Firstly, the following link has a good, basic discussion of the Prisoner's
Dilemma:
<A HREF="http://plato.stanford.edu/entries/prisoner-
dilemma/">http://plato.stanford.edu/entries/prisoner-dilemma/</A>
I think that while it might be interesting to analyze the model of the google
answers in terms of game theory, I think the prisoner's dilemma is a poor model
to use, for the following reasons:
(i) The prisoner's dilemma features two players. Google groups features
numerous players, with three broad categories of users.
("Askers", "Researchers", and "Google")
(ii) In the prisoner's dilemma, the crux of the dilemma resides in the fact
that there seems to be a fundamental symmetry between you and your opponent.
This symmetry is lacking in the case of Google Answers.
Fundamentally, Google's system reminds me more of the setting of Taxi drivers
in an urban environment. In the typical situation, the average cab driver is
either an independent entity, or a subcontractor who works for a taxi company.
The company is really a glorified dispatching agency, helping cab drivers to
meet up with potential fares. In less stable markets, cab drivers compete with
each other more fiercely (imagine a less developed country, where taxi cab
drivers offer all manner of different car sizes, speeds, safety records, or
degrees of cleanliness in the cabs they offer their customers!).
In more established markets, the cab rate has been set by some sort of
overseeing body -- the government, the dispatch agency, or a semi-professional
organization. Competition is still present, however, particular in places like
airports, where many cab drivers compete for fares.
The analogy is not perfect, of course, as the customers set the fare rather
than the cab driver, but many of the other elements are the same. Most
importantly, there is seldom any reason for the cab drivers to 'compete' with
the dispatching agency on a regular basis. Over time, the dispatching agency
and the taxi drivers would probably negotiate a specific amount of revenue to
go to the agency, but I'm not sure if it would be appropriate to speak of the
drivers 'competing' with the company at any time other than when the rates are
set.
Likewise, in the day to day operation of answers.google, the only real
competition possible is the percentage of the question revenue which Google
gets. The lion's share of the competition is naturally between the researchers,
who try to produce accurate and rapid answers as quickly as possible. Along
with these more "positive" strategies, researchers can also undercut one
another, by providing more accurate responses via commentary. They can cast the
original answer in a bad light, causing the customer to ask for clarification,
or they can simply be gaining a sort of reputation. ("Person Y gave me an
incomplete answer, but Person X gave me a complete one. Don't let Y answer my
next question!")
The utility of the researchers remains to be seen. A number of relevant factors
are:
(i) The value of personal time to the researcher. This also includes jobs where
researchers must be "on-call" to do other things -- for example, a fireman
might be able to work on his Google job while awaiting calls, and still end up
with a standard work week.
(ii) The value of the service to the askers, particularly in regards to the
amount they're willing to pay.
(iii) The amount of intra-researcher competition. (This is presumably a
function of the number of researchers.)
(iv) The fraction of the pot Google takes.
In regards to the specific fate of the researchers, consider that the internet
has seen a fair-sized minority of its population able and willing to spread
information to other users. We see this in the creation of the open directory
project, the open source movement, or even file sharing networks. In all these
cases, the only really pay-off is the vague intangible that is fame. It seems
more than this, though, given how many of these people labor without
recognition. Regardless of the psychological profiles of these 'contributors',
however, is a simple fact: Google Answers allows them to continue doing what
they were doing for free -- but now they can receive renumeration from it. The
public nature of the answers and the commentaries also seems to play into what
these sorts of people are really into in the first place -- the public
dissemination of information writ large.
Finally, on a personal note, I think that the mathematics of the Google Answers
model are very interesting, and might prove a case study in game theory. The
previous comparison to taxi cab drivers, dispatchers, and fares is a strong one
I believe, but more remains to be said given a detailed analysis.
Some relevant reading:
1. <A HREF="http://www-irps.ucsd.edu/irps/faculty/auctions.html">http://www-
irps.ucsd.edu/irps/faculty/auctions.html</A>
This link discusses auctioning off public resources, and has a bit about how an
example of this is the selling of cab medallions in places like New York City.
It suggests that Google might one day have to limit the number of researchers,
in order to make the job lucrative enough to support them.
2.<A HREF="http://www.constitution.org/prisdilm.htm">
http://www.constitution.org/prisdilm.htm</A>
This link is a hub which links to a number of Prisoner's Dilemma web pages,
including numerous connections to sites that simulate individual runs of the
PD, or large simulations of various strategies. |
