Return to Colloquia & Seminar listing
A User's Guide for Fixing Elections with Discrete Mathematics
Student-Run Research| Speaker: | Isaiah Lankam, UC Davis |
| Location: | 693 Kerr |
| Start time: | Wed, Oct 15 2003, 12:10PM |
Description
Abstract: Economist Kenneth Arrow proved in the 1951 the much celebrated
"Arrow's Impossibility Theory". This theorem essentially says that there can
never be a "fair" election with three or more candidated when one makes
certain mathematically necessary assumptions. Given the predominately
two-party political system in the United States, it doesn't seem overly
surprising that this result is not better known to Americans. However, the
content of the theorem has perhaps never before been so important to
Californians as during the recent Gubernatorial Recall Election.
In this intentionally light-hearted talk, we will first examine what
reasonably constitutes a fair election in so-called Arrovian Social Choice
Theory. This will natually include a brief discussion of various voting
systems and in particular an examination of just how poor the Plurality
system currently employed in political election is in this respect. Then we
will closely examine Arrow's assumptions and consider several ways of both
trying to weaken these conditions and how one can try to prove the theorem
in such contexts. Finally, we will illustrate this somewhat surprising
result by showing how certain example elections can be modified via a
careful examination of a graphical representation of voter preference so
that any one of the three or more candidates becomes the winner. These
examples will, of course, include the 2000 US Presidental Election (in which
we will show, e.g., how Ralph Nader could have won) and the 2003 California
Gubinatorial Recall Election (in which we will show, e.g., how any of the
leading candidates could have won).
