| “A core task of an economy is the reallocation of
goods, services and resources. This plays an
important role not only between customers and
manufacturers within a national or international
economy, but also within a company resources
need to be used efficiently in order to be
competitive. Even in a traditionally less commercial
environment, such as health care, the efficient
allocation of goods and services is becoming ever
more important. In principle, modern ICT
environments, such as Enterprise Resource
Planning (ERP) and calendar systems, offer good
support in solving allocation problems. Accordingly,
many of us now more or less automatically plan a
large number of our appointments using the
electronic calendar.
My academic background is in mathematical
optimization, a field of applied mathematics. In this
field, with a central goal in mind, we examine
planning and attempt to use mathematical models
to calculate a good, if not the best, solution.
However, the structure of modern ICT
environments, especially the internet, does not
always allow for central planning. There are too
many parties involved, each with their own
preferences. Calendar systems such as Outlook do
not, therefore, plan on the basis of optimization
models, but instead on the actions and preferences
of individuals without there being one central goal.
Even ERP systems, despite their name, are often
unequipped with a planning module to take
account of a central goal, instead only collecting
decisions about where and when particular goods
and services can be used, and they look out for
consistency. Consider a hospital, for example: a
patient telephones to make an appointment. The
assistant checks the calendar system to find a
possible date, taking into account the doctor’s
wishes (e.g. ‘preferably not after 4 p.m.’) and other
logistical data (e.g., whether the examination room
is free). The patient in turn tries to plan the
appointment for the most suitable time for him. In
short, it is a combination of computers, people and
individual wishes that eventually leads to the
appointment date. The question then remains of
how good the outcome of such a decentralised
approach is. Every organisation wants the time of
all its employees to be used in the most efficient
way. And, expensive resources as, for example, the
operating theatre of a hospital must be in as much
use as possible. My research concentrates on the
design of ICT systems that are able to maintain a
balance between individual preferences and
overarching goals. At the start of these kinds of
systems are mathematical models that combine
game theory and operations research. It also
involves a significant amount of computer science,
in order to take account of computing times in the
system. This field, known as Algorithmic
Mechanism Design, is still quite new, although it
can hark back to results from microeconomics and
computer sciences. In this field we also learn a lot
about systems in their ‘natural state’, such as the
community of users of eBay or other internet
markets. In developing new systems, we attempt
to consider various issues simultaneously. Firstly,
users have differing opinions on the outcome of
decisions: you have to deal with very many possibly
contradictory preferences, and thereby search for
compromises. However, a compromise is not the
best option for everyone. People then demonstrate
strategic behaviour and attempt to influence that
decision. Reverting to the example of the hospital,
you could imagine that the patient may lie and say:
“I can’t make that appointment”. Using strategic
behaviour, people conceal facts, lie and manipulate.
One way of avoiding this is to allocate a proportion
of profits to the players involved. So you could offer
extra incentives to a patient to attend a midday
appointment. In a commercial application this
would be done by making a payment. This is what
happens in auctions. To use eBay as an example: it
is clear that the winner of an auction always pays
only a little bit more than the price his closest rival
bidder was prepared to pay. The difference
between the maximum he was prepared to pay and
the price actually paid is, in fact, a payment from
the seller to the winner. This difference is a profit
for the bidder and eliminates his incentives for
strategic behaviour. At least, as far as theory is
concerned, since there is always an element of
irrationality here, which needs to be investigated
through Experimental Economics.
To summarise, my research covers the following
themes:
- How do I optimise a central goal without centralised
decisions?
- How do I apportion the central value between the
players to encourage them to cooperate properly?
- How can I construct mechanisms so that
communications and the processing of information
sources is not too broad, i.e. limit the quantity of
information and computer capacity?”
|
