Re: [Ag Meinungsfindungstool] Tailored Intervention in case of Violation of Human Dignity and Human Rights

[Scott Raney <scott AT metacard.com>]

On Sat, Aug 23, 2014 at 3:03 PM, Thomas <entropy AT heterarchy.net> wrote:

> I think it would be inappropriate to assume an "objective true score"
> for certain behavior or statements in such highly subjective contexts as
> politics. The ratings depend too much on individual values, priorities,
> knowledge and sympathies.

Exactly!

> Furthermore there would be no need have a collective decision on
> disruptive behavior if every participant can fine tune her filters
> individually (online).

Right: In a sense what this does is provide a collective "vote" on the
quality of a submission.  An open question, however, is whether this
filtering should also work on individuals rather than just their
posts.  My experience with the comments sections on news sites tells
me that it would be very useful, since it's the same individuals who
reliably exhibit troll behavior.  But it does start to blur into
censorship at that point.  Even so, far better *I* be able to make
this decision than some moderator or some "jury".

> I recommend to consider modern collaborative filtering approaches
> http://en.wikipedia.org/wiki/Collaborative_filtering
> for such ratings.

This technology is really in its infancy and what little we do know
about it probably won't transfer very well to the types of government
decision-making systems we're proposing.  It's going to require a lot
of experimentation.  Fortunately we've got lots of time and lots of
guinea pigs ;-)

> Every participant would rate sets of both self-selected and
> randomly-assigned items (statements etc) based on which the system could
> * predict how other items would be rated by the individual based on
> similar (in terms of rating behavior) participants
> * find more "objectively" good or bad items which are rated similarly
> even by very dissimilar participants ...

My inclination is a strong adherence to the KISS rule (Keep It Simple,
Stupid!)  Last thing we want is for people to start distrusting the
system because they can't predict how the algorithm is working.
Facebook already suffers from this, and it's getting worse all the
time as their algorithms get more sophisticated.  In the case of
public-policy decisionmaking, it'll seriously compromise the project
if people start believing that the system itself is working to promote
particular viewpoints (people kind of accept being manipulated for
commercial purposes ;-)

> There would be no binary global sanction mechanism but every participant
> could set a threshold for item display based on his predicted ratings.
> No jury would need to be explicitly elected and trusted.
> This way trolls would quickly be ignored (i.e. have nearly no influence)
> by other participants but would still have a chance to
> improve their reputation by changing their behavior.

Exactly.  And even if the system has flaws at least the errors be
somewhat random, as opposed to a "jury" or "moderator" system where
that special group not only gets to define the rules, but decide how
to enforce them.  That would lead to systematic bias, and most likely
toward authoritarian behavior (because, again, those are the people
that will want to be involved in this part of the system).

> IMHO it doesn't make sense to use the term "democratic" in this context.
> In democracy every voter has (ideally) the same influence on the decision.
> But in individual decision making the influence of other people on you
> varies considerably.

But it sort of does: In this case the "decision" is whether a post (or
a person) is a valuable contribution to the process of making a
decision on a larger proposal.  Making even those small decisions
collectively is the essence of democracy.
  Regards,
    Scott

>> authoritarianism, I highly recommend Bob Altemeyer's book on the
>> subject:
>> http://home.cc.umanitoba.ca/~altemey/
>
> +1
>
> best regards,
>
>
> --
> Ag-meinungsfindungstool mailing list
> Ag-meinungsfindungstool AT lists.piratenpartei.de
> https://service.piratenpartei.de/listinfo/ag-meinungsfindungstool