[unrev-II] The Need for Maleable Archives

From: Eric Armstrong (eric.armstrong@eng.sun.com)
Date: Tue Jul 24 2001 - 13:09:43 PDT

  • Next message: Rod Welch: "[unrev-II] Maleable Archives"

    In the discussion I had with Lee last week, a couple
    of thoughts came out that I would like to post to a
    wider audience. (As usual, I'm thinking in email. I'll
    post as an HTML page later.)

    The core is this: The concept of an eternal and unchanging
    archive of exactly what was expressed, when, and by whom,
    is highly overrated, in my opinion. In my opinion, we think
    that way because we are only used to two options: full
    history, and no history. Of the two, we prefer having a

    Let's consider that we have a collaborative email-ish
    discussion going on, most likely based on an IBIS-style
    investigation of issues, and that we have our dream data
    engine (Nodal, in all liklihood) powering it.

    Now, lets consider a few cases.

    CASE #1: Wrong audience.
      I post a message to the group that was intended for an
      individual. Or I accidentally post to the wrong group.
      As the author of the message, I have a perfect right to
      press the "retract" button, as it were, and remvoe the
      noise from the wires.

    CASE #2: Wrong assertion.
      I post a message that says, "A is true", and someone
      replies, "no it's not", with canonical proof that I
      am wrong. Well darn it all, I want to remove that
      hare-brained, addle-pated, wrong-headed, idiotic claim,
      so I don't look a fool for all eternity. And as its
      author, I have a right to do that!

      Now maybe I just have to file the "desire to delete",
      and when the person responding takes away their
      contradiction, then the whole thing can disappear.

      Or maybe I can at least edit the thing to say that
      "you'd think, on the surface, that A is true, but as
       Ralph so alerty points out, ..." Or words to that
       effect. Details to be decided. But you get the idea.

    CASE #3: Civil Discourse.
      Here, tempers flare. I say, "you rat-brained son of a
      skunk". He/she says, "you pig-eating, flaming goose
      neck", and we're off to the wars. After a while,
      everyone calms down. I edit my message to read, "I'm
      not sure you're right about that". He/she edits their
      message to read "I have to disagree". Later on, reading
      the archive, the information content is the same. It is
      only the emotions that have been extracted.

    The bottom line here is that the overall readability and
    usability of the archive is improved as a result of the
    editing. At the moment, an archive is a huge field of
    wheat, with needles strewn about. You search the field
    looking for needles, and of course you spend a lot more
    time "hopeless separating the chaff from the chaff".

    Message threads begins to organize the wheat into haystacks.
    But as we all know, often a message that belongs in a
    thread is posted outside of it. The ability to edit the
    archive makes it possible to put it where it belongs,
    which reduces the number of haystacks and makes them a
    more accurate reflection of the dialog.

    Similarly, removing material helps to reduce the size of
    the haystack. Both reorganization and removal help you
    narrow your search and make it more effective.

    The alternative -- the unedited archive -- is a mass of
    detail that no one ever sees. If you can link to it --
    but you can't find anything in it, and you don't even
    want to start looking, then the archive loses much of its
    potential value.

    The archive, should, over time, become a "narrative
    history" of the discussion. What options did we consider,
    which ones did we reject and why. What did we learn. What
    did we end up with, and why.

    Like any narrative -- a story told in a novel or over a
    campfire -- it benefits from revision and editing. Rough
    drafts are simply not all that much use, except to the

    That leads to the first of two arguments in favor of fixed
    archives -- the value of maintaining a complete history.
    The second argument is over the possibility of broken links.

    COUNTER ARGUMENT #1: History
      For the process-gurus, a history is invaluable. How did
      we get here? How did that group *actually* arrive at
      those conclusions. For such folks, the fixed archive, IN

      Tracing the evolution of the archive lets them see when
      tempers flared, how things were retracted, the deals that
      were made. "I'll remove my comment if you'll remove yours",
      etc. Then, when the whole ball of wax disappears, the
      process folks can dissect the social processes that made it

      That kind of archeaology can certainly be valuable. And
      it makes it worth considering retaining the original
      archive via versioning, as the edited archive is constructed.
      But it is important to remember that for everyone *except*
      the process historian, it is the readable version of the
      archive that is most valuable -- it contains the distilled
      collection of "knowledge nuggets", with the minimum of
      excess material.

      There is a space consideration, though. Since the archiving
      and versioning consumes additional space, resource constraints
      may play a role in the deciding whether or not to retain
      the unedited archive.

      Then, too, when I accidentally post a note that was intended
      for my girlfriend, I unequivocally reserve the right to
      remove it, I care not what!

      This kind of case happened just the other day.
      At home, I have the alias "sun" for my work address. At
      work, I have (HAD) the alias "sunstatus" for the folks who
      stay abreast of what I'm working on. I was creating an alpha
      test-list for my learn-by-ear, see-how-its-played tune teaching
      program, and mailing it back and forth between the two accounts
      as I thought of additional people to put on the list. At work,
      I typed "sun" without thinking, and autocompletion expanded it
      to "sunStatus" and mailed it to the gang of folks I do projects
      for. Not good! In such cases, "retract" is really necessary.)

    COUNTER ARGUMENT #1: Broken links
      The second counter argument concerns broken links. That is
      of course a serious technical consideration. What if I have
      a pointer to a node, and that node has disappeared? Given a
      data engine like Nodal, I believe that the issue is largely

      First, note that Nodal specifies that the default linking is
      to the latest version of a node. So when a node is replaced
      by an edited version, no broken link results.

      Second, since the back link facility keeps track of who is
      has pointers to the node, the engine knows when the node can
      be retired, or when it needs to be kept around.

         Implementation note:
         The node would be marked as deleted. If no links exist to
         the node, it would be targeted as a *candidate* for
         removal. But there could still be external links who have
         not "synced" with the system recently. The "ok to eradicate"
         process would have to work like this:
         a) For each node, maintain a list of folks to whom
             he node is potentially visible.
         b) Maintain a master list of delete candidates.
         c) As each person syncs up, check the master list. For
            any node on it they haven't referenced, remove them
            from the potentiallyVisibleTo list.
         d) When the potentiallyVisibleTo list is empty, the node
            can be garbage collected.

    The value of an archive, like the value of a novel, depends on
    its *readability*. Editable archives allow for discussions
    that, over time, reflect the best possible reasoning and
    explication of issues. Whether or not an unedited version is
    retained underneath, it is the edited version which should be
    the topmost, publically visible layer.

    ------------------------ Yahoo! Groups Sponsor ---------------------~-->
    Small business owners...
    Tell us what you think! http://promo2.yahoo.com/sbin/Yahoo!_BusinessNewsletter/survey.cgi

    Community email addresses:
      Post message: unrev-II@onelist.com
      Subscribe: unrev-II-subscribe@onelist.com
      Unsubscribe: unrev-II-unsubscribe@onelist.com
      List owner: unrev-II-owner@onelist.com

    Shortcut URL to this page:

    Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/

    This archive was generated by hypermail 2b29 : Tue Jul 24 2001 - 13:36:23 PDT