Game theory: Army of agents to tackle corrupt
officials, tax evaders, terrorists
October
20, 2016
Credit: University of Warwick
Game theory has long been used to
apply mathematical models of conflict and cooperation between intelligent
rational decision-makers.
However, our world has evolved from
great power conflicts to one where many of our major problems are spawned not
from monolithic blocks of self-interest, but from a vast array of single
entities making highly individual choices: from lone wolf terrorists to corrupt
officials, tax evaders, isolated hackers or even armies of botnets and packages
of malware.
Game theory needs to catch up and
new research by mathematicians, led by Professor Vassili Kolokoltsov at the
University of Warwick, has just found the way to do that by giving game theory
calculations an enormous army of "agents".
In a paper, entitled 'The
evolutionary games of pressure (or interference), resistance and
collaboration', Professor Kolokoltsov, from Warwick's Department of Statistics,
has been able to take Game Theory far beyond some of its early applications of two
opposing sides in zero sum games, and equipped it with the ability to model the
impact of a vast array of individual actors - an "infinite state-space of
small players".
The paper says the new tool can be
"applied to the analysis of the processes of inspection, corruption,
cyber-security, counter-terrorism, banks and firms merging [...] and many
others"
To take just one application: tax
fraud costs the UK government £16bn a year, according to the National Audit. HM
Revenue & Customs (HMRC) has faced questions about both how it decides to
deal with individual large companies and how it balances its efforts between
pursuing large corporations and individual tax payers.
This evolution of Game theory could
greatly assist it in simultaneously model the best approach to manage the great
number of participants in the process and create efficient disincentives for
both individual and corporate tax evasion.
The modelling tools this evolution
of Game Theory will provide can also deal with a tax system's budget inputs and
the potential for corruption within any tax system.
Professor Vassili Kolokoltsov
comments:
"Our method has a potential to
be used in a variety of situations where one big player, referred to as the
principal agent, confronts the behaviour of a large pool of individuals with
different agendas."
"Of course, as usual for the
applications of mathematical tools to socio-economic systems, any concrete
applications of the method would require a serious additional input of concrete
experimental data to feed various key parameters the model relies upon,"
he continues.
Professor Kolokoltsov is now working
with colleagues to apply the new Game Theory technique to specific types of
problem such as internet Denial of Service attacks by botnets.
No comments:
Post a Comment