[SharedKnowing] Wiki lessons learned

Ben Kovitz bkovitz at acm.org
Sun Dec 30 19:04:27 CST 2007


Inspired by Larry's example questions on
http://mail.citizendium.org/mailman/listinfo/sharedknowing#more, I'll
propose six lessons learned from the last eight years or so of wikiing.

1. Error and imperfection drive wikis.  Preventing error kills wikis.

    When someone posts a factual error or poorly worded description, that
    motivates others to correct it or improve it.  A perfected, polished
    piece of writing does not trigger improvement.  Trying to prevent
    poor or inaccurate text from ever getting into the system would
    undermine collaboration before it starts.

    Wikis are peculiarly positioned to turn disagreement into growth:
    instead of one party backing down, or merely reverting a poor edit, a
    good editor will incorporate what's good in an imperfect edit into 
new
    content or structure that no one previously anticipated.  This is the
    essence of what makes wikis so fertile.  Text on a good wiki is not a
    compromise, but true synergy.

    Consequently, wikis trade off between reliability and fertility.
    Wikipedia is not completely reliable, but it is this very
    unreliability that has triggered its unprecedented growth of useful
    content.

2. Wikis are bad for resolving disagreements or determining consensus.
    "The wiki renders no verdict."

    Because any individual can change anything, a fair representation of
    a consensus is impossible to achieve.  Edit wars happen as people try
    to make the text speak an opinion for the whole community.  We can
    see this principle at work on Wikipedia where the pages about the
    most controversial pages are plagued by edit wars and low quality.

    Some ways to resolve this are: limit the topics only to factually
    indisputable matters ("NPOV" on Wikipedia) or simply declare that the
    wiki renders no verdict on anything but has space for all opposing
    ideas.

3. DocumentMode should be separate from ThreadMode.

    That's now-archaic terminology for "the end product that people are
    collaborating on" (e.g. encyclopedia pages) should be separated from
    people's signed statements in the first person (e.g. discussion 
pages).

    To put this another way, the glory of wikis is that there is a truly
    collaborative product that no one owns, rather than people making
    first-person statements that they own individually.  c2 is pretty
    much all ThreadMode now--which is to say, people shouting back and
    forth at each other, and not really collaborating.

    This anonymity or cross-authorship makes it impossible for wikis to
    support pedigreed statements (this credentialed person said so,
    therefore the statement is reliable).

4. Free-form text is especially good for collaboration: infinite degrees
    of freedom, but it's easy for anyone to make small changes.

    More-structured media like SQL databases offer less wiggle room for
    people to grow the data in unanticipated directions.  Disagreement in
    such a medium can only be resolved by someone backing down, the
    opposite of wiki-style growth in new directions (#1).  Extremely rich
    media like video don't allow small changes to be made without
    completely reworking the whole.

5. Intervention from above should be very sparing.  The growth and
    direction of a wiki are not very controllable from the executive
    level.

    The more control a leader has, the less the wiki can grow
    spontaneously into something beyond what any planner can fully
    envision.  As an example, I think here of Richard Drake, founder of
    the Why Clublet wiki in 2000.  When the wiki drifted beyond his
    ability to make sense of it, he clamped down control and ultimately
    threw nearly everybody off to rescue his vision.  The resulting wiki
    became stale: just one person's self-publishing forum, lacking
    "wiki-magic".

    This need for a light touch at the executive level results from the
    unpredictable way growth of a wiki is triggered by error (#1).

6. A wiki's growth has two principal "cusps": reaching a critical mass
    when it takes on a life of its own (beyond the founder's personal
    friends and ability to control the vision), and another one where
    people show up who violate the social norms of collaborative writing
    and threaten to destroy it.

    The many small wikis that contain just a little haphazard content are
    ones that never reached critical mass.  Ward Cunningham's original
    wiki at http://www.c2.com/cgi/wiki?RecentChanges illustrates a wiki
    long past the stage where the riff-raff showed up and ended the
    culture of mutual respect.  Wikipedia is a great success, in large
    part because when the riff-raff showed up, a strong-enough group of
    citizens enacted rules and set up dispute resolution--alienating some
    people but preserving the cooperative spirit of the wiki.

I doubt that everyone will agree with these.  I'd love to hear some
counterexamples--or, if you do think these are genuine lessons, perhaps
an interesting illustration of them.  Better yet, I'd love to hear other
lessons you've learned from watching or participating on wikis.

Ben Kovitz



More information about the SharedKnowing mailing list