Introduction: my e-mail to Harrie Verbon
BEE, Game Theory, Gaming, Simulation: A personal view
199?: BEE course (by Werner Guth?) My conclusion: BEE is related to ‘management gaming’ (business & military games or ‘Kriegsspiele’ ) 1974: My postdoc visit to IBM Research, San Jose (Cal.): Q.: ’What is the benefit of computer info?’ A.: IBM game (origin: Martin Shubik, Yale) Problem: No players Solution: ‘Robot’ players References: MAB 1976; Chapter 9 in Kleijnen (1980), Computers & profits, Addison-Wesley
Jack Kleijnen Department of Information Systems & Management CentER, 20 February 2003
2/21/2003
2. Themes at CentER
Kleijnen: BEE; CentER 20 Feb. 2003
3
2/21/2003
Kleijnen: BEE; CentER 20 Feb. 2003
4
(Blackboard/Simulation/ Course documents/chapter 12)
Kleijnen (1974): Robot players in IBM game Kleijnen, Sanchez, Lucas, Cioppa (2002): Computerized agents with (AI) ‘rules’ Example rules: Propensity to move toward a location, a friend, an enemy Applications: Counter-terrorism, e-market Demo: NPS (urban warfare) Platoon: Which type of men & weapons? Kleijnen: BEE; CentER 20 Feb. 2003
1. Game theory: Rational agents in simplistic ‘world’ 2. BEE: Real human players in simplistic ‘world’ Goal of players: Money, fun, etc.? Goal of researcher: Discover behavioral rules 3. Gaming: Human players in simplified ‘world’ Goal of players: Training, fun Goal of researcher: Insight into measurable economic response (profit, market share) But: Minnesota games: ‘confidence’ in decision
3. Experiments: My view
2. Agent-based simulation
2/21/2003
2
2. BEE & management games
Theme 1: BEE; chaired by Jan Potters Theme 2: Supply Chains (SC); Jack Kleijnen Theme 3: Design science; Sjoerd Romme Sub 2 & 3: Tools A. Simulation: Dynamic non-linear models a: Discrete-event (example: Ericsson SC by Fredrik Persson, Linköping) b. System dynamics (example: Wolstenholme) c. Econometric (CPB) & micro simulations (EIM) B: Management games: Simulation model & players Examples: Info game (Casimir), VIP game (Gremmen), beer game (Romme?), VU game (Frans van Schaik) 2/21/2003
Kleijnen: BEE; CentER 20 Feb. 2003
5
Factor 1: f1 Factor 2: f2 Factor k: fk
Simulation model (computer program)
Response R1 Response R2
PRN seed(s): S Input 2/21/2003
Black box Kleijnen: BEE; CentER 20 Feb. 2003
Output 6
1
Black box: My consulting cases Ericsson supply chain (Linköping & KLICT) Nuclear waste disposal (Sandia Labs, ABQ, NM) Naval mine search (TNO, The Hague) Metal tube manufacturing (VBF, Oosterhout) Phone-line capacities & queuing (PTT, The Hague) Container harbor capacity (ECT, Rotterdam) Global warming (RIVM, Bilthoven) Conclusion: Simulation, not gaming (also see next slide) 2/21/2003
Kleijnen: BEE; CentER 20 Feb. 2003
7
Black box: My limited gaming experience Jack Kleijnen, Lin Barbosa, Jean-Paul Jacob: IBM (Shubik) game Rommert Casimir: Info game (Ph.D. Tilburg 1995) Van Schaik: VUMAS game (Ph.D. Delft 1988) ???: ‘Junior Chamber Tilburg ‘ game (term project); BEE partner? 2/21/2003
Black box: BEE???
Kleijnen: BEE; CentER 20 Feb. 2003
8
References
Problems in BEE: 1. Human players: Weak experimental control 2. Rules: hard to compare quantitatively? 3. ??? Benefits of DOE: Scientific design & analysis of experiment: Results: General (not ad hoc, casuistic) Reference on games, gaming, simulation: Shubik, Operations Research, 2002 2/21/2003
Kleijnen: BEE; CentER 20 Feb. 2003
9
Kleijnen, J. (1980), Computers and profits: quantifying financial benefits of information. Addison-Wesley, Reading (Massachusetts) Kleijnen, Sanchez, Lucas & Cioppa (2002), A user's guide to the brave new world of designing simulation experiments Shubik, M. (2002), Game theory and Operations Research: some musings 50 years later. Operations Research, 50, no. 1, pp. 192-196 Gotts, et al. (2003), Agent-based simulation in the study of social dilemmas. AI Review , 19, no. 1, pp. 3-92 2/21/2003
Kleijnen: BEE; CentER 20 Feb. 2003
10
2