
variety of interesting and important community related information.

Please feel to take a look and respond with with any comments or feedback.


Hi Sharan,

Thanks so much for sharing.  I think you did a very nice job presenting the culture and you are asking the right basic question.

Before getting into the analysis portion, I have one comment on the culture description part that I think deserves mention.  Under "merit" you have "People contribute to the ASF as individuals" and in a sense this fully captures our "diversity" requirement; but at least as an assumption the idea that our projects are independent (not controlled by external entities) should be called out somewhere.

Here are some things to think about relative to the analysis portion.

First, the pony factor is an interesting measure, but It's a stretch really to see it as indicative of cultural transmission. I know you are not really claiming that and it may be a nicely correlated measure; but without a deeper analysis of actual understanding / "acculturation" among volunteers in communities with high PF values, I am not sure what it means.  It is quite possible, for example for a fast-growing company to pay a lot of committers and efficiently divide up work among them so the PF goes way up but the community is effectively closed to outsiders. It's also possible that one or two committers to make many trivial commits pushing that statistic down.  Again, on average, it is an interesting statistic; but all that it measures is how commits are distributed.  You are right that high PF indicates growth in the committer base (or at least stable dispersion of commits); but that's all that it indicates.   There are some historical examples where high PF was not correlated with successful acculturation.  I will mention only the dead one:  Jakarta.  That project was wildly successful in attracting contributors and commit was not that hard to get; but "we" were pretty clueless :)

The text analysis is also interesting, but I am afraid that just extracting key phrases and sentient analysis may not accurately reflect cultural transmission / evolution.  An interesting follow-up might be to select some projects that appear to be in good shape in terms of culture from your measures and others that appear not so good and really dig into the lists, JIRAs, board reports, etc. to see if the same picture emerges.  In other words, integrate qualitative analysis of the project communications (and contribution history) to validate that your empirical results are valid and / or to better train your textual analysis algorithms. You seem to be headed in that direction in some of the suggestions for further research.

Thanks again for sharing and for stimulating us thinking about how our communities are growing.  One final comment is that while the basic ideas that you have at the beginning are not likely to change quickly, we need to make sure that allow our culture itself to grow and evolve.  That's another reason to be careful about developing empirical measures around how our communities embody the culture.

Phil


Thanks
Sharan


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@community.apache.org
For additional commands, e-mail: dev-h...@community.apache.org




---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@community.apache.org
For additional commands, e-mail: dev-h...@community.apache.org

Reply via email to