[Textop] New essay: The New Politics of Knowledge

[Philippe MARTIN]

> I've posted a new essay on my blog, called "The New Politics of Knowledge":
> http://www.dufoundation.org/blog/?p=96

Nice article!
However, it is based on the implicit assumption that there is a need
for each user of an information repository to let other persons than
herself "decide who determines what should pass for knowledge
or reliable information" and, more generally, arbitrate conflicts 
that she has with other users (even the "anarchy" system of governance:
"People manage themselves and resolve disputes on a piecemeal basis, 
or not, perhaps with the help of people both parties agree on").
The (sub-)assumption that I question is that a collaboratively-built
information repository necessarily implies that conflicts have to occur 
or have be solved in order to help each user increase her trust in
the stored information.
The conflicts that occur in current collaboratively-built information
repositories have three related causes:
- there is no efficient way for each user to filter out information
  according to her own combination of criteria about (i) the authors
  of the information (e.g., one user may want to discard information
  authored by people who do not have a PhD and/or who have authored
  something that she thinks particularly stupid) and (ii) the 
  information itself (e.g., one user may want to only see a piece of
  information if it has been argumented with arguments that have not
  been counter-argumented and/or if it has been voted as "relevant" 
  and "very original" by people who, according to the user, have not
  so far demonstrated bad taste);
- information is very poorly structured (e.g., ideas/facts/sentences 
  are not related by argumentation/specialisation relations) and has 
  very poor metadata (e.g., no associated source and/or source 
  interpreter, no vote on the originality, veracity or relevance of 
  the pieces of information, no ontology for the concepts in each
  piece of information) and hence adding information is "too cheap"
  (precision, consistency and deep thinking is neither required nor
  encouraged) and each addition lead to more redundancies in the
  repository and hence makes information more difficult to retrieve,
  compare, organise or filter out;
- the only efficient way used for dealing with possibly problematic
  information (e.g., considered as false, redundant, irrelevant, 
  too technical, offensive, ... by some persons) is either to let a 
  group of proxies decide to delete it or not and thus impose their
  own criteria to all users, or let each user the possibility to 
  delete anything and thus impose her criteria to all users (which
  leads to "edit wars").
A collaboratively-built information repository requiring
information to be very structured (e.g., by asking each new 
sentence to be related to an existing one by a semantic relation),
permitting votes on various criteria, and permiting personalised
knowledge filtering, would permit each user to specify and hopefully
see what she wants to see given her current mood, time constraints
and applications, and it would encourage the information providers
to be more cautious about what they write since their reputations 
are at stake (the provided "default" procedure for evaluating each
bit of information and each author may be used as a guide; in
http://www.webkb.org/doc/papers/iccs05/iccs05.html#cooperation
I give a template for such a procedure; the only real problem I 
currently see for this information filtering based approach is to
efficiently distinguishing genuine spam from things malevolently 
voted as spam by "new" users without making the filter too agressive 
and hence miss out on possibly interesting information). 
With such an approach, the use of proxies is undesirable for any user
and hence the use of a "system of governance" (be it a "constitutional
representative democracy", a "pure democracy" or an "anarchy") seems
useless and undesirable.
Thanks in advance for anyone's rebuttals, they'll help me refine my
views or my arguments.

Philippe