tools-yak
[Top] [All Lists]

[tools-yak@collab] grass roots peer review experiment

To: tools-yak@xxxxxxxxxxxxxxxxxxx
From: Eugene Eric Kim <eekim@xxxxxxxxxxxx>
Date: Thu, 24 Jul 2003 23:16:53 -0700
Message-id: <20030725061653.GC5058@douge.blueoxen.net>
As I reported on the yak@collab list, we had an informal gathering of
folks in the Bay Area, and one of the topics of discussion was Brian
Lincoln's idea for grass roots peer review.    (01)

  http://collab.blueoxen.net/forums/yak/2003-07/msg00159.html#nid06    (02)

Being a Collaboratory topic, the discussion eventually morphed into
the question: Can we set up an experiment to test the idea of
collaborative filtering using personal trust networks?    (03)

First, some observations:    (04)

1. Grass roots peer review is one answer to the question Peter Jones
   recently posed on the yak list: What good is FOAF?    (05)

     http://collab.blueoxen.net/forums/yak/2003-07/msg00149.html#nid03    (06)

   FOAF is a standard for declaring personal trust networks.  With
   such a standard, search engines like Google, Blogdex, or Technorati
   could download your FOAF file and filter content based on the
   ratings of people within your trust network.    (07)

2. People are doing this.  I think that NewsMonster allows you to do
   exactly this for blog content.  I know that Affero wants to
   eventually do this, although I'm not sure how far they've gotten.
   I'm quite certain there are others.    (08)

All this said, I think the question remains relevant.  I think that
there's a small, concrete experiment that we could do to play with the
notion of grass roots peer review, and I'd like to do it.    (09)

The hard part is defining the experiment.  There are two elements to
such an experiment: defining the trust network and building a ratings
engine.  The first part is easy: People create FOAF files for
themselves.  The second part is hard, as we discovered in the course
of our discussion.    (010)

Why is it hard?  There are three major problems:    (011)

1. What are you rating?  Interest?  Relevance?  Are you rating the
   messages or the messenger?  Are you rating based on categories?  If
   so, you need to have a shared ontology of categories.  Are you
   using a five point scale or a two point scale?    (012)

2. The UI for rating.  If you're rating e-mail messages (for example,
   the messages sent to this list), what's the right UI for rating
   them?  If you have a web-based rating system, people would have to
   go from their e-mail clients to their browser to rate.  You could
   build an e-mail gateway, but to do it right, people would have to
   format their e-mails correctly.    (013)

3. Scale.  The Blue Oxen Collaboratories aren't large enough --
   people-wise or content-wise -- for grass roots peer review to
   create meaningful results.  In a sense, our collaboratories are
   already a form of grass roots peer review.  People join it because
   they value what the other members of the community have to say.  So
   if we're not using our own content, whose content should we use?    (014)

Cool stuff, eh?  I was jazzed by the conversation tonight, and I'd
like to continue it.  I think there's an idea here somewhere, and I'd
like to explore it and test it.  I have a feeling some really cool
things could come out of it.    (015)

-Eugene    (016)

-- 
Was I helpful?  Let others know.    (017)

http://svcs.affero.net/rm.php?r=eekim&p=EnablingOnlineCommunities    (018)

-- 
This message is archived at:    (019)

http://collab.blueoxen.net/forums/cgi-bin/mesg.cgi?a=tools-yak&i=20030725061653.GC5058@douge.blueoxen.net    (020)
<Prev in Thread] Current Thread [Next in Thread>