There are four main speakers; two 'for' the motion (known as the Proposer and Proposing Second) and two 'against' (the Opposer and the Opposing Second). The debate commences with brief statements from (in order) the Proposer, the Opposer, the Proposing Second and the Opposing Second. The debate is then opened to 'the floor'. No questions are allowed from the floor, only statements can be made, which will be of limited duration and which should normally, as suggested above, support either the Proposer's or the Opposer's case.
At the last stage before the vote is taken, the Opposer and the Proposer (in that order) are allowed to summarise their case, and to include any rebuttal they wish to make of points that have been raised either by their opponents or from the floor. There is a neutral Chair who keeps the speakers to time and ensures as far as possible that the debate remains relevant to the motion. In the event of a tied vote, the Chair has the dubious honour of casting the deciding vote!
The best results (in terms of the standard of debate, and making the event one in which it is fun to participate) are achieved by choosing a motion which is clear and contentious, and by choosing forceful advocates as the four main speakers. It is helpful if the Chair is generally fair and has experience of managing such a debate.
For the ICS'97 event, we have chosen to look at the topic of 'data-sharing in a single-address-space' versus 'message-passing between multiple-address-spaces' as competing means of developing programs for scalable parallel computers. This is an issue on which opinions seem to have swung backwards and forwards, from one extreme to the other, for almost a decade now. Neither scheme seems to have wholly 'won' the practical argument (by being universally adopted by the end-user communities); indeed, there have been continuous changes of stance by key players (e.g., the manufacturers) during this period (and some manufacturers have apparently supported both sides of the argument at the same time!). Yet the criteria by which the argument should be 'won' seem quite clear to those concerned. So, it certainly seems that we have a topic that fits the needs of the medium of formal debate. The formal motion that we have chosen to embody this debate is the following:
"This House believes that the development of effective programs for massively parallel computers is better achieved by means of message-passing between multiple-address-spaces than by means of data-sharing within a single-address-space."
For our advocates, we have invited the following eminent scientists:
For the motion: Proposers
Dennis Gannon, Professor of Computer Science, Indiana University,
Bloomington, Indiana, USA.
Piyush Mehrotra, Research Fellow, Institute for Computer Applications in Science and Engineering, NASA Langley Research Centre, Hampton, Virginia, USA.
Against the motion: Opposers
William Jalby, Professeur d'Informatique, Universite de Versailles,
Saint Quentin, and Chief Executive, International Supercomputer Technology
Institute, Mulhouse, France.
David Snelling, Research Manager, Fujitsu European Centre for Information Technology, Uxbridge, UK.
The debate itself will occupy a total of 2 hours. The Proposer and Opposer will be allowed 8 minutes each to present their initial case, plus 5 minutes summary at the end. The Seconds will each be allowed 7 minutes to make their initial case. Contributions from the floor will be limited to 3 minutes maximum, leaving time for at least 25 such contributions. We strongly recommend you bring paper and a pen, with which to prepare your constribution so as to make maximum use of the limited available time.
The Chair will be John Gurd, Professor of Computer Science, University of Manchester, UK.