[PREV - BEAT_BREAK] [TOP]
REALITY_RANK
September 21, 2010
August 07, 2013
August 17, 2013
Somone writing a comment in Paul
Krugman's blog suggested it would
be nice if he could block comments
from particular people.
In other words, he reinvented the
idea for "kill files", available
for years in usenet clients.
This would work to screen out the crazies, but
not the professional, hired jammers who have
been a problem for some time: they already TEN_YEARS_LATER
create new accounts with bogus identities at
will (a site that doesn't verify identities is
a site that's essentially a toy, which is to THE_TOY_WEB
say they're all toys).
And running a kill file on trolls doesn't help
improve the general level of discourse: the
trolls work by acting on people who haven't
learned not to take the bait. If you've decided
to place a troll in your personal a kill-file,
you're already well on your way to being immune
to the troll.
Myself, I find it's just as easy to
use a "mental kill file", and just
not look at the ones from people
I don't care about.
Krugman often laments that the mainstream
medium-- with professional editors and such--
don't take the trouble to screen out
commentators based on elementary criteria
like contact with reality.
Is it possible to find a way to do this in
a quasi-automated, crowd-sourced sort of way?
Remember: it wouldn't have to be very good to
do a better job than the editors at the New
York Times and the Washington Post.
Take it as a given: you would like to
see a score ranking commentators based
not on how polite they are, not how
"reasonable" they sound, not on whether
they parrot conventional wisdom; but
based on things like how often they get
things right.
What would that mean?
How could you get something
to work like that?
Or at least to approximate it?
Some ideas:
(1)
a forum where you must place three futures bets
a year in order to play.
once again: must be tied to meatspace,
or it's too easy to hedge.
(2)
The masters of reality tribunal:
they stamp people with an R
score. High R people become
lieutenants, capable of handing
out R scores to others.
(3)
Don't ignore the obvious:
"like", "recommended" buttons,
so the churning crowd can
gradually contribute to your
score. Note: this is less
easy to game with real ids. REAL_ID
Still has the problem of using
conventionality as proxy for
the reasonable. (the family
feud epistemology).
--------
[NEXT - REAL_ID]