Political kidnaping as a deadly game

Simulation & Gaming http://sag.sagepub.com/

Political Kidnaping as a Deadly Game Lawrence C. Hamilton Simulation Gaming 1980 11: 387 DOI: 10.1177/104687818001100401 The online version of this article can be found at: http://sag.sagepub.com/content/11/4/387

Published by: http://www.sagepublications.com

On behalf of: Association for Business Simulation & Experiential Learning International Simulation & Gaming Association Japan Association of Simulation & Gaming

North American Simulation & Gaming Association Society for Intercultural Education, Training, & Research

Additional services and information for Simulation & Gaming can be found at: Email Alerts: http://sag.sagepub.com/cgi/alerts Subscriptions: http://sag.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://sag.sagepub.com/content/11/4/387.refs.html

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

>> Version of Record - Dec 1, 1980 What is This?

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

POLITICAL KIDNAPING AS A DEADLY GAME LAWRENCE C. HAMILTON University of New Hampshire

INTRODUCTION On August 28, 1968, two vehicles forced U.S. Ambassador John Gordon Mein’s car to the side of a street in downtown Guatemala City. As guerrillas armed with automatic weapons approached, Mein jumped out the car’s opposite door and began to run. He was shot and killed. The following day the Fuerzas Armada Rebeldes (FAR) issued a statement saying that they had planned to kidnap Mein and ransom him for an imprisoned FAR leader, Camilio Sanchez. The Mein incident brought a new, international dimension to the series of political kidnapings that occurred as Latin American guerrillas turned to the cities in the late 1960s. Together with the right-wing abduction of a Catholic prelate in March of 1968, it also demonstrated the intense publicity and official anxiety which

AUTHOR’S NOTE: I am indebted to John Bramsen, Otomar Bartos, Brian Jenkins, and Edward Mickolusfor helpful suggestions during the course of this work. An earlier version of the paper was presented at the annual meeting of the Pacific Sociological Association in Sacramento, April 1977. SIMULATION & GAMES, Vol. I1 No. 4, December 1980 387402 @ 1980 Sage Publications, Inc

387 Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

388

could be generated by the kidnaping of important persons. The tactic was soon successfully copied in Brazil, where a group headed by Carlos Marighela (author of the widely-read Minimanual of the Urban Guerrilla) exchanged U.S. Ambassador Elbrick for 1 ~ prisoners and the publication of a manifesto. Other incidents followed as the practice spread rapidly through Latin America and began to appear elsewhere. Political kidnaping is only one of the several new tactics adopted by international terrorists in recent years. Compared with such other innovations as bomb mailings, aircraft hijackings, and rocket attacks, it has not in itself been particularly lethal or destructive. It is, however, one of the most successful of the new tactics, for reasons inherent in the structure of the kidnaping ~‘~an~e.&dquo; Kidnapings have been profitable: Argentinian terrorists reportedly received over $14 million for an Exxon representative (Jenkins, 1974) and $60 million for two Bunge and Brn

officials (Alexander, 1976). Political kidnapings have also demonstrated a formidable ability to attract public attention. The publicity is particularly important to terrorist groups, because it emphasizes their own strength (small though it may actually be) and their target’s weakness, thereby encouraging supporters and intimidating

Company

cause. Taking the Symbionese Liberation point, Jenkins (1974: 3) has remarked on the gap that may develop between &dquo;the actual amount of violence and the greatly amplified effects of that violence.&dquo; He goes on to note: Terrorism is psychological warfare. It is theater. Publicity is one of the major rewards sought by terrorists. A prominent kidnaping virtually guarantees this reward. Terrorism, and political kidnaping in particular, is a battle of threats more than weapons. Its primary damage is not in lives

opponents of their

Army as

a case

in

-

lost-these may be few-bnt in the effects of its coercion. Since terrorist groups are generally too small to seize state power themselves, they aim to provoke the state into repressive countermeasures that will make life miserable enough for the masses to push them toward the terrorist’s cause. This is the classic strategy of terrorism, clearly articulated in (for example) Marighela’s Minimanual. The structural power which hostage situations give

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

389

the kidnaper,and the wide publicity which international victims insure, have made such incidents quite effective in provoking heavy-handed reactions. The government of Uruguay moved from a democracy (the &dquo;Switzerland of Latin America&dquo;) to a military dictatorship under pressure from an impressive Tupamaro terrorist campaign. In 1970, the Canadian government responded to the Cross and Laporte kidnapings by putting the country on a wartime footing and making some 500 arrests. Following the massacre of Israeli athletes at Munich in 1972, Israeli jets bombed Lebanese refugee camps, shot down a Libyan airliner with 107 people aboard, and sent out teams of assassins against Palestinian operatives in Europe. These acts cost Israel some of its much-needed international support. It is undeniable that terrorists have frequently succeeded in provoking governments into unpopular and undemocratic countermeasures. But whether or not this ultimately advances the insurgents’ causes, it clearly involves them in grave short-run risks. When Uruguay’s military rulers finally called in the army, the Tupamaros--which had been considered one of the most successful and most innovative terrorist groups in the worldwere quickly destroyed. By 1~’~~, a journalist could complain of being unable to find a single member of this once flamboyant group still in the country (McKnight, 1974). In the wake of Laporte’s murder, the Canadian government crushed the Quebec Liberation Front (FLQ). The Israeli counterterror offensive kiiled a number of prominent Black September agents, and forced many others into a defensive and fugitive lifestyle (Tinnin and Christensen, 1976). Marighela himself was killed in the police crackdown that followed the successful Elbrick kidnaping in Brazil. The grim doings of terrorists and counterterrorists seem far from the activities normally visualized as &dquo;games.&dquo; However, some of the central elements of political ~idnapin~-~onflicts of

to

interest, alternative choices, and anticipations regarding the consequences of those choices-are common to the family of &dquo;games of strategy&dquo; (see Rapoport, 1960: 108). The abstract theory of games may therefore provide useful insights into the logical structure behind all-too-concrete situations of political

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

390

kidnaping. The parallels between the two are strengthened by the importance of threats, commitment, and bargaining in both the theory of games and the practice of political kidnaping. Political kidnapings are modeled below as two-person nonzero-sum games of policy. The models provide a framework for examining the short-term power and long-term vulnerability of the kidnaper; the short- and long-term vulnerabilities of the ransom target; and the overall consequences of repeated plays for both players. The game models employed are extremely simple ones, in contrast to the elaborate theoretical games which have

been have

developed elsewhere. Although simple, these models do advantages in transparency and in clarity of empirical

referents.

Describing kidnaping as a &dquo;game&dquo; is emphatically not meant to imply trivialization of such life-and-death confrontations. In this connection, it may be noted that a prominent earlier application of game theory was to the infinitely more serious (but, structurally, not altogether unrelated) problems of nuclear deterrence (for examples, see Ellsberg,1960; Williams, 1966; also criticisms in Green, 1966; and Rapoport, 1964). POLICY GAMES MODELS ~

For the theoretical purposes, the

kidnaping

situation will be

reduced to the simplest possible terms, as a two-person, two. move game with four possible outcomes. The game begins assuming that a kidnaping situation already exists; the decision to play such a game (i.e., the decision of the kidnaper to kidnap) will be examined later. The first choice appears to be that of the ransom target, who may pay or not pay whatever ransom is demanded. Once that is done, the kidnaper moves, by freeing or executing the hostage. Actually, the individual moves at this level may be necessary consequences of an earlier commitment to a policy, and the real game may consist of the choices of these

policies.

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

391

Figure 1: Terrorism

and

Target Outcome Preferences With Resulting Policy Games

There are a number of unique features of political kidnaping which encourage the formation of such policies. A political kidnaping, unlike a merely criminal kidnaping, often represents only one move in a larger, continuing game. The utilities (valued objectives) of both players are therefore influenced by considerations of the effect any single outcome may have on future plays. Also, the motives and constraints of players in political kidnapings differ from those in criminal cases. The criminal kidnaper’s main concern is to achieve the difficult task of getting away with

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

392

the

money. The political kidnaper, who may demand in payment prisoners, publicity, or &dquo;charitable&dquo; contributions and be able to seek shelter in a neighboring country, can afford to be less preoccupied with the problems of escape. Because political kidnapings, like other terrorist acts, are incidents in a multiple-move battle of threats, it may be admantageous for both players to convince the other thay they have made a policy commitment. &dquo;In bargaining, the commitment is a device to leave the last clear chance to decide the outcome with the other party&dquo; (Schelling, 1960: 37). Such policy commitments provide a basis for describing kidnaping in formal gametheoretical terms, as developed in Figure 1. In the A and B matrices shown, two possible orders in which the terrorists could prefer the outcomes are shown. The utilities depicted are strictly ordinal (4 is preferred to 3, which is preferred to 2, which is preferred to I). The matrices at this point are not intended to represent a game. Matrix A displays a short-sighted set of preferences, similar to those that would be expected for a criminal kidnaper. For such a kidnaper, ransom is the major goal, and the fate of the hostage is relatively un~~nlaortant. With repeated plays, however, terrorists might be forced to reorder their priorities toward the configuration of matrix B. This is because a pay ransom kill hostage outcome would discourage the target from ever paying ransom a~~~u~ubvi~u~ly to the tcrrorransom

From this point of view, a no ransom/free hostage might be even worse. Matrix B describes terrorists whose primary concerns are with perpotuating the game and enhancing their future chances of winning (i.e., achieving pay ransom/free hostage). One such group, the Tupamaros of Uruguay, explained their 1 ~?t~ murder of a kidnaped and unransomed American police advisor, Dan Mitrione, in the following terxn~;

~~t’s longterm

disadvantage.

outcome

logic of the technique of kidnapings to get the release of prisoners has to be followed all the way if it is to remain effective [Moss, 19&dquo;~~: 228].

The

A similar

shift,

over

time, in the utility

structure of ransom

targets may also be expected. In a one-shot kidnaping affair (such

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

393

most criminal cases), the target’s main concern is for the safety of the hostage. Ordinal utilities representing that concern are shown in matrix D (Figure I), Payment of ransom is, understandably, the almost universal choice of ransom targets (often relatives of the hostage) in criminal kidnapings. As will be discussed later, some terrorists have deliberately acted to insure as

that the targets they picked would possess these short-term outlooks and preferences. The utilities of the D matrix will result in moves of pay ransom, which allow the terrorist to continually win. The result of such submission is an inevitable loss of power and ransom to the terrorist; demands may well escalate too. As noted earlier, the cash ransoms paid to terrorists in Argentina grew rapidly in the early 1970s. In Brazil, after receiving 1 ~ prisoners for Elbrick, the American ambassador, kidnapers successively demanded, and got, 40 prisoners for a West German and 70 for a Swiss. These pressures must eventually push frequent targets toward harder, no-ransotn approaches as the only means of stopping the victories of their opponents. Such attitudes would result from the preferences shown in matrix C (Figure 1). With the ordinal utilities described ill matrices A; B, C, and D, it is possible to formulate the policy games that could arise; such games are pictured at the bottom of Figure 1. Ransom targets have two possible policy choices: commitment to paying ransom whenever it is demanded, or commitment to not paying ransom under any circumstances. The latter policy has become official with such popular targets as the U.S., Israel, and Turkey. Other governments, including Canada, Brazil, and Uruguay have also adopted hard no-ransom policies. The terrorist policy options are slightly more complex. They can free the hostages unconditionally. They can kill the hostages unconditionally. They can free the hostages if ransom is paid (&dquo;free-if in matrices B and F) and kill them if it is not. Or they can kill the hostages if ransom is paid and free them if ? is not. This last alternative makes no sense in the present context, and will be

discarded. Several deductions are now possible. The game played between the short-sighted A terrorist and the short-sighted D target is

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

394

by matrix E; the target has no dominant strategy.2 The terrorists, on the other hand, can maximize their payoff by playing &dquo;free-if’ no matter what the target does. It is in the terrorists’ interest to convince their target (perhaps through previous plays of the game) that they are committed to do this. Since such a commitment is highly plausible, &dquo;pay unconditiana.lly&dquo; is the most rational move for the target in matrix E; it alone will insure the hostage’s safety. Pay/free-if becomes the equilibrium or saddle point of the game: both players can only do worse by changing their move. In individual incidents, the equilibrium would result in a series of pay ransom/free hostage shown

~ ‘

outcomes, or continual terrorist wins. It is expected that, over time, both players’ preferences will change. The game then shifts towards that produced by a meeting of the B terrorist and the C target. In this game (shown as matrix F in Figure 1), the terrorist’s dominant strategy remains a policy of &dquo;free-if.&dquo; However, this time the target too has a dominant strategy: pay no ransom. No ransom free-if is the new equilibrum point, resulting in a sequence of no ransom/ kill hostage outcomes. Both players lose on each play, and long-term victory becomes a matter of which player can continue to accept their losses the longest. The prediction resulting from these arguments is simple: over many plays involving the same or related players, political kidnapings will tend to shift from the E game to the F one, and hence their outcomes will shift from pay ransom/free hostage to no ransom, kill hostage. Advocates of hard-line, no ransom responses add a third stage: the outcome will eventually shift to no kidnapings. Their conclusion assumes that the ransom target can absorb the damage of lost hostages longer than the kidnapers can absorb the damage of police crackdowns, loss of public sympathy, etc. There is some reason to believe that this is so. Past instances in Canada, Brazil, and Uruguay, where the kidnapers chose to kill an unransomed hostage, proved ultimately disastrous for the kidnapers. The issue invites more formal empirical study. Good data sets on terrorism are not yet widely available but may soon become so.

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

395

In the interim

make do with somewhat older data. Figures supplied by the International Civil Aviation Organization covering 1969-1974 skyjackings and sabotage provide some support for the game-theoretical prediction of a shift toward no ransom/ kill hostage outcomes. ’While skyjacking and sabotage incidents substantially decreased during this period, the ones that did occur were more deadly. Both trends presumably reflect heightened airport security precautions, a version of the noransom policy that led to some bloody airport shoot-outs. Whether the new security (no ransom) emphasis had this effect because it more frequently forced incidents to a violent conclusion, or because it deterred hijackers less prone to violence in the first place is unclear. Bust the ratio of casualties per attempt increased more than tenfold from 19711 to 1973. Data on diplomatic kidnapings from 1968-1974 (from Jenkins and Johnson, 1975) show no such clear trend, partly because kidnapers facing a committed no-ransom government became more flexible at directing their ransom demands elsewhere. For example, Palestinian terrorists shifted from demands aimed at Israel to demands aimed at less committed third parties, such as the governments of Austria, Belgium, or Greece. As more Latin American governments adopted no-ransom policies, guerrillas there shifted their demands to the hostages’ families and employers. The new targets are much less likely to take the long view of things; their priorities are those shown in D (Figure 1~~ whatever the government’s position, A more detailed analysis of specific countries is needed to explore the question further. one can

INCIDENT DECISION MODELS

As has been stated, a strategy of commitment may be most rational for both players, in the long run. Most players, however, do not yet appear to have made such commitments and continue to decide their moves on a case-by-case basis. It is, therefore, of interest to more closely examine the details of such decisions.

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

396

Figure 2: Outcomes of International Political Kidnapings, 1969-1974 SOURCE: Jenkins and Johnson, 1975.

Figure 2 shows the numbers and proportions of choices made in 75 international kidnaping incidents, from 1969 to 1974 (data from Jenkins and Johnson, 1975). Ransom targets exhibited two principal tactics of noncooperation: doing nothing (i.e., paying no ransom), and attacking the kidnaper. During their heyday, the secure Tupamaros of Uruguay were able to add a second noncooperative option for kidnapers: they kept a hostage indefinitely, in a &dquo;people’s prison.&dquo; Recent developments in Iran have taken this alternative to unanticipated extremes.3 Some preliminary generalizations are possible. If ransom was paid, the terrorists were almost certain (p .98) to free the hostage. They were somewhat likely (p .69) to free the hostage anyway if the target ignored them, and about equally likely (p .62) to kill the hostages if they were attacked. Since attacking the kidnaper carries such a high likelihood of getting the hostages killed, it may be motivated by more than a simple desire for their unransomed release. Presumably, attacks =

=

=

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

397

also serve other functions sueh as dramatizing the target’s power and uncooperativeness, and intimidating future terrorists, regardldss of the co9t, The positive American response to President Ford’s handling of the Mayaguez incident, in which more Americans died than were freed, illustrates the potential success of this tactic. In his film State of Seige, a fictionalized account of the 1970 Mitrione kidnaping in Uruguay, the French director Costa-Gravas suggested that the death of an &dquo;expendable&dquo; hostage may actually be desirable to the target (in that case, the U.S.), because it mobilizes public opinion against the kidnapers and in favor of strong governmental countermeasures. The empirical success of the &dquo;do nothing&dquo; tactic in obtaining the release of hostages is also interesting. Baldwin’s (1976: 408) observations about hijackers are probably relevant here:

People suffer from fatigue, but governments do not. This asymmetry of the vulnerability of participants to fatigue and pressure may be one reason why a strategy of &dquo;waiting the hijacker out&dquo; has seemed to work so well.... Most hijackers are unable to create a situation in which the &dquo;last clear chance&dquo; to avert disaster lies-or appears to lie-with the government. are important weaknesses in the terrorists’ position; if initial threats fail, then they have little remaining leverage with which to insure their own safety. The argument also supports the view that most long-term battles of will between equally hardline terrorists and targets will result in target victories, if no other forces intervene. Ric Blacksten (1974: 10) has proposed a more detailed gametheoretical model with some similarities to those of Figure 1. In his analysis, both the Nash threat and the Nash bargaining solutions led to the conclusion that pay transom free hostage was the &dquo;fair&dquo; (in the sense of balanced-power) outcome of the kidnaping game: &dquo;The fact that the guerrillas are granted their demands simply reflects the power of their position.&dquo; But, empirically, the guerrillas are not always granted their demands, and it is evident that when they aren’t the &dquo;power of their

These

position&dquo; may become a liability. A kidnaper can then punish the

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

398

target, by killing the hostage, but the target may be able to survive such punishment better than the terrorist can when the target no reason to refrain from acting out its counterthreats. If observed proportions such as those shown in Figure 2 are

longer has

taken to represent the a priori probabilities of an opponent’s responses, then they could serve as a basis for identifying the most rational decisions. To do so, one more bit of information is needed: the ratio utilities of the players. For purposes of demonstration only, these will be assumed to equal the ordinal utilities specified in matrix F of Figure 1. That is:

&dquo;keep hostage&dquo; option is left calculations, though it could readily be included. To

simplify,

the

out

of these

Terrorist-it is rational to kidnap if .68(4) + .32(3) is larger than the utility of the status quo; (2) Target-it is rational to pay ransom if .98(2) + .02( 1) is largest; to attack if .37(4) + .62(3) is largest; or to do nothing if .69(4) +.25(3) is largest.

(1) I

If, in keeping with these arbitrary values the terrorists’ &dquo;status quo&dquo; is assigned a utility of 3.5, then it is rational for the terrorist to kidnap (3.68 > 3.5). It is then rational for the target to do nothing (3.51 > 3.34 > 1.98). Players capable of specifying their actual utilities could, in

principle, use the methods outlined above to determine the most rational choices. The analysis might be elaborated to include estimates of opponent utilities, and of opponents’ perceptions

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

399

about one’s own utilities and probabilities, and Otomar Bartos (1976: 490) has observed,

so on.

But,

as

ironically, publications of scientific findings of this kind may influenced the negotiation process itself and, in so doing, render the original findings inapplicable. A destructive circularity emerges when each player bases his or own decisions on assumptions about the opponent’s rationality. The specification of utilities would pose a considerable her

methodological problem anyway, and on balance it is hard to imagine real-life terrorists or targets placing a great deal of faith in such calculations. These obstacles have not deterred some methodologically sophisticated recent efforts to generate em-

Figure 3: Payoffs

in the

Kidnaping Game, Relative to the Status Quo laxesl

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

400

pirical utility estimates for hostage situations (Saaty and Bennett, 1978). The

above may be seen merely as some small use in the stillof subjective process decision-making. A larger and better sample would make it possible to use specifically relevant subsets o~ the total cases, such as previous behavior of one organization, or the most recent worldwide trends. The question would then arise of exactly which subsets were the most relevant.

probabilities computed

additional knowledge, possibly of

CONCLUSIONS

.

Figure 3 shows joint payoffs for possible outcomes of a kidnaping game played by two hard-line antagonists (matrices B and C in Figure 1). As before, I have treated the ordinal utilities as if they were ratio level for purposes of illustration. Even for a hard-line terrorist, the status quo is likely to seem preferable to outcomes of no ransom/ kill hostage. Although such outcomes demonstrate the kidnapers’ resolve, they have historically been very costly to kidnapers in terms of public support and government suppression. Hence the status quo for terrorists is arbitrarily assigned a utility of 3.5, between that of no ransom j kill hostage (3) and pay ransom/ fr0e hostage (4). For targets, hard-line or otherwise, the best to be hoped for is the freeing of an unransomed hostage-a return to the disturbed status quo. The target status quo is therefore given the same value (4) as the no ransom/free hostage outcome. The specific numerical values are not otherwise important to the conclusions that may be drawn from Figure 3. Figure 3 dramatizes the fact that whereas the kidnapers may hope to &dquo;win&dquo; (achieve a payoff superior to the prekidnaping status quo) at this game, the target cannot. This no doubt contributes to the attractiveness of kidnaping as an insurgent tactic. However, if the target moves to prevent a kidnaper &dquo;win,&dquo; by paying no ransom, then the game is necessarily negative-sum. If the kidnapers maintain their hard line, both players must lose

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

401

(accept worse than status-quo payoffs) at every turn. That is the meaning of the equilibrium point in the game of Figure IF. Seen in this light, the kidnaping tactic should look less attractive. The public participates in the kidnaping game in many roles including these of kidnaping victims and their relatives, innocent bystanders, and subjects of government countermeasures. If political kidnaping become a negative-sum game for kidnapers and their targets, its consequences will also be negative, perhaps much more so, four the society at large. °

NOTES 1. The Minimanual is reprinted as an appendix to Moss (1971). 2. A dominant strategy is one which will result in the highest possible payoff no matter what the opponent does. The "free-if" strategy is dominant for the terrorist in 1E because the terrorist cannot do better by another strategy, regardless of whether the target does or does not ransom. The target in 1 E has no dominant strategy: if the terrorist chooses "freeif," it is best for the target to pay; but if the terrorist chooses "free" or "kill," it is best for the target not to pay. 3. Iran is certainly the most important instance of political kidnaping to date. However, it is singular in every respect and has been left outside of the scope of the present

analysis.

REFERENCES ALEXANDER, Y. [ed.] (1976) International Terrorism. New York: Praeger. BALDWIN, D. A. (1976) "Bargaining with airline hijackers," in I. W. Zartman (ed.) The 50% Solution. Garden City, NY: Anchor/ Doubleday. BARTOS, O. J. (1976) "How predictable are negotiations?" in I. W. Zartman (ed.) The 50% Solution. Gardon City, NY: Anchor/Doubleday. BLACKSTEN, R. and R. ENGLER (1974) "Hostage studies." Arlington, VA: Kentron. ELLSBERG, D. (1960) "The crude analysis of strategic choices." Santa Monica, CA: Rand.

GREEN, P. (1966) Deadly Logic. Columbus: Ohio State Univ. Press. JENKINS, G. (1974) "Terrorism and kidnapping." Santa Monica, CA: Rand. and J. JOHNSON (1975) International Terrorism: A Chronology, 1968-1974. Santa Monica, CA: Rand. McKNIGHT, G. (1974) The Terrorist Mind. Indianapolis: Bobbs-Merrill. ———

Downloaded from sag.sagepub.com at UNIV OF NEW HAMPSHIRE on March 3, 2014

402

MOSS, R. (1972) The War for the Cities. New York: Coward, McCann & Geoghegan. (1971) "Urban guerrilla warfare." Adelphi Papers No. 79. London: International ———

Institute for

Strategic Studies.

RAPOPORT, A. (1964) Strategy and Conscience. New York: Harper & Row. (1960) Fights, Games, and Debates. Ann Arbor: Univ. of Michigan Press. SAATY, T. L. and J. P. BENNETT (1978) "Theoretical aspects of negotiating with ter-

———

at The Institute of Management Science/Operations Research of America, New York. SCHELLING, T. C. (1960) The Strategy of Conflict. Cambridge, MA: Harvard Univ. Press. TINNIN, D. B. and D. CHRISTENSEN (1976) The Hit Team. Boston: Little Brown. WILLIAMS, J. D. (1966) The Compleat Strategist. New York: McGraw-Hill.

rorists." Presented

Society