Indefatigable Cost-Benefit Analysts, Immersed in and Endless Succession of Subtle Non-Zero-Sum Games
May 1, 2005
An unusual presentation of the prisoner�s dilemma. The players�Student (the author) and Dragon�have the options of staying mum (M) or confessing (C). The open circles are the players� decision nodes; the solid circles are terminal nodes or the end of the game; the numbers are the payoffs to the players. From the appendix, Games Prisoners Play.
Book Review
James Case
Games Prisoners Play: The Tragicomic Worlds of Polish Prison. By Marek M. Kaminski, Princeton University Press, Princeton, New Jersey, 2004, xvi + 207 pages, $29.95.
Glad tidings I bring for readers consumed by curiosity concerning Polish prison culture during the waning years of communist rule. Everything you ever wanted to know about the subject, and much, much more, is revealed in Marek Kaminski�s sometimes hilarious, often horrific, occasionally disgusting, yet immensely revealing expos� of the culture and conditions that evolved in Polish penal institutions during the decades preceding the fall of the Berlin Wall. The author�s central thesis is that prison life, by its very nature, obliges inmates to function as indefatigable cost�benefit analysts immersed in a seemingly endless succession of subtle non-zero-sum games, of which the endlessly discussed �prisoner�s dilemma� is but one example.
It was quite by accident that Kaminski acquired the expertise needed to write such a book. He and his driver, �Dragon,� were apprehended by Warsaw police on March 12, 1985, in a van filled with anticommunist (Solidarnosc) literature. This was hardly surprising, as Kaminski (a.k.a. �Student�) was at the time running an underground publishing house known as STOP, which occupied about twenty full-time and perhaps a hundred part-time employees. Between 1982 and 1989, STOP published some thirty-five clandestine titles, of which it managed to sell about 100,000 copies. It was also part of a decentralized network of pamphleteers, trade union organizers, and anticommunist activists�including several Nobel Prize winners�--along with an ever-changing menu of underground theaters, art galleries, and video rental agencies. Members described this network as an �independent society.�
As matters were explained to him by a cellmate�--whom he later discovered to be a snitch placed there by authorities�--the state had plenty of evidence to convict him of possession and sentence him to a year in prison, but would have been unable to make a more serious �intent-to-distribute� charge stick without extracting a confession from either him or his driver. Unable to communicate with one another, he and Dragon faced the classic prisoner�s dilemma: to rat or not to rat. Dragon did, the author didn�t, and his sojourn at Bialolecka Jail commenced a mere three days after his apprehension. He raises some interesting questions, near the end of the book, concerning the degree to which prisoner�s dilemma models the decision problems newly apprehended prisoners actually face. How, for instance, were he and Dragon to account for the possibility that (as actually happened after only five months) he would be able to use the medical knowledge and connections of his parents (both MDs) to gain an early medical release?
By the second day of his incarceration, Kaminski had already resolved to make the best of his personal misfortune by performing a sociological study of the bizarre, terrifying world of rapes, knife fights, suicides, blunt talk, and self-inflicted injuries he was about to enter. He felt marginally qualified to undertake such a study, at the somewhat tender age of 22, having switched his undergraduate major (some two years earlier) from mathematics to social science. To his considerable surprise, he found prison research to be an excellent survival strategy, sparing him endless hours of useless introspection, and helping him to endure his new role as an inmate while insulating him from the more debilitating aspects of prison culture. In retrospect, he recommends prison research to anyone facing a lengthy sentence.
Following his release, Kaminski resumed his undergraduate studies, drawing on his prison experience for several term papers--�drafts of which he had had the foresight to smuggle out of prison�--a thesis, and (in time) his first published research. Yet he remained frustrated by what he perceived as a lack of analytic depth in his work. Only when a graduate thesis adviser suggested that he construct game-theoretic models of critical inmate interactions did he feel that he had begun to penetrate the core of his subject. As a result, he returned to his previously abandoned mathematical studies, becoming a game theorist specializing in applications to social science. (He is now an assistant professor of political science and mathematical behavioral science at the University of California, Irvine.) Without denying that game-theoretic models have a disturbing tendency to reveal only the obvious, he wrote the book under review in order to share the insights to which his more penetrating analyses have guided him. The bulk of those insights concern the rigid caste system around which Polish prison culture is organized.
The dominant caste, which Kaminski labels �grypsmen,� maintains and enforces an elaborate code of conduct reputedly derived from an earlier code applicable only to thieves. A �gryps� is a secret message, addressed either to a fellow inmate or to someone on the outside. Grypsmen are prisoners deemed trustworthy enough to convey such messages to and from their peers. If intercepted by authorities, such messages can expose sender, messenger, and recipient alike�--not to mention cellmates and other innocent bystanders�--to harsh punishment.
Members of the (smaller) middle caste are called �non-grypsmen,� or �suckers.� Grypsmen can lie to, cheat, or steal from them without fear of reprisal. To the non-grypsmen are assigned all manner of menial tasks, and from them is withheld anything remotely resembling sensitive information. The lowest (and smallest) caste consists of prisoners known to have provided sexual services to fellow inmates at some time in the (possibly remote) past. These �untouchables� are confined to remote corners of the cells they share with higher ranking inmates, and are typically allowed to sleep only on the floor. Skin-on-skin contact with such individuals�-even an innocent handshake�-can bring (permanent and irrevocable) demotion to the lowest caste. Suckers can occasionally be promoted to grypsmen, but members of the lowest caste are ineligible for promotion.
Because Polish prison authorities regularly transfer inmates from one prison or cell to another, suckers and members of the lowest caste have both opportunity and incentive to pose as grypsmen. While intramural transfers can usually be authenticated within hours, via the prison grapevine, those from distant facilities can take weeks to be vetted by inmates. And, because the presence of a possible snitch represents a serious threat to cellmates, determined efforts are made to identify impostors. To this end, grypsmen typically scrutinize a new cellmate�s knowledge of and adherence to the grypsman�s code of conduct. Kaminski points out that a number of the standard reliability tests constitute games between new and old inhabitants of a cell.
Because such contests involve small numbers of decisions from short lists of alternatives, Kaminski is able to treat them as �games in extensive form,� exhibiting the complete decision tree for each one. He contrives to do so even for contests with novel features�-e.g., options of which the player exercising them is not necessarily aware, or rewards of which the player might not know the correct value�-features ordinarily assumed away by game-theoretic model builders. As a result, Kaminski is able to reduce the mathematical complexity of his account to a bare minimum. His eminently readable book can be recommended to anyone with an interest in either prison life or unusual applications of game theory.
James Case writes from Baltimore, Maryland.
