A Critique of Wargaming as a research tool.

Dug up and Reproduced by Jim Wallman

I was looking through my old files the other day, when I came across a yellowing folder that I had called 'Theoretical Papers'. Having no recollection of what that might include, I spend some time reading through a collection of random scrps gleaned from various sources. Little of it was of any interest, except the following, which was a section of a RAND memorandum (RM-3489-PR) called 'War Gaming' by E.W.Paxson, dated February 1963 - which I found in the MoD Library some time in 1979 I think. I recall being very influenced by it at the time.
As you probably know, the RAND corporation was (and probably still is) an important think-tank for the US military-industrial complex.
I thought this extract is interesting, and I have re-typed it here because I think you might take some amusement from it, and perhaps see that concerns in 1963 are not too far from those of 2002. I have reproduced the extract exactly as copied, and resisted the temptation to insert my own comments and corrections in the text - though it was hard. In re-reading it a lot of points occur, but I'd like to read member's thoughts in MilMud first.



So that the conventions may be observed, something must be said about the strengths and weaknesses of war gaming. We are, of course, concerned only with gaming as a research tool. Consider first the advantages.
Gaming gives full scope to the enemy's actions and reactions, to his capability and his persistence. No staff planner writes a paper showing how his plan will fail and no engineer points out how a weapon system he has designed can be circumvented.(1) In these one-sides analyses, replete with detailed calculations, assume a somewhat chuckleheaded and rather friendly enemy whose course of action is transparently clear and rigid and who seems quite vague about what he will do later on as the plan or weapon system moves smoothly to success. Gaming, on the contrary, supplies the most hostile of reviews.
Gaming provides a framework within which experts in the relevant field can systematically apply and pool their ideas and techniques. This framework requires all relevant fields to be represented so that the logistician can tether the weapon designer, and the political scientist frustrate the operational planner. Each expert is forced to see a context which may modify his dogmatism and in fact stimulate his ingenuity.
By definition, a play of a war game is a sequence of moves each involving a concrete situation. Generalities and pontifications must show their applicability. Doctrine and belief are explicitly tested. The sequence of events forces a "thinking through" of an operation or campaign to the end.
Like Chinese medicine, the military profession is most successful if the client does not become a patient. War gaming can assist in designing a balanced deterrent posture, because we make our mistakes on paper and we can check plans and force structures for feasibility and balance. Since there is little secret about our military posture, one can hope that the enemy will conduct his own war games based on the inherent capabilities of that posture - not on its intentions - and detect no soft spots.(2)
But now it is the Criticics' turn. There are many of these people. Polemics, as well as semantics, is a characteristic of operations research. A discussion of their structures may bring out aspects of war gaming other than those reviewed above.

Criticic: Because of the enthusiasm and conviction war games engender in the participants, they lose all sense of logic, they confuse their model with reality.

Protagonist: Gaming is hard work. The gamers I know were either cynical when they started or lost their glow in a few weeks and never found it again.(3) They are well aware that the game is only a model; albeit sufficiently close to operations in the real world so that some guidance and insight are provided. If they are Players, they argue with Control about the validity of the assumptions and the assessments. They continually find branches that should be explored. If they are Controllers, they may beg the players to put on their old uniforms in order to reproduce lags arising from interservice "considerations."

Critic: Your players are completely persuaded by the elaborate detail and work put into play.

Protagonist: You are repeating yourself. Some of the players are not persuaded after ten plays. War gaming is not an exercise in self-hypnosis. But the Players do become convinced that a war game is an experiment, undoubtedly imperfect. They know that intuition and judgement intervene. This is equally true in any scientific endeavour. Any conclusions from the game or games must be subject to Critical review, independent of the details of the game. The important point is that the game produces new ideas or phrases alternative concepts sharply and sketches explicitly possible consequences.

Critic: Even so, your people are assuming roles they never approximated in the real world. Hence they will "learn" in unpredictable ways during the game sequence.

Protagonist: What kind of people do you hire to do responsible gaming? As for "learning", this is precisely the feature desired in gaming, although it should be called generating new research ideas and finding out what mistakes could be made. Out-and-out strategic or tactical errors must of course be located and the game re-assessed. Although a game may be finished, the gamers are not finished with the game.

Critic: You simply cannot produce a set of realistic rules. Your people will exploit them, they will "play the game" against itself.

Protagonist: Negative. You are dealing with a group which is no only research minded but which has (or should have) an appreciable continuity. When one side does not point our flagrant unrealities, the other certainly will. In the course of time, refinement of rules and procedures is inevitable.

Critic: Your games are larded with deceptive detail. Each detail is a parameter and each must have a value. You don't know these values and even worse, in the ensemble, you have a hopelessly uncontrolled situation in the statistical sense, because you can't possibly have enough plays. You should stick to analytic methods. Or if you insist on games, strip away all your ornaments and find a simpler formulation which will give you a precise statement of problems and results, and which will let you understand what is going on.

Protagonist: You are talking as emotionally as a war gamer. But your points are all worth attention. In fact, you have arrived at the right conclusion by the wrong course or argument.
First, there is a lot of detail in war games. The reason is that the designer does not know in advance what factors may prove Critical. He strives instead to cover what must be a range of contextual factors more or less clearly relevant to the game's objective. Reasoning by imperfect analogy is one of the thirty-four dishonest ways to win an argument, but here, comparing this phase of war gaming with dimensional analysis fail utterly if one essential physical quality is omitted. Systems analysis may produce a completely worked-out and "optimum" system which may also be completely wide of the mark because of operational factors not considered.
Second, each parameter must be assigned a value, or even better, a range of values. Choosing this value is a matter of workmanship, and exactly the same problem arises in the study of any problem in military operations research by any method whatsoever.
Third, the conscientious war gamer must always be conscious of his tolerances. He must calculate or assess, if only by seminaring, the sensitivity of his over-all conclusions to variation in parameter values. Of course, he is never satisfied that he has completed this job.
Fourth, the number of plays of any one game that should be made is usually grossly overexaggerated by appeal to superficial combinatorial arguments. Once the main assumption for a given branch of the game are set out, the flexibility in strategy and tactics is not nearly so rich as is presupposed, if only the main lines of the over-all course of events are to be made out. Perturbations of the main line can usually be assessed satisfactorily by those experienced in the game. A war game is no more chess than chess is a war game - in regard to the Criticicality of the fine structure.
Fifth, you are completely correct in saying that the real problem is to understand what is going on. Analysis of a sequence of war games is performed to deduce those few dominant key factors, assumptions, and general results whose relative weights and interactions are actually pertinent to the conclusions drawn. Most of the detail of the game now seems irrelevant.
Finally, war gaming is only one tool of military operations research. But there are many problems which must be tackled for which no other tool is yet available. Above all, gaming is a preresearch technique, generating ideas and sketching their outlines. Military operations research of necessity must be conducted with suspense dates in view. Gaming can make major contributions to pressing questions which cannot await refinements in methodology.

(1) In all fairness, in preparing his Estimate of the Situation the wise Commander may foster interaction between his intelligence and operations sections to beget opposing points of view.
(2) Unlike Admiral Yamamoto's war gaming of Pearl Harbour
(3) (sotto voce) Some are hard losers.