Newsletter No. 70

2 No. 70 19th May 1995 CUHK Newsletter Comments form Senior Administration T h e I n t e r n a l R e s e a r c h A s s e s s m e n t E x e r c i s e From the Chairman of the Research Committee The University ha s always emphasize d research, together with teaching and service, in the assessment of individual members of the academi c staff . Ove r th e las t fe w months, th e Universit y fo r th e first tim e undertook a n assessment o f th e research performance of each department as a unit. The immediate impetus to the internal research assessmen t exercis e wa s th e adoption of a decentralized one-line budget for each academic department starting with the academi c yea r 1995-96 . Thi s development consist s o f tw o importan t elements. First , by devolving to faculties and department s th e responsibilit y t o manage their own affairs within an overall budget, efficiency and productivity shoul d improve. Secondly , the overall budget is to contain a portion that is performance-based, so that there will be suitable incentives as well as accountability. Having accepted this move to decentralization, the Administrative and Planning Committe e instructe d th e Research Committe e t o desig n an d implement an internal assessment exercise to assist the Resource Allocation Committee in budget allocation. Althoug h 30 per cent of the budget for academic departments is notionally tied to research, for 1995-96 only 5 per cent will be allocated on the basis of the internal research assessment exercise, with the remaining 25 per cent allocated on a pro-rata basis (weighted according to the nature of the subjects). The modalitie s o f the assessment exercise, as well as the results for each department in relation to it s ow n facult y average and the overall university average, have already been communicated to departmen t chairs, and it is expected that thes e woul d hav e been passed on t o th e academic staff . Therefore it woul d b e mor e useful to concentrate on overall impressions and deal wit h a few frequently asked questions. First an d foremost , th e assessmen t shows that across the University in virtually all departments, there is very active research; minor variation s amon g differen t unit s should not be allowed to cloud this finding. There ar e man y example s o f work s o f significant impac t i n all fields, as well as evidence that colleagues are now aiming at extremely hig h quality. I n retrospect, the quality threshold s adopte d in the curren t exercise could have been raised; they almost certainly will be in any future exercise. Secondly, it appears that this exercise has helped departments to take stock and set goals for themselves, as for instance in the qualit y o f publications . Moreover , departments hav e no w accepte d th e responsibility and the challenge of raising a proportion of their research funds externally —n o t onl y a s a means o f enhancin g resources for research, but also as a way of subjecting projects to disinterested external scrutiny. Department s were free to set their own targe t figure s fo r externa l grants , though in the majority of cases these turned out to be essentially the same as the ones suggested by the review panels. Two myths need to be debunked. In this exercise, there was no list of 'approved' journals, an d no mechanica l yardstick ' to measure wha t i s good research output — no sensibl e academi c woul d believ e tha t such a set of criteria could be constructed for eve n one major discipline , le t alon e across the entire span of subjects covered at CUHK. Th e review panels agreed only on broad guidelines, tempered with flexibility and common sense in actua l application . For example , an article i n a reputed peer reviewed journal may be assumed to be of reasonably hig h quality , especiall y i f th e reviewing process has been at arms' length. In contrast, a conference presentation might not have gone through as rigorous a review. But, th e facult y dean s hav e ha d th e opportunity t o convinc e th e panel s otherwise i n particular circumstances , fo r example, becaus e o f th e natur e o f th e subject, becaus e th e conferenc e wa s especially prestigious , o r becaus e th e presentation was a specially invited keynote speech. Th e same goes for books of high quality which may not have gone through a formal refereeing process. The second myth i s that the exercise (like the earlier UGC exercise) demanded from each academic a n output o f a t least three items over four years as a quantitative goal. This was certainly not the case; rather, the panels woul d conside r at most three items, md in many cases a single significant item woul d b e judged t o b e mor e tha n adequate. In deciding that, in effect, afourth entry woul d no t eve n b e looked at , th e exercise was designed to stress quality rather than quantity. It i s likel y tha t there will b e futur e internal research assessment exercises, and that there will be a larger linkage to funding. However, it is thought that future exercises could probabl y b e don e i n a les s cumbersome and less mechanical way, and higher quality-target s coul d be set . Th e performance indicator s coul d als o b e improved. Fo r example , i n th e curren t exercise, th e supervisio n o f researc h postgraduate student s wa s take n int o consideration, bu t onl y throug h th e percentage of PhDs. The rationale was that the total number (including MPhils) had in the past been set by a centrally determined quota, an d therefor e di d no t reflec t performance. Wit h more places becoming available an d quota no longe r stric t a s of the academi c yea r 1995-96 , i t wil l mak e sense t o includ e MPhil s i n th e future . Nevertheless, i t ha s to be recognized that no formal assessment method can be perfect, and the results must not be treated as gospel. Such assessment s shoul d i n fac t be taken wit h a large grai n o f sal t a s far as individuals are concerned, and herein lies a difference betwee n thre e level s o f assessment: th e UG C assessmen t o f institutions, the internal research assessment of departments , an d th e assessmen t o f individuals fo r purpose s suc h a s substantiation or promotion. Th e larger the aggregate uni t o f assessment , th e mor e acceptable i t i s t o us e simpl e an d gross measures, in the knowledge that errors will tend to wash out in the average; the same methodology, i f blindl y transferre d t o individuals, woul d fai l t o tak e account o f special circumstance s an d could resul t i n injustice. The review panels therefore stated clearly that the results of this assessment will not b e mad e know n i n th e cas e o f individuals, an d under n o circumstance s should they be used in staff assessment. Indeed, th e rea l issu e — i n thi s assessment and in future assessments — is not about funding or about comparison and competition between different units. Eve n without th e linkag e t o funding , th e University need s to take a serious look at itself from time to time, to understand the strengths and weaknesses of each unit, and to set goals for the future. I n the end the real question s are : Will such an exercise signal the right message about the expected role of research in the University? Will this in turn lead to improvements in the future for the University as a whole? Th e answer would b e 'yes ' t o bot h question s — provided the purpose and the nature of the assessment ar e no t misunderstood . I t i s hoped that this articl e will contribut e i n a small way towards a better understanding. Kenneth Young (MAN v s C O M P U T E R IN CHESS cont'd) who work s i n th e Prince o f Wale s Hospita l (bu t i s affiliate d wit h th e Singapor e Ches s Federation). The match will be played at a rate of 40 moves in two hours each, followed by either the remaining moves in 30 minutes each, or game adjudication by a panel including International Master Michael Valvo and International Master David Levy. Saitek will also be giving members of the general public the opportunity to win Saitek computer chess machines in competitions held during the championship. The local organizing committee of the championship consists of six members and is chaired by Dr. H.K. Tsang of the Department of Electroni c Engineering, who is the Hong Kong Open Champion in 1994. Dr. Tsang did a major part of his schooling in the UK, and was introduced to chess at the comparatively lat e age of 14 by a teacher who himself is an avid and eminent chess player. He feels7—8 years to be the right age that a child can be initiated into chess, though children as young as 4-5 years are also known to comprehend the game an d mak e goo d moves . Dr . Tsan g quotes th e exampl e o f th e famou s Hungaria n educationist, Polgar, who introduced his three daughters to chess from infancy wit h very successful and encouraging results. His youngest daughter, Judit Polgar, is today one of the strongest players in the world. In Dr. Tsang's opinion, she is an exceptional and gifted player in a game which somehow does not seem to hold popular appeal for many women. This is surprising, given that chess is a game primarily of the mind and mental concentration, while requiring its share of physical stamina with a single game often requiring a player to sit for six to eight hours at a stretch. According to Dr. Tsang, though chess is a popular game in Hong Kong in high schools, it does not receive the emphasis it does in the US where studies have shown a high correlation between chess playing abilities and academic abilities. When asked about general attitudes towards, and acceptance of, computer chess v i s - à - vis chess between humans, Dr. Tsang felt that computer chess has been gaining strengt h rapidly in the last few years, and is challenging the strongest human players. IBM's Deep Blue Programme has already reached the standard of perhaps the top 200 human players in the world and could, in the next five to 1 0 years, equal that of the very best. But given a choice, Dr. Tsang himself would much rather play wit h another human, or at least, while playing with the computer, have the familiarity o f a real chess board beside him to replay each move on. In the imminent championshi p o n campus too , the chess games with the computers will be accompanied by real chess boards and clocks, with a computer operator replicating the moves made by the computer on the chess board, for the benefit of the human players. O f th e 2 4 compute r entries , som e wil l b e Pentiums an d the res t Engineerin g Workstations, with several of them using very strong multiprocessor computers. Five of the computers will not be physically present, but will instead be connecting by internet or telephone lines from their home locations in Europe or North America. Dr. Tsang has an FIDE (Federation Internationale Des Echecs) rating of 2,200. These ratings have points added to them with every win, and points deducted with every loss. It is based on the normal distribution curve, and the comparative ratings of two players can indicate a probability of the results, should they play against each other. Hypothetically, should Dr . Tsang and Kasparov (FIDE rating of 2,800) play a match of 100 games, the normal distribution curve shows a differenc e of 600 points to mean that Dr. Tsang would probably wi n two games. A difference of 300 points (eg. between IBM's Deep Blue with an FIDE rating of about 2,500, and Dr. Tsang) predicts a result of 15 wins out of 100 games for the lower rated player. The forthcoming championshi p promise s t o be an interesting one , with progress i n computers and computer chess programmes going from strength to strength. While Kasparov would perhaps be looking at , at most, 20 to 30 positions ahea d of the immediate one , a computer today would be looking at millions ！ Computer chess is also important for research, in helping find techniques to improve the speed of searching algorithms — which form the basis of computer applications 一 and hence facilitate applications in real world industries, like in integrated circuit layouts. For all those interested in chess, the opportunity to see some of the best computers and human players in action will be an enriching and memorable experience, and the organizing committee is looking forward to an enthusiastic response from the press and public alike. O Shalini Bahadur