Improving community health

discussion about organizational roles

Rewards

Communication and support

Help

Usability and interface

Outreach and user conversion

Hmmm ... proposals I don't like, but demonstrate understanding about community problems

Other favorites

Research and fact base

  • Participation/Attracting new participants and retaining existing participants
  • Participation/Participants of Wikimedia projects
  • Participation/Drivers of participation
  • Quality/Quality
  • Key questions/question list
  • Interviews/Summary of interviews
  • Wikimedia Report Card - August 2008-2009
  • Wikipedia: A quantitative analysis. <-- Great scholarly paper densely packed with information
    • Number of authors, articles, and contributions: the amount of authors and contributions have stabilized
      • "the total number of logged authors in (the English Wikipedia) overwhelms that of any of the (other languages) in this study, which contributes to reach this unusually high number of user discussion pages" (page 74)
      • "In 2007, the total number of contributions becomes stable in all versions" (page 76)
      • "in all language versions, the most common length of standard articles is situated around 1,5KB, indicating that this might be considered as the expected length of a Wikipedia article." (page 80)
      • "the length of articles tend to increase as more different articles revise them, but there exists no direct correlation between these two variables" (page 88)
    • Talk page activity: contributions to talk pages have stabilized, but we continue to see more and shorter active talk pages
      • "Figures 4.14 and 4.15, show respectively the evolution of monthly number of revisions received by talk pages, and the number of active logged authors in talk pages per month. The same deceleration in the steady growing rate of early years, already found in previous graphs, happens here too." (page 88-89
      • "Nevertheless, if we inspect Figure 4.16, depicting the number of active talk pages, we find a different situation. The number of active talk pages in all language versions has continued its steady growing trend, even in 2007. This a completely new, and unexpected pattern." (page 89)
      • "outstanding proportion of talk pages per article found in the English Wikipedia, well beyond any limits reached by the other language versions" (page 90)
      • "Contrary to what we saw in the case of articles, the median of the length of talk pages tend to become lower as the language versions evolve over time." (page 95)
    • Author and article inequality: in all languages, most activity comes from 'core' users, on a slightly larger set of popular articles
      • "the top ten language editions maintain a very skewed distribution, with less than 10% of the total number of authors performing more than 90% of the total number of contributions received by each version" (page 106)
      • "The graph shows that there exist very little differences in the inequality level exhibited by all communities under study, showing highly biased distributions towards a small core of very active logged authors in each language version" (page 114)
      • "The distribution of revisions among articles is more balanced, though there is a slight bias towards a group of more popular articles" (page 115)
    • Life cycle of authors compared to "core" authors: this is too interesting to summarize. read it all.
      • "for all language versions ... the increment in the monthly rate of deaths .. has overcome the number of births per month, from 2007 on" (page 116)
      • "Figure 4.36 shows that English core members are the fastest abandoning the project after leaving the core ... On the opposite side, former core members of the German still remain very active once the left the very active group of logged authors" (page 119)
      • "once the core author has surpassed the 100 days threshold, the probability that she remains in the core for a longer period of time raises dramatically, in all language versions." (page 121)
      • "the individual effect of contributing to FAs or talk pages do have some influence in enhancing the longevity of logged authors in the system. But the definitive improvement in authors lifetime is only registered for authors who both edited in FAs and participated in talk pages, at the same time" (page 121)
      • "It is interesting to see that the risk (of death) for young authors decrease following a log-linear pattern, for an age of less than 15 days, except for the English and German Wikipedias. The remarkably lower risk for young authors in these language editions might be a logical cause behind the more active production pattern in these language versions" (page 126)
      • "for all language editions, the restricted mean survival time to reach the core of very active contributors is concentrated around 200 days. Once the reached the core, the restricted mean of the time of membership (blue discontinuous line) varies between 200 and 400 days" (page 128)
      • "50% of authors who eventually joined the core in the top ten Wikipedias needed between 80 and 130 days to reach that status. On the other side, 50% of core members left the core after a period oscillating between 20 and 110 days" (page 129)
    • Patterns of FA contributors: established editors are the main contributors to FAs, but their growth pattern has stabilized just like all editors
      • "the lower population of younger authors has almost dissappeared completely, revealing that older authors are the main creation force behind the content revision process in FAs for all language versions." (page 135)
      • "It seems that the 'big crack' in the logged authors population, in summer 2006, had also a great impact in revisors of FAs as well" (page 140)
    • Lifecycle of Wikipedias: More articles are being handled by fewer editors
      • "the core of very active authors in progressively taking over a larger proportion of the creation process, though the differences among distinct years is not very significant" (page 149)
    • How does this help the community health task force?
      • The "health" analogy overlaps with a lot of the measurements from this study.
      • To improve community health, we need to:
        1. Reduce "infant mortality". (Increase uptake for Wikipedians who never really contribute, who leave within 15 days.)
        2. Reduce "child mortality", for those never reach "working age". (Increase uptake for Wikipedians who don't become core users, who leave within 200 days.)
        3. Increase "life expectancy" of working population. (Increase number of days that core Wikipedians, older than 200 days, stay with the project.) <-- Reduce "occupational hazards"?
        4. Increase "births". (Outside our scope. The reader conversion task force is focusing on this.)

To do list

On my radar

  • Empirical: Push for survey (in progress)
  • Empirical: Push for tools to study some contributions pages, looking for patterns (in progress)
  • Empirical: Do one last look for quantitative studies on Wikipedia
  • Empirical: Look for studies on dispute resolution, especially multi-party dispute resolution
  • Community: Go over proposals one last time. Be more sweeping this time.
  • Logical: Start a brainstorm for possible threats to community health
  • Solutions: look for research from conflict resolution. Especially data that distinguishes two-party from multi-party disputes.

Empirical approach

  1. Gather data
    • Secondary research
      • Find more data on factors that cause new users to leave
      • Find more data on factors that cause veteran users to leave
      • Look for a relationship between policy and community patterns
      • Look for a relationship between disputes and community patterns
      • ... these have not been easy
    • Original research
      • Survey editors as to why they are leaving
      • Examine contributions pages to find patterns
  2. Identify main problems
  • Pros: Accurate, uncontroversial
  • Cons: May not have enough data, might lead to narrow solutions

Community approach

  1. Gather data
  2. Identify main problems
  • Pros: Respect for Wiki process, leverages community creativity
  • Cons: Accepts beliefs of small group of contributors as facts

Logical approach

  1. Make assumptions
    • What might cause new users to leave?
    • What might cause veteran users to leave?
    • What else might interfere with an effective community?
  2. Identify main problems
  • Pros: No research, can work passed gaps in knowledge base
  • Cons: Problem identification may be determined by politics

Final steps

  1. Turn findings into 2 to 4 problems
  2. Turn problems into 2 to 4 solutions
  3. Expand solutions into detailed recommendations
    • Brainstorm ideas
    • Gather ideas from existing proposals
    • Browse qualitative research for insights
    • Support with data

Goals

  • Find causes of veteran burnout, and reduce their impact
  • Find causes of new user rejection, and reduce their impact
  • Make Wikipedia a nicer place to do otherwise hard work

Outside scope, but complementary goals

  • Task force/Reader Conversion
    • Improve outreach to potential users with other motivations for participation
    • Improve outreach to potential users from other demographic backgrounds