A current challenge in Command and Control is "Time-sensitive Targeting", in which assets (e.g., bomber planes) are assigned to targets (e.g., hostile sites) in a manner that optimizes the overall effectiveness of an operation. The problem is complicated by time pressure because assets must be assigned and often diverted from scheduled targets in order to destroy "pop-up" targets before these targets escape or attack. We are researching the cognitive challenges of the asset-target "pairing problem" from both a theoretical perspective and a practical perspective. From a theoretical perspective, we developed a game called Tictac Tank that captures the basic challenge so we can perform laboratory experiments on the "mental models" used to solve pairing problems.  

Play Tictac Tank

Of course...When you pick a cell (bar), this excludes other cells in the same column and row since each guard can be assigned to only one prisoner. This constraint, plus time pressure, is what makes the game hard. Below on the left is how one game looks at the start. Below on the right is one player's solution (yellow cells), which took about 10 seconds to get. Notice that one bar (row D, column c) is not too high. Could you do better? Could you do it faster? How close can you get to the optimal solution (computed by a mathematical algorithm) and how fast can you do it? Try the game (above left) to see for yourself.

From a practical perspective, we designed a "support system" to help targeteers solve similar problems of time-sensitive targeting in the real world. Below is a screen shot of the system, called Bar-Gain Boxes (Burns, 2004a; 2004b; Burns & Means, 2005).

Try Bar-Gain Boxes

You can try a demo version of Bar-Gain Boxes by clicking the icon or link to the left.  

More details on this support system are available in Burns, 2004a; 2004b; Burns & Means, 2005).

In its basic approach, this support system (Bar-Gain Boxes) for the "pairing problem" is much like another support system (Bayesian Boxes) that we designed for "probability problems". In both cases, the design is based on the fact that a system's user needs to see the reasons behind the system's results if he is to understand and use the system's advice. This commonsense notion, which is surprisingly ignored in the design of many computerized support systems, is expressed in a famous quote by Pablo Picasso:

"Computers are useless. They can only give you answers."

Although science and art often have different means and ends, they are not always at odds (see www. ask-how.org). Picasso's quote serves to highlight the major problem with most support systems - namely that they are not able to interact with their users in the same ways that people interact with each other. One way to help bridge this gap is with "dynamic diagrams" that are both informative and interactive - as described in a paper titled Painting Pictures to Augment Advice.  

In warfare, two critical challenges of Command and Control are: (1) Inference, of enemy positions and intentions from limited information, and (2) Investment, of limited assets against potential targets in order to achieve desired outcomes. In poker, the same two challenges arise because a player must make: (1) Inferences, about the strength of his own hand against his opponents' hands based on limited information obtained from his opponents' bets, (2) Investments, of his own chips against his opponents' chips in order to win chips from the pot. In fact, as noted by McDonald (1950), the theory of games originated in poker and poker remains the ideal model of the basic strategic problem in warfare.

Mathematically speaking, poker is a game of “imperfect information that ... reflects the decision making challenges of real world domains” (Billings, et al., 2002). Psychologically speaking, “Poker contains a greater element of skill than any other card game including Contract Bridge… because poker is a game of money management in addition to card management. It is a game where there is a correct technical play in every situation. It is a game where the best possible hand need not win the pot, due to a bluff.” (Scarne, 1980). These features make poker an ideal tool for our research on how people make decisions and how systems might help people make better decisions.  

The problem with standard poker, played with 5-card hands dealt from a 52-card deck, is that the game is too complex to compute the "optimal" strategy for any given situation. This makes it difficult to assess exactly how well people play and exactly how much better they might play if they were given various computer "support systems". Thus, our research uses a suite of Pared-down Poker games that were specifically designed for research on human judgment and decision making in prototypical situations of Command and Control. The basic design and early research on Pared-down Poker are documented in several papers (Burns, 2004c; 2005). 

You can play a demo version of Pared-down Poker by clicking the icon below:

We are currently planning experiments to measure and model human performance in Pared-down Poker with respect to three key topics in Command and Control, which are summarized below. 

Expected Value: “Resource Management” in Command and Control (like betting chips in poker) depends on the probability (P) and utility (Q) of possible outcomes (like winning the pot in poker) from available options (like betting or folding in poker). Mathematically speaking, the expected value of an option is given by P*Q, with losses having negative Q values. Psychologically speaking, the question is: How do people represent and calculate the Ps and Qs in order to choose the best option? We will answer this question by measuring people's judgments of “confidence” (P) and “consequence” (Q) in poker playing, along with the betting choices (i.e., bet, raise, fold) that they make in the context of their subjective beliefs.

Bayesian Inference: “Risk Assessment” in Command and Control (like judging odds in poker) depends on the ability to update probabilities (P) as situations unfold and as new information is gained. Our thesis is that people generally reason in accordance with the norms of Bayesian inference, but that their judgments are bounded by the "mind sets" they use to represent possibilities and the "short cuts" they use to estimate probabilities. The question is: How can these mind sets and short cuts be augmented by support systems? We will answer this question by developing and evaluating prototype systems in the context of our poker testbed.

Opponent Models: “Rational Engagement” in Command and Control (like knowing when to hold or fold in poker) depends on the ability to model (understand and anticipate) the behavior of one’s opponents. The question is: How do people construct and update their models of opponents, and how do they use these models to adjust their own decision making strategies? This problem is central to good poker playing, and in fact the problem of opponent modeling is the main reason that Artificial Intelligence has not achieved the same success in poker-playing programs that it has achieved in programs playing chess and other games of "perfect information" (which can be won by brute-force search algorithms that do not require opponent modeling). The problem of opponent modeling is also crucial in warfare, especially in asymmetric warfare like the war on terror where one’s self-image may not be a good model of one’s opponent.

Billings, D., Aaron, D., Schaeffer, J., & Szafron, D. (2002). The Challenge of Poker. Artificial Intelligence Journal, Volume 134, 201-240.  

Burns, K. (2004a). Bar-Gain Boxes: An Informative Illustration of the Pairing Problem. Proceedings of the 3rd International Conference on the Theory and Application of Diagrams. 

Burns, K. (2004b). Painting Pictures to Augment Advice. Proceedings of the 7th International Conference on Advanced Visual Interfaces. 

Burns, K. (2004c). Heads-Up Face-Off: On Style and Skill in the Game of Poker.  Proceedings of the American Association for Artificial Intelligence, Symposium on Style in Language, Art, Music and Design. 

Burns, K. (2005). Pared-down Poker: Cutting to the Core of Command and Control. Proceedings of the IEEE Symposium on Computational Intelligence and Games.

Burns, K., & Means, C. D. (2005).  Dynamic Diagrams for Time-Sensitive Targeting. Proceedings of the Second Annual Integrated Sensing and Decision Support Workshop.

McDonald, J. (1950). Strategy in Poker, Business and War (W. W. Norton).

Scarne, J. (1980). Scarne’s Guide to Modern Poker (Simon & Schuster).