Posted by ""Conrad, Stephen H"" <
shconra@sandia.gov>
Kim,
Let me try and take a stab at answering your questions:
- does this wisdom still work where there are strong dynamic effects,
e.g. heavy worse-before-better outcomes .. I can see how it might do,
with individuals spotting these effects 'infecting' others with their
own understanding
I don't recall there being any specific examples considered where there
were strong dynamic effects per se. As Surowiecki states in a recent
interview,
""The idea really came out of my writing on how markets work.
Markets are made up of diverse people with different levels of
information and intelligence, and yet when you put all those people
together and they start buying and selling, they come up with generally
intelligent decisions. Sometimes, though, they come up with remarkably
stupid decisions -- as they did during the stock-market bubble in the
late 1990s. I was interested in what explained the successes and the
failures of markets, and as I got further into it I realized that it
wasn't just markets that were smart. In fact, crowds of all sorts were
often remarkably wise.""
He addresses three types of group interaction problems: cognition,
coordination, and cooperation. Cognition problems are those where
there's (supposedly) a right, or best, answer. Examples:
1. what are the odds of U Man beating Arsenel?
2. which horse is most likely to win the Grand National?
3. how much is a share of BP worth?
The bets of people betting on a football match or a horse race determine
what the odds will be, and the choices of investors determine stock
prices. Really, any group of people who can act collectively and
independently to make decisions and solve problems -- like a company or
a government agency --can be considered to be a crowd. And according to
Surowiecki, the more diverse and independent each of the voices, the
better the decisions tend to be. So, let's consider all those
independent voices in a bit more detail. Even when considering the
solution to an inherently dynamical problem:
-- some may tend not to give much consideration to the dynamical aspects
at all, instead looking to historical analogues
-- some may use choose to ignore the dynamics because they favor a more
statistical approach such as using correlation
-- some because of their experience and expertise, may be more
sophisticated in their consideration of the dynamics, but perhaps only
intuitively using systems thinking (without explicitly using either
qualitative or quantitative systems analysis tools to help them)
-- and some, perhaps those more formally trained in SD, may be inclined
to explicitly consider system structure and favor a more mechanistic
interpretation in forming their opinions
Surowiecki would claim that experts, no matter how smart or
sophisticated their analysis tools, only have limited amounts of
information. They also, like all of us, have biases. Add together enough
independent opinions and the biases tend to cancel out. The lesson
Surowiecki would have us take away is not to rely on one or two experts
when making difficult decisions. That doesn't mean that expertise is
irrelevant. It just means that together all of us know more than any one
of us does. (I guess I believe this -- mostly. The contrarian in me
still has some trouble acquiescing all the way. As Mark Twain said,
""whenever you find you are on the side of the majority, it is time to
pause and reflect."")
- is this phenomenon itself amenable to being modeled [and is there any
value in doing so?]
Well, there seem to be lots of agent based models that attempt to
explain the behavior of markets. In SD, we tend to attempt modeling
broader problems such as the interaction of commodity markets with the
behavior of suppliers and consumers (such as in the Meadows' hogs
model).
Maybe it would be interesting to model how an organization currently
solicits opinions when making important strategic decisions. Then, look
for policies that have the effect of encouraging broader participation.
I know in my company, some decisions tend to be top-down, while others
tend to be made bottom-up. Occasionally, we bring in outside experts to
provide an independent viewpoint. I think it'd sometimes be worthwhile
to have some combination of all applied to addressing the same problem.
Also, I would think that the coordination and cooperation problems that
Surowiecki considers would be particularly amenable to agent based
modeling. An example coordination problem is negotiating traffic. An
example of good traffic coordinating behavior: I avoid the rush hour
when commuting to work (because nobody really cares exactly when I show
up). I avoid the worst of the traffic (good for me) and the traffic
during rush hour is a very little bit smaller because I'm not driving
then (marginally good for everyone else). An example of bad traffic
coordinating behavior: New Mexicans seem to insist on driving slowly in
the left lane. (Italians driving on the Autostrade seem to coordinate
much better because the collective agreement seems to be that faster
cars have the right of way in the passing lane.) As for cooperation
problems, people cooperate to uphold and enforce the law even when it's
more rational (at least in the short term) for them to turn a blind eye
or let others do the work. It seems that eliciting and programming the
decision rules of various individual agents -- and then allowing those
agents to interact and learn -- can provide insight about emergent
behaviors found when studying coordination and cooperation problems.
(Emergent behavior is defined as the collective, aggregate, sometimes
surprising, behavioral manifestation of all those little individual
decision rules.) And I would say that it is the explanation and
definition of these emergent behaviors that we operationalize as table
functions to summarize aggregate behavior in an SD model.
Interesting questions you ask. Hope this is not overkill.
Steve
Posted by ""Conrad, Stephen H"" <
shconra@sandia.gov>
posting date Fri, 17 Jun 2005 20:05:46 -0600