On Thu, Dec 27, 2001 at 03:05:30PM -0500, Mike Lorrey wrote:
> Anders Sandberg wrote:
> > As for po2, I think I can at least throw out a first iteration: trust is
> > about long-term relations. As Axelrod suggested, you can build trust in
> > IPD-like situations since participants know they will be around each
> > other for a long time, and hence it is in their interests to cooperate.
> > I think it was Novak who later on did some game theory on how reputation
> > also helps: if your actions become known to others, then it pays to show
> > yourself trustworthy since other members in your community will take
> > your past actions into account even when they were aimed at somebody
> > else. So trust can be promoted by making past interactions clearer and
> > well known, and by the awareness of that there is a long future ahead.
> Also consider that trust between strangers evolves from each individuals
> characteristics, which is one reason why 'discrimination' or 'prejudice'
> is a natural phenomenon of human relations. If your experience tells you
> that men or women are more trustworthy than the other, you will trust
> individuals in the future based on that experience. If your experience
> is that you can trust blacks more than whites, you will do so with
> blacks you don't know as well (and vice versa if your experience is the
The problem is when prejudices do not become postjudices.
In order to function in the world we need to have prior expectations about
nearly everything, and we acquire them in all sorts of ways even if we have
never directly encountered the object of expecation - through culture,
through stories about the experiences of others, deductions and sometimes
random preferences. These expectations then influence how we act, producing
experiences we can learn from. A rational being adjusts their expectations
to fit experience, ideally taking metacognitive factors into account such as
the risk of confirmation bias etc. The prejudice (prior probability
estimate) has become a postjudice (a posterior probability estimate), and
the process can continue.
When we normally talk about prejudices, we do not refer to prior estimates
but rather to the tendency of not updating estimates - especially about
other people. This is an irrational strategy, and in itself a reason to
trust prejudiced people less (since their actions have a persistent
irrational component and are not based on correct estimates of the state of
> This is why bin Laden's terrorism handbooks advise infiltrators of the
> US to groom and dress themselves as westerners, obtain western
> identities and documentation, etc: not only does this tactic take
> advantage of this implicit trust, but it also amplifies the effect that
> subsequent terrorist attacks have by causing Americans to question this
> implicit trust of people appearing in western guise.
Which may actually be a good thing, since it is not a reliable predictor of
trust anyway (ask any conman). In fact, it might be worth examining what
trust predictors people commonly use that are *not* valid and see if they
could be made valid or be recognized as invalid. Better trust signals can
make our society more trusting.
> > The problem we face here is that we now have a far larger and diverse
> > society with larger tensions - suddenly everything is global, your
> > neighbour is a gay zoroastrian sociologist and the smallpox genome can
> > be found on the Internet. Solutions trying to fix things by making the
> > world a smaller place (isolationism) are practiclly unworkable (the
> > benefits are enormous), as are attempts to put genies back in bottles
> > (relinquishment). Attempts to make people less diverse are not just
> > ethically questionable and impractical, they also tend to breed
> > resistance that produces just the kind of nasty backlashes we have come
> > to fear. Hence any solution has to deal with strengthening those webs of
> > trust.
> Yes. Part of this is mechanisms to better filter out those
> untrustworthies who have infiltrated the webs of trust. As a chain is as
> strong as its weakest link, the solution is to check each and every link
> for integrity.
A better solution is to have a web rather than a chain. A web is stronger
than its weakest link.
A problem of current attempts to find infiltrators is that they will attack
non-infiltrators too from time to time. While some such errors are
unavoidable and acceptable, there is a risk that pervasive infiltrator-hunts
undermine trust more than they protect them. Suspicion does not breed
-- ----------------------------------------------------------------------- Anders Sandberg Towards Ascension! email@example.com http://www.nada.kth.se/~asa/ GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
This archive was generated by hypermail 2b30 : Sat May 11 2002 - 17:44:32 MDT