Notes on Methods and Motives: Exploring Links between Transnational Organized Crime & International Terrorism

In preparation for the work on this report, we reviewed a significant body of academic research on the structure and behavior of organized crime and terrorist groups. By examining how other scholars have approached the issues of organized crime or terrorism, we were able to refine our methodology. This novel approach combines a framework drawn from intelligence analysis with the tenets of a methodological approach devised by the criminologist Donald Cressey, who uses the metaphor of an archeological dig to systematize a search for information on organized crime.7 All the data and examples used to populate the model have been verified, and our findings have been validated through the rigorous application of case study methods.

 

While experts broadly accept no single definition of organized crime, a review of the numerous definitions offered identifies several central themes.8 There is consensus that at least two perpetrators are in- volved, but there is a variety of views about the way organized crime is typically organized as a hierarchy or as a network.9

 

Organized crime is a continuing enterprise, so does not include conspiracies that perpetrate single crimes and then go their separate ways. Furthermore, the overarching goals of organized crime groups are profit and power. Groups seek a balance between maximizing profits and minimizing their own risk, while striving for control by menacing certain businesses. Violence, or the threat of violence, is used to enforce obligations and maintain hegemony over rackets and enterprises such as extortion and narcotics smuggling. Corruption is a means of reducing the criminals’ own risk, maintaining control and making profits.

few definitions challenge the common view of organized crime as a ‘parallel government’ that seeks power at the expense of the state but retains patriotic or nationalistic ties to the state. This report takes up that challenge by illustrating the rise of a new class of criminal groups with little or no national allegiance. These criminals are ready to pro- vide services for terrorists as has been observed in European prisons.10

We prefer the definition offered by the UN Convention Against Transnational Organized Crime, which defines an organized crime group as “a structured group [that is not randomly formed for the im- mediate commission of an offense] of three or more persons, existing for a period of time and acting in concert with the aim of committing one or more serious crimes or offences [punishable by a deprivation of liberty of at least four years] established in accordance with this Convention, in order to obtain, directly or indirectly, a financial or other material benefit.”

we prefer the notion of a number of shadow economies, in the same way that macroeconomists use the global economy, comprising markets, sectors and national economies, as their basic unit of reference.

terrorism scholar Bruce Hoffman has offered a comprehensive and useful definition of terrorism as the deliberate creation and exploitation of fear through violence or the threat of violence in the pursuit of political change.15 Hoffman’s definition offers precise terms of reference while remaining comprehensive; he further notes that terrorism is ‘political in aims and motives,’ ‘violent,’ ‘designed to have far-reaching psychological repercussions beyond the immediate victim or target,’ and ‘conducted by an organization with an identifiable chain of command or conspiratorial cell structure.’ These elements include acts of terrorism by many different types of criminal groups, yet they clearly circumscribe the violent and other terrorist acts. Therefore, the Hoffman definition can be applied to both groups and activities, a crucial distinction for this methodology we propose in this report.

Early identification of terror-crime cooperation occurred in the 1980s and focused naturally on narcoterrorism, a phrase coined by Peru’s President Belaunde Terry to describe the terrorist attacks against anti-narcotics police in Peru.

the links between narcotics trafficking and terror groups exist in many regions of the world but that it is difficult to make generalizations about the terror- crime nexus.

 

International relations theorists have also produced a group of scholarly works that examine organized crime and terrorism (i.e., agents or processes) as objects of investigation for their paradigms. While in some cases, the frames of reference international relations scholars employed proved too general for the purposes of this report, the team found that these works demonstrated more environmental or behavioral aspects of the interaction.

2.3 Data collection

Much of the information in the report that follows was taken from open sources, including government reports, private and academic journal articles, court documents and media accounts.

To ensure accuracy in the collection of data, we adopted standards and methods to form criteria for accepting data from open sources. In order to improve accuracy and reduce bias, we attempted to corroborate every piece of data collected from one secondary source with data from a further source that was independent of the original source — that is, the second source did not quote the first source. Second, particularly when using media sources, we checked subsequent reporting by the same publication to find out whether the subject was described in the same way as before. Third, we sought a more heterogeneous data set by examining foreign-language documents from non-U.S. sources. We also obtained primary- source materials such as declassified intelligence reports from the Republic of Georgia, that helped to clarify and confirm the data found in secondary sources.

Since all these meetings were confidential, it was agreed in all cases that the information given was not for attribution by name.

For each of these studies, researchers traveled to the regions a number of times to collect information. Their work was combined with relevant secondary sources to produce detailed case studies presented later in the report. The format of the case studies followed the tenets outlined by Robert Yin, who proposes that case studies offer an advantage to researchers who present data illustrating complex relationships – such as the link between organized crime and terror.

 

2.4. Research goals

This project aimed to discover whether terrorist and organized crime groups would borrow one another’s methods, or cooperate, by what means, and how investigators and analysts could locate and assess crime-terror interactions. This led to an examination of why this overlap or interaction takes place. Are the benefits merely logistical or do both sides derive some long-term gains such as undermining the capacity of the state to detect and curtail their activities?

preparation of the investigative environment (PIE), by adapting a long-held military practice called intelligence preparation of the battlespace (IPB). The IPB method anticipates enemy locations and movements in order to obtain the best position for a commander’s limited battlefield resources and troops. The goal of PIE is similar to that of IPB—to provide investigators and analysts a strategic and discursive analytical method to identify areas ripe for locating terror and crime interactions, confirm their existence and then assess the ramifications of these collaborations. The PIE approach provides twelve watch points within which investigators and analysts can identify those areas most likely to contain crime-terror interactions.

The PIE methodology was designed with the investigator and analyst in mind, and thus PIE demonstrates how to establish investigations in a way that expend resources most fruitfully. The PIE methodology shows how insights can be gained from analysts to help practitioners identify problems and organize their investigations more effectively.

2.5. Research challenges

Our first challenge in investigating the links between organized crime and terrorism was to obtain enough data to provide an accurate portrayal of that relationship. Given the secrecy of all criminal organizations, many traditional methods of quantitative and qualitative research were not viable. Nonetheless we con- ducted numerous interviews, and obtained identified statements from investigators and policy officials. Records of legal proceedings, criminal records, and terrorist incident reports were also important data sources.

The strategy underlying the collection of data was to focus on the sources of interaction wherever they were located (e.g., developing countries and urban areas), rather than on instances of interaction in developed countries like the September 11th or the Madrid bombing investigations. In so doing, the project team hoped to avoid characterizing the problem “from out there.”

 

All three case studies high- light patterns of association that are particularly visible, frequent, and of lengthy duration. Because the conflict regions in the case studies also contribute to crime in the United States, our view was these models were needed to perceive patterns of association that are less visible in other environments. A further element in the selection of these regions was practical: in each one, researchers affiliated with the project had access to reliable sources with first-hand knowledge of the subject matter. Our hypothesis was that some of the most easy to detect relations would be in these societies that are so corrupted and with such limited enforcement that the phenomena might be more open for analysis and disclosure than in environments where this is more covert.

3. A new analytical approach: PIE

Investigators seeking to detect a terrorist activity before an incident takes place are overwhelmed by data.

A counterterrorist analyst at the Central Intelligence Agency took this further, noting that the discovery of crime-terror interactions was often the accidental result of analysis on a specific terror group, and thus rarely was connected to the criminal patterns of other terror groups.

IPB is an attractive basis for analyzing the behavior of criminal and terrorist groups because it focuses on evidence about their operational behavior as well as the environment in which they operate. This evidence is plentiful: communications, financial transactions, organizational forms and behavioral patterns can all be analyzed using a form of IPB.

the project team has devised a methodology based on IPB, which we have termed preparation of the investigation environment, or PIE. We define PIE as a concept in which investigators and analysts organize existing data to identify areas of high potential for collaboration between terrorists and organized criminals in order to focus next on developing specific cases of crime-terror interaction—thereby generating further intelligence for the development of early warning on planned terrorist activity.

While IPB is chiefly a method of eliminating data that is not likely to be relevant, our PIE method also provides positive indicators about where relevant evidence should be sought.

3.1 The theoretical basis for the PIE Method

Donald Cressey’s famous study of organized crime in the U.S., with the analogy of an archeological dig, was the starting point for our model of crime-terror cooperation.35 As Cressey defines it, archeologists first examine documentary sources to collect what is known and develop a map based on what is known. That map allows the investigator to focus on those areas that are not known—that is, the archeologist uses the map to focus on where to dig. The map also serves as a context within which artifacts discovered during the dig can be evaluated for their significance. For example, discovery of a bowl at a certain depth and location can provide information to the investigator concerning the date of an encampment and who established it.

The U.S. Department of Defense defines IPB as an analytical methodology employed to reduce un- certainties concerning the enemy, environment, and terrain for all types of operations. Intelligence preparation of the battlespace builds an extensive database for each potential area in which a unit may be re- quired to operate. The database is then analyzed in detail to determine the impact of the enemy, environment, and terrain on operations and presents it in graphic form.36 Alongside Cressey’s approach, IPB was selected as a second basis of our methodological approach.

Territory outside the control of the central state such as exists in failed or failing states, poorly regulated or border regions (especially those regions surrounding the intersection of multiple borders), and parts of otherwise viable states where law and order is absent or compromised, including urban quarters populated by diaspora communities or penal institutions, are favored locales for crime-terror interactions.

3.2 Implementing PIE as an investigative tool

Organized crime and terrorist groups have significant differences in their organizational form, culture, and goals. Bruce Hoffman notes that terrorist organizations can be further categorized based on their organizational ideology.

In converting IPB to PIE, we defined a series of watch points based on organizational form, goals, culture and other aspects to ensure PIE is flexible enough to compare a transnational criminal syndicate or a traditional crime hierarchy with an ethno-nationalist terrorist faction or an apocalyptic terror group.

The standard operating procedures and means by which military units are expected to achieve their battle plan are called doctrine, which is normally spelled out in great detail as manuals and training regimens. The doctrine of an opposing force thus is an important part of an IPB analysis. Such information is equally important to PIE, but is rarely found in manuals nor is it as highly developed as military doctrines.

Once the organizational forms, terrain and behavior of criminal and terrorist groups were defined at this level of detail, we settled on 12 watch points to cover the three components of PIE. For example, the watch point entitled organizational goals examines what the goals of organized crime and terror groups can tell investigators about potential collaboration or overlap between the two.

Investigators using PIE will collect evidence systematically through the investigation of watch points and analyze the data through its application to one or more indicators. That in turn will enable them to build a case for making timely predictions about crime-terror cooperation or overlap. Conversely, PIE also provides a mechanism for ruling out such links.

The indicators are designed to reduce the fundamental uncertainty associated with seemingly disparate or unrelated pieces of information. They also serve as a way of constructing probable cause, with evidence triggering indicators.

Although some watch points may generate ambiguous indicators of interaction between terror and crime, providing investigators and analysts with negative evidence of collusion between criminals and terrorists also has the practical benefit of steering scarce resources toward higher pay-off areas for detecting cooperation between the groups.

3.3. PIE composition: Watch points and indicators

The first step for PIE is to identify those areas where terror-crime collaborations are most likely to occur. To prepare this environment, PIE asks investigators and analysts to engage in three preliminary analyses. These are first to map where particular criminal and terrorist groups are likely to be operating, both in physical geographic terms and through information traditional and electronic media; secondly, to develop typologies for the behavior patterns of the groups and, when possible, their broader networks (often represented chronologically as a timeline); thirdly, to detail the organizations of specific crime and terror groups and, as feasible, their networks.

The geographical areas where terrorists and criminals are highly likely to be cooperating are known in IPB parlance as named areas of interest, or localities that are highly likely to support military operations. In PIE they are referred to as watch points.

A critical function of PIE is to set sensible priorities for analysts.

The second step of a PIE analysis concentrates on the watch points to identify named areas of inter- action where overlaps between crime and terror groups are most likely. The PIE method expresses areas of interest geographically but remains focused on the overlap between terrorism and organized crime.

the three preliminary analyses mentioned above are deconstructed into watch points, which are broad categories of potential crime-terror interactions.

the use of PIE leads to the early detection of named areas of interest through the analysis of watch points, providing investigators the means of concentrating their focus on terror-crime interactions and thereby enhancing their ability to detect possible terrorist planning.

The third and final step is for the collection and analysis of information that indicates organizational, operational or other nodes whereby criminals and terrorists appear to interact. While watch points are broad categories, they are composed of specific indicators of how organized criminals and terrorists might cooperate. These specific patterns of behavior help to confirm or deny that a watch point is applicable.

If several indicators are present, or if the indicators are particularly clear, this bolsters the evidence that a particular type of terror-crime interaction is present. No single indicator is likely to provide ‘smoking gun’ evidence of a link, although examples of this have occasionally arisen. Instead, PIE is a holistic approach that collects evidence systematically in order to make timely predictions of an affiliation, or not, between specific criminal and terrorist groups.

For policy analysts and planners, indicators reduce the sampling risk that is unavoidable for anyone collecting seemingly disparate and unrelated pieces of evidence. For investigators, indicators serve as a means of constructing probable cause. Indeed, even negative evidence of interaction has the practical benefit of helping investigators and analysts manage their scarce resources more efficiently.

3.4 The PIE approach in practice: Two Cases

the process began with the collection of relevant information (scanning) that was then placed into the larger context of watch points and indicators (codification) in order to produce the aforementioned analytical insights (abstraction).

 

Each case will describe how the TraCCC team shared (diffusion) its findings in or- der to obtain validation and to have an impact on practitioners fighting terrorism and/or organized crime.

3.4.1 The Georgia Case

In 2003-4, TraCCC used the PIE approach to identify one of the largest money laundering cases ever successfully prosecuted. The PIE method helped close down a major international vehicle for money laundering. The ability to organize the financial records from a major money launderer allowed the construction of a significant network that allowed understanding of the linkages among major criminal groups whose relationship has not previously been acknowledged.

Some of the information most pertinent to Georgia included but that was not limited to:

1. Corrupt Georgian officials held high law enforcement positions prior to the Rose Revolution and maintained ties to crime and terror groups that allowed them to operate with impunity;
2. Similar patterns of violence were found among organized crime and terrorist groups operating in Georgia;
3. Numerous banks, corrupt officials and other providers of illicit goods and services assisted both organized crime and terrorists
4. Regions of the country supported criminal infrastructures useful to organized crime and terrorists alike, including Abkhazia, Ajaria and Ossetia.

Combined with numerous other pieces of information and placed into the PIE watch point structure, the resulting analysis triggered a sufficient number of indicators to suggest that further analysis was warranted to try to locate a crime-terror interaction.

 

The second step of the PIE analysis was to examine information within the watch points for connections that would suggest patterns of interaction between specific crime and terror groups. These points of interaction are identified in the Black Sea case study but the most successful identification was found from an analysis of the watch point that specifically examined the financial environment that would facilitate the link between crime and terrorism.

The TraCCC team began its investigation within this watch point by identifying the sectors of the Georgian economy that were most conducive to economic crime and money laundering. This included such sectors as energy, railroads and banking. All of these sectors were found to be highly criminalized.

Only by having researchers with knowledge of the economic climate, the nature of the business community and the banking sector determined that investigative resources needed to be concentrated on the “G” bank. By knowing the terrain, investigative focus was focused on “G” bank by the newly established financial investigative unit of the Central Bank. A six-month analysis of the G bank and its transactions enabled the development of a massive network analysis that facilitated prosecution in Georgia and may lead to prosecutions in major financial centers that were previously unable to address some crime groups, at least one of which was linked to a terrorist group.

Using PIE allowed a major intelligence breakthrough.

First, it located a large facilitator of dirty money. Second, the approach was able to map fundamental connections between crime and terror groups. Third, the analysis highlighted the enormous role that purely “dirty banks” housed in countries with small economies can provide as a service for transnational crime and even terrorism.

While specific details must remain sealed due to deference to ongoing legal proceedings, to date the PIE analysis has grown into investigations in Switzerland, and others in the US and Georgia.

the PIE approach is one that favors the construction and prosecution of viable cases.

the PIE approach is a platform for starting and later focusing investigations. When coupled with investigative techniques like network analysis, the PIE approach supports the construction and eventual prosecution of cases against organized crime and terrorist suspects.

3.4.2 Russian Closed Cities

In early 2005, a US government agency asked TraCCC to identify how terrorists are potentially trying to take advantage of organized crime groups and corruption to obtain fissile material in a specific region of Russia—one that is home to a number of sensitive weapons facilities located in so-called “closed cities.” The project team assembled a wealth of information concerning the presence and activities of both criminal and terror groups in the region in question, but was left with the question of how best to organize the data and develop significant conclusions.

The project’s information supported connections in 11 watch points, including:

• A vast increase in the prevalence of violence in the region, especially in economic sectors with close ties to organized crime;
• Commercial ties in the drug trade between crime groups in the region and Islamic terror groups formerly located in Afghanistan;
• Rampant corruption in all levels of the regional government and law enforcement mechanisms, rendering portions of the region nearly ungovernable;
• The presence of numerous regional and transnational crime groups as well as recruiters for Islamic groups on terrorist watch lists;

employment of the watch points prompted creative leads to important connections that were not readily apparent until placed into the larger context of the PIE analytical framework. Specifically, the analysis might not have included evidence of trust links and cultural ties between crime and terror groups had the PIE approach not explained their utility.

When the TraCCC team applied the PIE to the closed cities case, the team found using the technologies reduced time analyzing data while improving the analytical rigor of the task. For example, structured queries of databases and online search engines provided information quickly. Likewise, network mapping improved analytical rigor by codifying the links between numerous actors (e.g., crime groups, terror groups, workers at weapons facilities and corrupt officials) in local, regional and transnational contexts.

3.5 Emergent behavior and automation

The dynamic nature of crime and terror groups complicates the IPB to PIE transition. The spectrum of cooperation demonstrates that crime-terror intersections are emergent phenomena.

PIE must have feedback loops to cope with the emergent behavior of crime and terror groups

when the project team spoke with analysts and investigators, the one deficiency they noted was the ability to conduct strategic intelligence given their operational tempo.

4. The terror-crime interaction spectrum

In formulating PIE, we recognized that crime and terrorist groups are more diverse in nature than military units. They may be networks or hierarchies, they have a variety of cultures rather than a disciplined code of behavior, and their goals are far less clear. Hoffman notes that terrorist groups can be further categorized based on their organizational ideology.

Other researchers have found significant evidence of interaction between terrorism and organized crime, often in support of the general observation that while their methods might converge, the basic motives of crime and terror groups would serve to keep them at arm’s length—thus the term “methods, not motives.”41 Indeed, the differences between the two are plentiful: terrorists pursue political or religious objectives through overt violence against civilians and military targets. They turn to crime for the money they need to survive and operate.

Criminal groups, on the other hand, are focused on making money. Any use of violence tends to be concealed, and is generally focused on tactical goals such as intimidating witnesses, eliminating competitors or obstructing investigators.

In a corrupt environment, the two groups find common cause.

Terrorists often find it expedient, even necessary, to deal with outsiders to get funding and logistical support for their operations. As such interactions are repeated over time, concerns arise that criminal and terrorist organizations will integrate and might even form new types of organizations.

Support for this point can be found in the seminal work of Sutherland, who has argued that the “in- tensity and duration” of an association with criminals makes an individual more likely to adopt criminal behavior. In conflict regions, where there is intensive interaction between criminals and terrorists, there is more shared behavior and a process of mutual learning that goes on.

The dynamic relationship between international terror and transnational crime has important strategic implications for the United States.

The result is a model known as the terror-crime interaction spectrum that depicts the relationship between terror and criminal groups and the different forms it takes.

Each form of interaction represents different, yet specific, threats, as well as opportunities for detection by law enforcement and intelligence agencies.

An interview with a retired member of the Chicago organized crime investigative unit revealed that it had investigated taxi companies and taxicab owners as cash-based money launderers. Logic suggests that terrorists may also be benefiting from the scheme. But this line of investigation was not pursued in the 9/11 investigations although two of the hijackers had worked as taxi drivers.

Within the spectrum, processes we refer to as activity appropriation, nexus, symbiotic relationship, hybrid, and transformation illustrate the different forms of interaction between a terrorist group and an organized crime group, as well as the behavior of a single group engaged in both terrorism and organized crime.

While activity appropriation does not represent organizational linkages between crime and terror groups, it does capture the merger of methods that were well-documented in section 2. Activity appropriation is one way that terrorists are exposed to organized crime activities and, as Chris Dishman has noted, can lead to a transformation of terror cells into organized crime groups.

Applying the Sutherland principle of differential association, these activities are likely to bring a terror group into regular contact with organized crime. As they attempt to acquire forged documents, launder money, or pay bribes, it is a natural step to draw on the support and expertise of the criminal group, which is likely to have more experience in these activities. It is referred to here as a nexus.

terrorists first engage in “do it yourself” organized crime and then turn to organized crime groups for specialized services like document forgery or money laundering.

In most cases a nexus involves the criminals providing goods and services to terrorists for payment although it can work in both directions. A typically short-term relation- ship, a nexus does not imply that the criminals share the ideological views of the terrorists, merely that the transaction offers benefits to both sides.

After all, they have many needs in common: safe havens, false documentation, evasive tactics, and other strategies to lower the risk of being detected. In Latin America, transnational criminal gangs have employed terrorist groups to guard their drug processing plants. In Northern Ireland, terrorists have provided protection for human smuggling operations by the Chinese Triads.

If the nexus continues to benefit both sides over a period of time, the relationship will deepen. More members of both groups will cooperate, and the groups will create structures and procedures for their business transactions, transfer skills and/or share best practices. We refer to this closer, more sustained cooperation as a symbiotic relationship, and define it as a relationship of mutual benefit or dependence.

In the next stage, the two groups continue to cooperate over a long period and members of the organized crime group begin to share the ideological goals of the terrorists. They grow increasingly alike and finally they merge. That process results in a hybrid or dark network49 that has been memorably described as terrorist by day and criminal by night.50 Such an organization engages in criminal acts but also has a political agenda. Both the criminal and political ends are forwarded by the use of violence and corruption.

These developments are not inevitable, but result from a series of opportunities that can lead to the next stage of cooperation. It is important to recognize, however, that even once the two groups have reached the point of hybrid, there is no reason per se to suspect that transformation will follow. Likewise, a group may persist with borrowed methods indefinitely without ever progressing to cooperation. In Italy and elsewhere, crime groups that also engaged in terrorism never found a terrorist partner and thus remained at the activity appropriation stage. Eventually they ended their terrorist activities and returned to the exclusive pursuit of organized crime.

Interestingly, the TraCCC team found no example where a terrorist group engaging in organized crime, either through activity appropriation or through an organizational linkage, came into conflict with a criminal group.51 Neither archival sources nor our interviews revealed such a conflict over “turf,” though logic would suggest that organized crime groups would react to such forms of competition.

The spectrum does not create exact models of the evolution of criminal-terrorist cooperation. In- deed, the evidence presented both here and in prior studies suggests that a single evolutionary path for crime-terror interactions does not exist. Environmental factors outside the control of either organization and the varied requirements of specific organized crime or terrorist groups are but two of the reasons that interactions appear more idiosyncratic than generalizable.

Using the PIE method, investigators and analysts can gain an understanding of the terror-crime intersection by analyzing evidence sourced from communications, financial transactions, organizational charts, and behavior. They can also apply the methodology to analyze watch points where the two entities may interact. Finally, using physical, electronic, and data surveillance, they can develop indicators showing where watch points translate into practice.

5. The significance of terror-crime interactions in geographic terms

Some shared characteristics arose from examining this case. First, both neighborhoods shared similar diaspora compositions and a lack of effective or interested policing. Second, both terror cells had strong connections to the shadow economy.

the case demonstrated that each cell shared three factors—poor governance, a sense of ethnic separation amongst the cell (supported by the nature of the larger diaspora neighborhoods), and a tradition of organized crime.

U.S. intelligence and law enforcement are naturally inclined to focus on manifestations of organized crime and terrorism in their own country, but they would benefit from studying and assessing patterns and behavior of crime in other countries as well as areas of potential relevance to terrorism.

When turning to the situation overseas, one can differentiate between longstanding crime groups and their more recently formed counterparts according to their relationship to the state. With the exception of Colombia, rarely do large, established (i.e., “traditional”) crime organizations link with terrorists. These groups possess long-held financial interests that would suffer should the structures of the state and the international financial community come to be undermined. Through corruption and movement into the lawful economy, these groups minimize the risk of prosecution and therefore do not fear the power of state institutions.

Developing countries with weak economies, a lack of social structures, many desperate, hungry people, and a history of unstable government are both relatively likely to provide ideological and economic foundations for both organized crime and terrorism within their borders and relatively unlikely to have much capacity to combat either of them. Conflict zones have traditionally provided tremendous opportunities for smuggling and corruption and reduced oversight capacities, as regulatory and enforcements be- come almost solely directed at military targets. They are therefore especially vulnerable to both serious organized crime and violent activity directed at civilian populations for political goals – as well as cooperation between those engaging in pure criminal activities and those engaging in politically-motivated violence.

Post-conflict zones are also likely to spawn such cooperation; as such areas often retain weak enforcement capacity for some time following an end to formal hostilities.

these patterns of criminal behavior and organization can arise from areas as diverse as conflict zones overseas (which then tend can replicate once they arrive in the U.S.) to neighborhoods in U.S. cities. The problematic combinations of poor governance, ethnic separation from larger society, and a tradition of criminal activity (frequently international) are the primary concerns behind this broad taxonomy of geographic locales for crime-terror interaction.

6. Watch points and indicators

Taking the evidence of cooperation between organized crime and terrorism, we have generated 12 specific areas of interaction, which we refer to as watch points. In turn these watch points are subdivided into a number of indicators that point out where interaction between terror and crime may be taking place.

These watch points cover a variety of habits and operating modes of organized crime and terrorist groups.

We have organized our watch points into three categories: environmental, organizational, and behavioral. Each of the following sections details one of the twelve watch points.

 

Watch Point 1: Open activities in the legitimate economy

Watch Point 2: Shared illicit nodes

Watch Point 3: Communications

Watch Point 4: Use of information technology (IT)

Watch Point 5: Violence

Watch Point 6: Use of corruption

Watch Point 7: Financial transactions & money laundering

Watch Point 8: Organizational structures

Watch Point 9: Organizational goals

Watch Point 10: Culture

Watch Point 11: Popular support

Watch Point 12: Trust

 

6.1. Watch Point 1: Open activities in the legitimate economy

The many indicators of possible links include habits of travel, the use of mail and courier services, and the operation of fronts.

Organized crime and terror may be associated with subterfuge and secrecy, but both criminal types engage legitimate society quite openly for particular political purposes. Yet in the first instance, criminal groups are likely to leave greater “traces,” especially when they operate in societies with functioning governments, than do terrorist groups.

Terrorist groups usually seek to make common cause with segments of society that will support their goals, particularly the very poor and the disadvantaged. Terrorists usually champion repressed or dis- enfranchised ethnic and religious minorities, describing their terrorist activities as mechanisms to pressure the government for greater autonomy and freedom, even independence, for these minorities… the openly take responsibility for their attacks, but their operational mechanisms are generally kept secret, and any ongoing contacts they may have with legitimate organizations are carefully hidden.

Criminal groups, like terrorists, may have political goals. For example, such groups may seek to strengthen their legitimacy through donating some of their profits to charity. Colombian drug traffickers are generous in their support of schools and local sports teams.5

criminals of all types could scarcely carry out criminal activities, maintain their cover, and manage their money flows without doing legal transactions with legitimate businesses.

Travel: Frequent use of passenger carriers and shipping companies are potential indicators of illicit activity. Clues can be gleaned from almost any pattern of travel that can be identified as such.

Mail and courier services: Indicators of interaction are present in the tracking information on international shipments of goods, which also generate customs records. Large shipments require bills-of-lading and other documentation. Analysis of such transactions, cross-referenced with in- formation on crime databases, can identify links between organized crime and terrorist groups.

Fronts: A shared front company or mutual connections to legitimate businesses are clearly also indicators of interaction.

Watch Point 2: Shared illicit nodes

 

The significance of overt operations by criminal groups should not be overstated. Transnational crime and terror groups alike carry out their operations for the most part with illegal and undercover methods. There are many similarities in these tactics. Both organized criminals and terrorists need forged pass- ports, driver’s licenses, and other fraudulent documents. Dishonest accountants and bankers help criminals launder money and commit fraud. Arms and explosives, training camps and safe houses are other goods and services that terrorists obtain illicitly.

Fraudulent Documents. Groups of both types may use the same sources of false documents,

or the same techniques, indicating cooperation or overlap. A criminal group often develops an expertise in false document production as a business, expanding production and building a customer base.

 

Some of the 9/11 hijackers fraudulently obtained legitimate driver’s licenses through a fraud ring based at an office of DMV in the Virginia suburbs of Washington, DC. Ac- cording to an INS investigator, this ring was under investigation well before the 9/11 attacks, but there was insufficient political will inside the INS to take the case further.

Arms Suppliers. Both terror and organized crime might use the same supplier, or the same distinctive method of doing business, such as bartering weapons or drugs. In 2001 the Basque terror group ETA contracted with factions of the Italian Camorra to obtain missile launchers and ammunition in return for narcotics.

Financial experts. Bankers and financial professionals who assist organized crime might also have terrorist affiliations. The methods of money laundering long used by narcotics traffickers and other organized crime have now been adopted by some terrorist groups.

 

Drug Traffickers. Drug trafficking is the single largest source of revenues for international organized crime. Substantial criminal groups often maintain well-established smuggling routes to distribute drugs. Such an infrastructure would be valuable to terrorists who purchased weapons of mass destruction and needed to transport them.

 

Other Criminal Enterprises. An increasing number of criminal enterprises outside of narcotics smuggling are serving the financial or logistical ends of terror groups and thus serve as nodes of interaction. For example, piracy on the high seas, a growing threat to maritime commerce, often depends on the collusion of port authorities, which are controlled in many cases by organized crime.

These relationships are particularly true of developed countries with effective law enforcement, since criminals obviously need to be more cautious and often restrict their operations to covert activity. In conflict zones, however, criminals of all types feel even less restraint about flaunting their illegal nature, since there is little chance of being detected or apprehended.

Watch Point 3: Communications

 

The Internet, mobile phones and satellite communications enable criminals and terrorists to communicate globally in a relatively secure fashion. FARC, in concert with Colombian drug cartels, offered training on how to set up narcotics trafficking businesses used secure websites and email to handle registration.

Such scenarios are neither hypothetical nor anecdotal. Interviews with an analyst at the US Drug Enforcement Administration revealed that narcotics cartels were increasingly using encryption in their digital communications. In turn, the agent interviewed stated that the same groups were frequently turning to information technology experts to provide them encryption to help secure their communications.

Nodes of interaction therefore include:

• Technical overlap: Examples exist where organized crime groups opened their illegal communications systems to any paying customer, thus providing a service to other criminals and terrorists among others. For example, a recent investigation found clandestine telephone exchanges in the Tri-Border region of South America that were connected to Jihadist networks. Most were located in Brazil, since calls between Middle Eastern countries and Brazil would elicit less suspicion and thus less chance of electronic eavesdropping.
• Personnel overlap: Crime and terror groups that recruit common high-tech specialists to their cause. Given their ability to encrypt messages, criminals of all kinds may rely on outsiders to carry the message. Smuggling networks all have operatives who can act as couriers, and terrorists have networks of sympathizers in ethnic diasporas who can also help.

Watch Point 4: Use of information technology (IT)

 

Organized crime has devised IT-based fraud schemes such as online gambling, securities fraud, and pirating of intellectual property. Such schemes appeal to terror groups, too, particularly given the relative anonymity that digital transactions offer. Investigators into the Bali disco bombing of 2002 found that the laptop computer of the ringleader, Imam Samudra, contained a primer he authored on how to use online fraud to finance operations. Evidence of terror groups’ involvement is a significant set of indicators of cooperation or overlap.

Indicators of possible cooperation or nodes of interaction include:

Fundraising: Online fraud schemes and other uses of IT for obtaining ill-gotten gains are already well-established by organized crime groups and terrorists are following suit. Such IT- assisted criminal activities serve as another node of overlap for crime and terror groups, and thus expand the area of observation beyond the brick-and-mortar realm into cyberspace (i.e., investigators now expect to find evidence of collaboration on the Internet or in email as much as through telephone calls or postal services).

• Use of technical experts: While no evidence exists that criminals and terrorists have directly cooperated to conduct cybercrime or cyberterrorism, they are often served by the same technical experts.

Watch Point 5: Violence

 

Violence is not so much a tactic of terrorists as their defining characteristic. These acts of violence are designed to obtain publicity for the cause, to create a climate of fear, or to provoke political repression, which they hope will undermine the legitimacy of the authorities. Terrorist attacks are deliberately highly visible in order to enhance their impact on the public consciousness. Indiscriminate violence against innocent civilians is therefore more readily ascribed to terrorism.

no examples exist where terrorists have engaged criminal groups for violent acts.

A more significant challenge lies in trying to discern generalities about organized crime’s patterns of violence. Categorizing patterns of violence according to their scope or their promulgation is suspect. In the past, crime groups have used violence selectively and quietly to achieve their goals, but then have also used violence broadly and loudly to achieve other goals. Neither can one categorize organized crime’s violence according to goals as social, political and economic considerations often overlap in every attack or campaign.

Violence is therefore an important watch point that may not yield specific indicators of crime-terror interaction per se but can serve to frame the likelihood that an area might support terror-crime interaction.

Watch Point 6: Use of corruption

 

Both terrorists and organized criminals bribe government officials to undermine the work of law enforcement and regulation. Corrupt officials assist criminals by exerting pressure on businesses that refuse to cooperate with organized crime groups, or by providing passports for terrorists. The methods of corruption are diverse on both sides and include payments, the provision of illegal goods, the use of compromising information to extort cooperation, and outright infiltration of a government agency or other target.

Many studies have demonstrated that organized crime groups often evolve in places where the state cannot guarantee law or order, or provide basic health care, education, and social services. The absence of effective law enforcement combines with rampant corruption to make well-organized criminals nearly invulnerable.

Colombia may be the only example of a conflict zone where a major transnational crime group with very large profits is directly and openly connected to terrorists. The interaction between the FARC and ELN terror groups and the drug syndicates provides crucial important financial resources for the guerillas to operate against the Colombian state – and against each another. This is facilitated by universal corruption, from top government officials to local police. Corruption has served as the foundation for the growth of the narcotics cartels and insurgent/terrorist groups.

In the search for indicators, it would be simplistic to look for a high level of corruption, particularly in conflict zones. Instead, we should pose a series of questions:

Cooperation Are terrorist and criminal groups working together to minimize cost and maximize leverage from corrupt individuals and institutions?

Division of labor Are terrorist and criminal groups purposefully corrupting the areas they have most contact with? In the case of crime groups, that would be law enforcement and the judiciary; in the case of terrorists, the intelligence and security services.

• Autonomy Are corruption campaigns carried out by one or both groups completely independent of the other?

These indicators can be applied to analyze a number of potential targets of corruption. Personnel that can provide protection or services are often mentioned as the target of corruption. Examples include law enforcement, the judiciary, border guards, politicians and elites, internal security agents and Consular officials. Economic aid and foreign direct investment are also targeted as sources of funds by criminals and terrorists that they can access by means of corruption.

 

Watch Point 7: Financial transactions & money laundering

 

despite the different purposes that may be involved in their respective uses of financial institutions (organized crime seeking to turn illicit funds into licit funds; terrorists seeking to move licit funds to use them for illicit means), the groups tend to share a common infrastructure for carrying out their financial activities. Both types of groups need reliable means of moving, and laundering money in many different jurisdictions, and as a result, both use similar methods to move money internationally. Both use charities and front groups as a cover for money flows.

Possible indicators include:

• Shared methods of money laundering
• Mutual use of known front companies and banks, as well as financial experts.

Watch Point 8: Organizational structures

 

The traditional model of organized crime used by U.S. law enforcement is that of the Sicilian Mafia – a hierarchical, conservative organization embedded in the traditional social structures of southern Italy… among today’s organized crime groups the Sicilian mafia is more of an exception than the rule.

Most organized crime now operates not as a hierarchy but as a decentralized, loose-knit network – which is a crucial similarity to terror groups. Networks offer better security, make intelligence-gathering more efficient, cover geographic distances and span diverse memberships more effectively.

Membership dynamics Both terror and organized crime groups – with the exception of the Sicilian Mafia and other traditional crime groups (i.e., Yakuza) – are made up of members with loose, relatively short-term affiliations to each other and even to the group itself. They can readily be recruited by other groups. By this route, criminals have become terrorists.

Scope of organization Terror groups need to make constant efforts to attract and recruit new members. Obvious attempts to attract individuals from crime groups are a clear indication of co- operation. An intercepted phone conversation in May 2004 by a suspected terrorist called Rabei Osman Sayed Ahmed revealed his recruitment tactics: “You should also know that I have met other brothers, that slowly I have created with a few things. First, they were drug pushers, criminals, I introduced them to the faith and now they are the first ones who ask when the moment of the jihad will be…”

Need to buy, wish to sell Often the business transactions between the two sides operate in both directions. Terrorist groups are not just customers for the services of organized crime, but often act as suppliers, too. Arms supply by terrorists is particularly marked in certain conflict zones. Thus, any criminal group found to be supplying outsiders with goods or services should be investigated for its client base too.

Investigators who discovered the money laundering in the above example were able to find out more about the terrorists’ activities too. The Islamic radical cell that planned the Madrid train bombings of 2004 was required to support itself financially through a business venture despite its initial funding by Al Qaeda.

Watch Point 9: Organizational goals

 

In theory, their different goals are what set terrorists apart from the perpetrators of organized crime. Terrorist groups are most often associated with political ends, such as change in leadership regimes or the establishment of an autonomous territory for a subnational group. Even millenarian and apocalyptic terrorist groups, such as the science-fiction mystics of Aum Shinrikyo, often include some political objectives. Organized crime, on the other hand, is almost always focused on personal enrichment.

By cataloging the different – and shifting – goals of terror and organized crime groups, we can develop indicators of convergence or divergence. This will help identify shared aspirations or areas where these aims might bring the two sides into conflict. On this basis, investigators can ask what conditions might prompt either side to adopt new goals or to fall back to basic goals, such as self-preservation.

Long view or short-termism

Affiliations of protagonists

 

Watch Point 10: Culture

 

Both terror and criminal groups use ideologies to maintain their internal identity and provide external justifications for their activities. Religious terror groups adopt and may alter the teachings of religious scholars to suggest divine support for their cause, while Italian, Chinese, Japanese, and other organized crime groups use religious and cultural themes to win public acceptance. Both types use ritual and tradition to construct and maintain their identity. Tattoos, songs, language, and codes of conduct are symbolic to both.

Religious affiliations, strong nationalist sentiments and strong roots in the local community are often characteristics that cause organized criminals to shun any affiliation with terrorists. Conversely, the absence of such affiliations means that criminals have fewer constraints keeping them from a link with terrorists.

In any organization, culture connects and strengthens ties between members. For networks, cultural features can also serve as a bridge to other networks.

• Religion Many criminal and terrorist groups feature religion prominently.
• Nationalism Ethno-nationalist insurgencies and criminal groups with deep historical roots are particularly likely to play the nationalist card.
• Society Many criminal and terrorist networks adapt cultural aspects of the local and regional societies in which they operate to include local tacit knowledge, as contained in narrative traditions. Manuel Castells notes the attachment of drug traffickers to their country, and to their regions of origin. “They were/are deeply rooted in their cultures, traditions, and regional societies. …they have also revived local cultures, rebuilt rural life, strongly affirmed their religious feeling, and their beliefs in local saints and miracles, supported musical folklore (and were rewarded with laudatory songs from Colombian bards)…”

Watch Point 11: Popular support

 

Both organized crime and terrorist groups engage legitimate society in furtherance of their own agendas. In conflict zones, this may be done quite openly, while under the rule of law they are obliged to do so covertly. One way of doing so is to pay lip service to the interests of certain ethnic groups or social classes. Organized crime is particularly likely to make an appeal to disadvantaged people or people in certain professionals though paternalistic actions that make them a surrogate for the state. For instance, the Japanese Yakuza crime groups provided much-needed assistance to the citizens of Kobe after the serious earthquake there. Russian organized crime habitually supports cultural groups and sports troupes.

 

Both crime and terror derive crucial power and prestige through the support of their members and of some segment of the public at large. This may reflect enlightened self-interest, when people see that the criminals are acting on their behalf and improving their well-being and personal security. But it is equally likely to be that people are afraid to resist a violent criminal group in their neighborhood

This quest for popular support and common cause suggests various indicators:

• Sources Terror groups seek and sometimes obtain the assistance of organized crime based on the perceived worthiness of the terrorist cause, or because of their common cause against state authorities or other sources of opposition. In testimony before the U.S. House Committee on International Relations, Interpol Secretary General Ronald Noble made this point. One of his examples was that Lebanese syndicates in South America send funds to Hezbollah.
• Means Groups that cooperate may have shared activities for gaining popular support such as political parties, labor movements, and the provision of social services.
• Places In conflict zones where the government has lost authority to criminal groups, social welfare and public order might be maintained by the criminal groups that hold power.

 

Watch Point 12: Trust

Like business corporations, terrorist and organized crime groups must attract and retain talented, dedicated, and loyal personnel. These skills are at an even greater premium than in the legitimate economy because criminals cannot recruit openly. A further challenge is that law enforcement and intelligence services are constantly trying to infiltrate and dismantle criminal networks. Members’ allegiance to any such group is constantly tested and demonstrated through rituals such as the initiation rites…

We propose three forms of trust in this context, using as a basis Newell and Swan’s model for inter- personal trust within commercial and academic groups.94

Companion trust based on goodwill or personal friendships… In this context, indicators of terror-crime interaction would be when members of the two groups use personal bonds based on family, tribe, and religion to cement their working relationship. Efforts to recruit known associates of the other group, or in common recruiting pools such as diasporas, would be another indicator.

Competence trust, which Newell and Swan define as the degree to which one person depends upon another to perform the expected task.

Commitment or contract trust, where all actors understand the practical importance of their role in completing the task at hand.

7. Case studies

7.1. The Tri-Border Area of Paraguay, Brazil, and Argentina

Chinese Triads such as the Fuk Ching, Big Circle Boys, and Flying Dragons are well established and believed to be the main force behind organized crime in CDE.

CDE is also a center of operations for several terrorist groups, including Al Qaeda, Hezbollah, Islamic Jihad, Gamaa Islamiya, and FARC.

Watch points

Crime and terrorism in the Tri-Border Area interact seamlessly, making it difficult to draw a clean line be- tween the types of persons and groups involved in each of these two activities. There is no doubt, however, that the social and economic conditions allow groups that are originally criminal in nature and groups whose primary purpose is terrorism to function and interact freely.

Organizational structure

Evidence from CDE suggests that some of the local structures used by both groups are highly likely to overlap. There is no indication, however, of any significant organizational overlap between the criminal and terrorist groups. Their cooperation, when it exists, is ad hoc and without any formal or lasting agreements, i.e., activity appropriation and nexus forms only.

Organizational goals

In this region, the short-term goals of criminals and terrorists converge. Both benefit from easy border crossings and the networks necessary to raise funds.

Culture Cultural affinities between criminal and terrorist groups in the Tri-Border Area include shared ethnicities, languages and religions.

It emerged that 400 to 1000 kilograms of cocaine may have been shipped on a monthly basis through the Tri-Border Area on its way to Sao Paulo and thence to the Middle East and Europe

Numerous arrests revealed the strong ties between entrepreneurs in CDE and criminal and potentially terrorist groups. From the evidence in CDE it seems that the two phenomena operate in rather separate cultural realities, focusing their operations within ethnic groups. But nor does culture serve as a major hindrance to cooperation between organized crime and terrorists.

Illicit activities and subterfuge

The evidence in CDE suggests that terrorists see it as logical and cost-effective to use the skills, contacts, communications and smuggling routes of established criminal networks rather than trying to gain the requisite experience and knowledge themselves. Likewise, terrorists appear to recognize that to strike out on their own risks potential turf conflicts with criminal groups.

There is a clear link between Hong Kong-based criminal groups that specialize in large-scale trafficking of counterfeit products such as music albums and software, and the Hezbollah cells active in the Tri-Border Area. Within their supplier-customer relationship, the Hong Kong crime groups smuggle contraband goods into the region and deliver them to Hezbollah operatives, who in turn profit from their sale. The proceeds are then used to fund the terrorist groups.

Open activities in the legitimate economy

The knowledge and skills potential of CDE is tremendous. While no specific examples exist to connect terrorist and criminal groups through the purchase of legal goods and services, it is obvious that the likelihood of this is high, given how the CDE economy is saturated with organized crime.

Support or sustaining activities

The Tri-Border Area has an usually large and efficient transport infrastructure, which naturally assists organized crime. In turn, the many criminals and terrorists using cover require a sophisticated and reliable document forgery industry. The ease with which these documents can be obtained in CDE is an indicator of cooperation between terrorists and criminals.

Brazilian intelligence services have evidence that Osama bin Laden visited CDE in 1995 and met with the members of the Arab community in the city’s mosque to talk about his experience as a mujahadeen fighter in the Afghan war against the Soviet Union.

Use of violence

Contract murder in CDE costs as little as one thousand dollars, and the frequent violence in CDE is directed at business people who refuse to bend to extortion by terror groups. Ussein Mohamed Taiyen, president of the CDE Chamber of Commerce, was one such victim—murdered because he refused to pay the tax.

Financial transactions and money laundering in 2000, money laundering in the Tri-Border Area was estimated at 12 billion U.S. dollars annually.

As many as 261 million U.S. dollars annually has been raised in Tri-Border Area and sent overseas to fund the terrorist activities of Hezbollah, Hamas, and Islamic Jihad.

Use of corruption

Most of the illegal activities in the Tri-Border Area bear the hallmark of corruption. In combination with the generally low effectiveness of state institutions, especially in Paraguay, and high level of corruption in that country, CDE appears to be a perfect environment for the logistical operations of both terrorists and organized criminals.

Even the few bona fide anti-corruption attempts made by the Paraguayan government have been under- mined because of the pervasive corruption, another example being the attempts to crack down on the Chinese criminal groups in CDE. The Consul General of Taiwan in CDE, Jorge Ho, stated that the Chinese groups were successful in bribing Paraguayan judges, effectively neutralizing law enforcement moves against the criminals.122

The other watch points described earlier – including fund raising and use of information technology – can also be illustrated with similar indicators of possible cooperation between terror and organized crime.

In sum, for the investigator or analyst seeking examples of perfect conditions for such cooperation, the Tri-Border Area is an obvious choice.

7.2. Crime and terrorism in the Black Sea region

Illicit or veiled operations Cigarette, drugs and arms smuggling have been major sources of financing of all the terrorist groups in the region.

Cigarette and alcohol smuggling has fueled the Kurdish-Turkish conflict as well as the terrorist violence in both the Abkhaz and Ossetian conflicts.

From the very beginning, the Chechen separatist movement had close ties with the Chechen crime rings in Russia, mainly operating in Moscow. These crime groups provided and some of them still provide financial sup- port for the insurgents.

8. Conclusion and recommendations

The many examples in this report of cooperation between terrorism and organized crime make clear that the links between these two potent threats to national and global security are widespread, dynamic, and dangerous. It is only rational to consider the possibility that an effective organized crime group may have a connection with terrorists that has gone unnoticed so far.

Our key conclusion is that crime is not a peripheral issue when it comes to investigating possible terrorist activity. Efforts to analyze the phenomenon of terrorism without considering the crime component undermine all counter-terrorist activities, including those aimed at protecting sites containing weapons of mass destruction.

Yet the staffs of intelligence and law enforcement agencies in the United States are already over- whelmed. Their common complaint is that they do not have the time to analyze the evidence they possess, or to eliminate unnecessary avenues of investigation. The problem is not so much a dearth of data, but the lack of suitable tools to evaluate that data and make optimal decisions about when, and how, to investigate further.

Scrutiny and analysis of the interaction between terrorism and organized crime will become a matter of routine best practice. Aware- ness of the different forms this interaction takes, and the dynamic relationship between them, will become the basis for crime investigations, particularly for terrorism cases.

In conclusion, our overarching recommendation is that crime analysis must be central to understanding the patterns of terrorist behavior and cannot be viewed as a peripheral issue.

For policy analysts:

1. More detailed analysis of the operation of illicit economies where criminals and terrorists interact would improve understanding of how organized crime operates, and how it cooperates with terrorists. Domestically, more detailed analysis of the businesses where illicit transactions are most common would help investigation of organized crime – and its affiliations. More focus on the illicit activities within closed ethnic communities in urban centers and in prisons in developed countries would prove useful in addressing potential threats.
2. Corruption overseas, which is so often linked to facilitating organized crime and terrorism, should be elevated to a U.S. national security concern with an operational focus. After all, many jihadists are recruited because they are disgusted with the corrupt governments in their home countries. Corruption has facilitated the commission of criminal acts such as the Chechen suicide bombers who bribed airport personnel to board aircraft in Moscow.
3. Analysts must study patterns of organized crime-terrorism interaction as guidance for what maybe observed subsequently in the United States.
4. Intelligence and law enforcement agencies need more analysts with the expertise to understand the motivations and methods of criminal and terrorist groups around the globe, and with the linguistic and other skills to collect and analyze sufficient data.

For investigators:

1. The separation of criminals and terrorists is not always as clear cut as many investigators believe. Crime and terrorists’ groups are often indistinguishable in conflict zones and in prisons.
2. The hierarchical structure and conservative habits of the Sicilian Mafia no longer serves as an appropriate model for organized crime investigations. Most organized crime groups now operate as loose networked affiliations. In this respect they have more in common with terrorist groups.
3. The PIE method provides a series of indicators that can result in superior profiles and higher- quality risk analysis for law enforcement agencies both in the United States and abroad. The approach can be refined with sensitive or classified information.
4. Greater cooperation between the military and the FBI would allow useful sharing of intelligence, such as the substantial knowledge on crime and illicit transactions gleaned by the counterintelligence branch of the U.S. military that is involved in conflict regions where terror-crime interaction is most profound.
5. Law enforcement personnel must develop stronger working relationships with the business sector. In the past, there has been too little cognizance of possible terrorist-organized crime interaction among the clients of private-sector business corporations and banks. Law enforcement must pursue evidence of criminal affiliations with high status individuals and business professionals who are often facilitators of terrorist financing and money laundering. In the spirit of public-private partnerships, corporations and banks should be placed under an obligation to watch for indications of organized crime or terrorist activity by their clients and business associates. Furthermore, they should attempt to analyze what they discover and to pass on their assessment to law enforcement.
6. Law enforcement personnel posted overseas by federal agencies such as the DEA, the Department of Justice, the Department of Homeland Security, and the State Department’s Bureau of International Narcotics and Law Enforcement should be tasked with helping to develop a better picture of the geography of organized crime and its most salient features (i.e., the watch points of the PIE approach). This should be used to assist analysts in studying patterns of crime behavior that put American interests at risk overseas and alert law enforcement to crime patterns that may shortly appear in the U.S.
7. Training for law enforcement officers at federal, state, and local level in identifying authentic and forged passports, visas, and other documents required for residency in the U.S. would eliminate a major shortcoming in investigations of criminal networks.

 

 

 

 

 

 

 

 

 

 

 

A.1 Defining the PIE Analytical Process

In order to begin identifying the tools to support the analytical process, the process of analysis itself first had to be captured. The TraCCC team adopted Max Boisot’s (2003) I-Space as a representation for de- scribing the analytical process. As Figure A-1 illustrates, I-Space provides a three-dimensional representation of the cognitive steps that constitute analysis in general and the utilization of the PIE methodology in particular. The analytical process is reduced to a series of logical steps, with one step feeding the next until the process starts anew. The steps are:

1. Scanning
   2. Codification 3. Abstraction 4. Diffusion
   5. Validation 6. Impacting

Over time, repeated iterations of these steps result in more and more PIE indicators being identified, more information being gathered, more analytical product being generated, and more recommendations being made. Boisot’s I-Space is described below in terms of law enforcement and intelligence analytical processes.

A.1.1. Scanning

The analytical process begins with scanning, which Boisot defines as the process of identifying threats and opportunities in generally available but often fuzzy data. For example, investigators often scan avail- able news sources, organizational data sources (e.g., intelligence reports) and other information feeds to identify patterns or pieces of information that are of interest. Sometimes this scanning is performed with a clear objective in mind (e.g., set up through profiles to identify key players). From a tools perspective, scanning with a focus on a specific entity like a person or a thing is called a subject-based query. At other times, an investigator is simply reviewing incoming sources for pieces of a puzzle that is not well under- stood at that moment. From a tools perspective, scanning with a focus on activities like money laundering or drug trafficking is called a pattern-based query. For this type of query, a specific subject is not the target, but a sequence of actors/activities that form a pattern of interest.

Many of the tools described herein focus on either:

o Helping an investigator build models for these patterns then comparing those models against the data to find ‘matches’, or

o Supporting automated knowledge discovery where general rules about interesting patterns are hypothesized and then an automated algorithm is employed to search through large amounts of data based on those rules.

The choice between subject-based and pattern-based queries is dependent on several factors including the availability of expertise, the size of the data source to be scanned, the amount of time available and, of course, how well the subject is understood and anticipated. For example, subject-based queries are by nature more tightly focused and thus are often best conducted through keyword or Boolean searches, such as a Google search containing the string “Bin Laden” or “Abu Mussab al-Zarqawi.” Pattern-based queries, on the other hand, support a relationship/discovery process, such as an iterative series of Google searches starting at ‘with all of the words’ terrorist, financing, charity, and hawala, proceeding through ‘without the words’ Hezbollah and Iran and culminating in ‘with the exact phrase’ Al Qaeda Wahabi charities. Regard- less of which is employed, the results provide new insights into the problem space. The construction, employment, evaluation, and validation of results from these various types of scanning techniques will pro- vide a focus for our tool exploration.

A.1.2. Codification

In order for the insights that result from scanning to be of use to the investigator, they must be placed into the context of the questions that the investigator is attempting to answer. This context provides structure through a codification process that turns disconnected patterns into coherent thoughts that can be more easily communicated to the community. The development of indicators is an example of this codification. Building up network maps from entities and their relationships is another example that could sup- port indicator development. Some important tools will be described that support this codification step.

A.1.3. Abstraction

During the abstraction phase, investigators generalize the application of newly codified insights to a wider range of situations, moving from the specific examples identified during scanning and codification towards a more abstract model of the discovery (e.g., one that explains a large pattern of behavior or predicts future activities). Indicators are placed into the larger context of the behaviors that are being monitored. Tools that support the generation and maintenance of models that support this abstraction process

81

will be key to making the analysis of an overwhelming number of possibilities and unlimited information manageable.

A.1.4. Diffusion

Many of the intelligence failures cited in the 9/11 Report were due to the fact that information and ideas were not shared. This was due to a variety of reasons, not the least of which were political. Technology also built barriers to cooperation, however. Information can only be shared if one of two conditions is met. Either the sender and receiver must share a context (a common language, background, understanding of the problem) or the information must be coded and abstracted (see steps 2 and 3 above) to extract it from the personal context of the sender to one that is generally understood by the larger community. Once this is done, the newly created insights of one investigator can be shared with investigators in sister groups.

The technology for the diffusion itself is available through any number of sources ranging from repositories where investigators can share information to real-time on-line cooperation. Tools that take advantage of this technology include distributed databases, peer-to-peer cooperation environments and real- time meeting software (e.g., shared whiteboards).

A.1.5. Validation

In this step of the process, the hypotheses that have been formed and shared are now validated over time, either by a direct match of the data against the hypotheses (i.e., through automation) or by working towards a consensus within the analytical community. Some hypotheses will be rejected, while others will be retained and ranked according to probability of occurrence. In either case, tools are needed to help make this match and form this consensus.

A.1.6. Impacting

Simply validating a set of hypotheses is not enough. If the intelligence gathering community stops at that point, the result is a classified CNN feed to the policy makers and practitioners. The results of steps 1 through 5 must be mapped against the opposing landscape of terrorism and transnational crime in order to understand how the information impacts the decisions that must be taken. In this final step, investigators work to articulate how the information/hypotheses they are building impact the overall environment and make recommendations on actions (e.g., probes) that might be taken to clarify that environment. The con- sequences of the actions taken as a result of the impacting phase are then identified during the scanning phase and the cycle begins again.

A.1.7. An Example of the PIE Analytical Approach

While section 4 provided some real-life examples of the PIE approach in action, a retrodictive analysis of terror-crime cooperation in the extraction, smuggling, and sale of conflict diamonds provides a grounding example of Boisot’s six step analytical process. Diamonds from West Africa were a source of funding for various factions in the Lebanese civil war since the 1980s. Beginning in the late 1990s intelligence, law enforcement, regulatory, non-governmental, and press reports suggested that individuals linked to transnational criminal smuggling and Middle Eastern terrorist groups were involved in Liberia’s illegal diamond trade. We would expect to see the following from an investigator assigned to track terrorist financing:

1. Scanning: During this step investigators could have assembled fragmentary reports to reveal crude patterns that indicated terror-crime interaction in a specific region (West Africa), involving two countries (Liberia and Sierra Leone) and trade in illegal diamonds.
2. Codification: Based on patterns derived from scanning, investigators could have codified the terror- crime interaction by developing explicit network maps that showed linkages between Russian arms dealers, Russian and South American organized crime groups, Sierra Leone insurgents, the government of Liberia, Al Qaeda, Hezbollah, Lebanese and Belgian diamond merchants, and banks in Cyprus, Switzerland, and the U.S.
3. Abstraction: The network map developed via codification is essentially static at this point. Utilizing social network analysis techniques, investigators could have abstracted this basic knowledge to gain a dynamic understanding of the conflict diamond network. A calculation of degree, betweenness, and closeness centrality of the conflict diamond network would have revealed those individuals with the most connections within the network, those who were the links between various subgroups within the network, and those with the shortest paths to reach all of the network participants. These calculations would have revealed that all the terrorist links in the conflict diamond network flowed through Ibra- him Bah, a Libyan-trained Senegalese who had fought with the mujahadeen in Afghanistan and whom Charles Taylor, then President of Liberia, had entrusted to handle the majority of his diamond deals. Bah arranged for terrorist operatives to buy all diamonds possible from the RUF, the Charles Taylor- supported rebel army that controlled much of neighboring civil-war-torn Sierra Leone. The same calculations would have delineated Taylor and his entourage as the key link to transnational criminals in the network, and the link between Bah and Taylor as the essential mode of terror-crime interaction for purchase and sale of conflict diamonds.
4. Diffusion: Disseminating the results of the first three analytical steps in this process could have alerted investigators in other domestic and foreign law enforcement and intelligence agencies to the emergent terror-crime nexus involving conflict diamonds in West Africa. Collaboration between various security services at this junction could have revealed Al Qaeda’s move into commodities such as diamonds, gold, tanzanite, emeralds, and sapphires in the wake of the Clinton Administration’s freezing of 240 million dollars belonging to Al Qaeda and the Taliban in Western banks in the aftermath of the August 1998 attacks on the U.S. embassies in Kenya and Tanzania. In particular, diffusion of the parameters of the conflict diamond network could have allowed investigators to tie Al Qaeda fund raising activities to a Belgian bank account that contained approximately 20 million dollars of profits from conflict diamonds.
5. Validation: Having linked Al Qaeda, Hezbollah, and multiple organized crime groups to the trade in conflict diamonds smuggled into Europe from Sierra Leone via Liberia, investigators would have been able to draw operational implications from the evidence amassed in the previous steps of the analytical process. For example, Al Qaeda diamond purchasing behavior changed markedly. Prior to July 2001 Al Qaeda operatives sought to buy low in Africa and sell high in Europe so as to maximize profit. Around July they shifted to a strategy of buying all the diamonds they could and offering the highest prices required to secure the stones. Investigators could have contrasted these buying patterns and hypothesized that Al Qaeda was anticipating events which would disrupt other stores of value, such as financial instruments, as well as bring more scrutiny of Al Qaeda financing in general.
6. Impacting: In the wake of the 9/11attacks, the hypothesis that Al Qaeda engaged in asset shifting prior to those strikes similar to that undertaken in 1999 has gained significant validity. During this final step in the analytical process, investigators could have created a watch point involving a terror-crime nexus associated with conflict diamonds in West Africa, and generated the following indicators for use in future investigations:

• Financial movements and expenditures as attack precursors;
• Money as a link between known and unknown nodes;
• Changes in the predominant patterns of financial activity;
• Criminal activities of a terrorist cell for direct or indirect operational support;
• Surge in suspicious activity reports.

A.2. The tool space

The key to successful tool application is understanding what type of tool is needed for the task at hand. In order to better characterize the tools for this study, we have divided the tool space into three dimensions:

• An abstraction dimension: This continuum focuses on tools that support the movement of concepts from the concrete to the abstract. Building models is an excellent example of moving concrete, narrow concepts to a level of abstraction that can be used by investigators to make sense of the past and predict the future.
• A codification dimension: This continuum attaches labels to concepts that are recognized and accepted by the analytical community to provide a common context for grounding models. One end of the spectrum is the local labels that individual investigators assign and perhaps only that they understand. The other end of the spectrum is the community-accepted labels (e.g., commonly accepted definitions that will be understood by the broader analytical community). As we saw earlier, concepts must be defined in community-recognizable labels before the community can begin to cooperate on those concepts.
• The number of actors: This last continuum talks in term of the number of actors who are involved with a given concept within a certain time frame. Actors could include individual people, groups, and even automated software agents. Understanding the number of actors involved with the analysis will play a key role in determining what type of tool needs to be employed.

Although they may appear to be performing the same function, abstraction and codification are not the same. An investigator could build a set of models (moving from concrete to abstract concepts) but not take the step of changing his or her local labels. The result would be an abstracted model of use to the single investigator, but not to a community working from a different context. For example, one investigator could model a credit card theft ring as a petty crime network under the loose control of a traditional organized crime family, while another investigator could model the same group as a terrorist logistic sup- port cell.

The analytical process described above can now be mapped into the three-dimensional tool space, represented graphically in Figure A-1. So, for example, scanning (step 1) is placed in the portion of the tool space that represents an individual working in concrete terms without those terms being highly codified (e.g., queries). Validation (step 5), on the other hand, requires the cooperation of a larger group working with abstract, highly codified concepts.

A.2.1. Scanning tools

Investigators responsible for constructing and monitoring a set of indicators could begin by scanning available data sources – including classified databases, unclassified archives, news archives, and internet sites – for information related to the indicators of interest. As can be seen from exhibit 6, all scanning tools will need to support requirements dictated by where these tools fall within the tool space. Scanning tools should focus on:

• How to support an individual investigator as opposed to the collective analytical community. Investigators, for the most part, will not be performing these scanning functions as a collaborative effort;
• Uncoded concepts, since the investigator is scanning for information that is directly related to a specific context (e.g., money laundering), then the investigator will need to be intimately familiar with the terms that are local (uncoded) to that context;
• Concrete concepts or, in this case, specific examples of people, groups, and circumstances within the investigator’s local context. In other words, if the investigator attempts to generalize at this stage, much could be missed.

Using these criteria as a background, and leveraging state-of-the-art definitions for data mining, scanning tools fall into two basic categories:

• Tools that support subject-based queries are used by investigators when they are searching for specific information about people, groups, places, events, etc.; and
• Investigators who are not as interested in individuals as they are in identifying patterns of activities use tools that support pattern-based queries.

This section briefly describes the functionality in general, as well as providing specific tool examples, to support both of these critical types of scanning.

A.2.1.1. Subject-based queries

Subject-based queries are the easiest to perform and the most popular. Examples of tools that are used to support subject-based queries are Boolean search tools for databases and internet search engines.

Functionalities that should be evaluated when selecting subject-based query tools include that they are easy to use and intuitive to the investigator. Investigators should not be faced with a bewildering array of ‘ifs’, ‘ands’, and ‘ors’, but should be presented with a query interface that matches the investigator’s cognitive view of searching the data. The ideal is a natural language interface for constructing the queries. An- other benefit is that they provide a graphical interface whenever possible. One example might be a graphical interface that allows the investigator to define subjects of interest, then uses overlapping circles to indicate the interdependencies among the search terms. Furthermore, query interfaces should support synonyms, have an ability to ‘learn’ from the investigator based on specific interests, and create an archive of queries so that the investigator can return and repeat. Finally, they should provide a profiling capability that alerts the investigator when new information is found based on the subject.

Subject-based query tools fall into three categories: queries against databases, internet searches, and customized search tools. Examples of tools for each of these categories include:

• Queries from news archives: All major news groups provide web-based interfaces that support queries against their on-line data sources. Most allow you to select the subject, enter keywords, specify date ranges, and so on. Examples include the New York Times (at http://www.nytimes.com/ref/membercenter/nytarchive.html) and the Washington Post (at http://pqasb.pqarchiver.com/washingtonpost/search.html). Most of these sources allow you to read through the current issue, but charge a subscription for retrieving articles from past issues.
• Queries from on-line references: There are a host of on-line references now available for query that range from the Encyclopedia Britannica (at http://www.eb.com/) to the CIA’s World Factbook (at http://www.cia.gov/cia/publications/factbook/). A complete list of such references is impossible to include, but the search capabilities provided by each are clear examples of subject-based queries.
• Search engines: Just as with queries against databases, there are a host of commercial search engines available for free-format internet searching. The most popular is Google, which combines a technique called citation indexing with web crawlers that constantly search out and index new web pages. Google broke the mold of free-format text searching by not focusing on exact matches between the search terms and the retrieved information. Rather, Google assumes that the most popular pages (the ones that are referenced the most often) that include your search terms will be the pages of greatest interest to you. The commercial version of Google is available free of charge on the internet, and organizations can also purchase a version of Google for indexing pages on an intranet. Google also works in many languages. More information about Google as a business solution can be found at http://www.google.com/services/. Although the current version of Google supports many of the requirements for subject-based queries, its focus is quick search and it does not support sophisticated query interfaces, natural language queries, synonyms, or a managed query environment where queries can be saved. There are now numerous software packages available that provide this level of support, many of them as add-on packages to existing applications.

o Name Search®: This software enables applications to find, identify and match information. Specifically, Name Search finds and matches records based on personal and corporate names, social security numbers, street addresses and phone numbers even when those records have variations due to phonetics, missing words, noise words, nicknames, prefixes, keyboard errors or sequence variations. Name Search claims that searches using their rule-based matching algorithms are faster and more accurate than those based only on Soundex or similar techniques. Soundex, developed by Odell and Russell, uses codes based on the sound of each letter to translate a string into a canonical form of at most four characters, preserving the first letter.

Name Search also supports foreign languages, technical data, medical information, and other specialized information. Other problem-specific packages take advantage of the Name Search functionality through an Application Programming Interface (API) (i.e., Name Search is bundled). An example is ISTwatch. See http://www.search-software.com/.

o ISTwatch©: ISTwatch is a software component suite that was designed specifically to search and match individuals against the Office of Foreign Assets Control’s (OFAC’s) Specially Designated Nationals list and other denied parties lists. These include the FBI’s Most Wanted, Canadian’s OSFI terrorist lists, the Bank of England’s consolidated lists and Financial Action Task Force data on money-laundering countries. See

http://www.intelligentsearch.com/ofac_software/index.html

All these tools are packages designed to be included in an application. A final set of subject-based query tools focus on customized search environments. These are tools that have been customized to per- form a particular task or operate within a particular context. One example is WebFountain.

o WebFountain: IBM’s WebFountain began as a research project focused on extending subject- based query techniques beyond free format text to target money-laundering activities identified through web sources. The WebFountain project, a product of IBM’s Almaden research facility in California, used advanced natural language processing technologies to analyze the entire internet – the search covered 256 terabytes of data in the process of matching a structured list of people who were indicted for money laundering activities in the past with unstructured in- formation on the internet. If a suspicious transaction is identified and the internet analysis finds a relationship between the person attempting the transaction and someone on the list, then an alert is issued. WebFountain has now been turned into a commercially available IBM product. Robert Carlson, IBM WebFountain vice president, describes the current content set as over 1 petabyte in storage with over three billion pages indexed, two billion stored, and the ability to mine 20 million pages a day. The commercial system also works across multiple languages. Carlson stated in 2003 that it would cover 21 languages by the end of 2004 [Quint, 2003]. See: http://www.almaden.ibm.com/webfountain

o Memex: Memex is a suite of tools that was created specifically for law enforcement and national security groups. The focus of these tools is to provide integrated search capabilities against both structured (i.e., databases) and unstructured (i.e., documents) data sources. Memex also provides a graphical representation of the process the investigator is following, structuring the subject-based queries. Memex’s marketing literature states that over 30 percent of the intelligence user population of the UK uses Memex. Customers include the Metropolitan Police Service (MPS), whose Memex network that includes over 90 dedicated intelligence servers pro- viding access to over 30,000 officers; the U.S. Department of Defense; numerous U.S. intelligence agencies, drug intelligence Groups and law enforcement agencies. See http://www.memex.com/index.shtml.

A.2.1.2. Pattern queries

Pattern-based queries focus on supporting automated knowledge discovery (1) where the exact subject of interest is not known in advance and (2) where what is of interest is a pattern of activity emerging over time. In order for pattern queries to be formed, the investigator must hypothesize about the patterns in advance and then use tools to confirm or deny these hypotheses. This approach is useful when there is expertise available to make reasonable guesses with respect to the potential patterns. Conversely, when that expertise is not available or the potential patterns are unknown due to extenuating circumstances (e.g., new patterns are emerging too quickly for investigators to formulate hypotheses), then investigators can auto- mate the construction of candidate patterns by formulating a set of rules that describe how potentially interesting, emerging patterns might appear. In either case, tools can help support the production and execution of the pattern queries. The degree of automation is dependent upon the expertise available and the dynamics of the situation being investigated.

As indicated earlier, pattern-based query tools fall into two general categories: those that support investigators in the construction of patterns based on their expertise, then run those patterns against large data sets, and those that allow the investigator to build rules about patterns of interest and, again, run those rules against large data sets.

Examples of tools for each of these categories include

1. Megaputer (PolyAnalyst 4.6): This tool falls into the first category of pattern-based query tools, helping the investigator hypothesize patterns and explore the data based on those hypotheses. PolyAnalyst is a tool that supports a particular type of pattern-based query called Online Analytical Processing (OLAP), a popular analytical approach for large amounts of quantitative data. Using PolyAnalyst, the investigator defines dimensions of interest to be considered in text exploration and then displays the results of the analysis across various combinations of these dimensions. For example, an investigator could search for mujahideen who had trained at the same Al Qaeda camp in the 1990s and who had links to Pakistani Intelligence as well as opium growers and smuggling networks into Europe. See http://www.megaputer.com/.
2. Autonomy Suite: Autonomy’s search capabilities fall into the second category of pattern-based query tools. Autonomy has combined technologies that employ adaptive pattern-matching techniques with Bayesian inference and Claude Shannon’s principles of information theory. Autonomy identifies the pat- terns that naturally occur in text, based on the usage and frequency of words or terms that correspond to specific ideas or concepts as defined by the investigator. Based on the preponderance of one pattern over another in a piece of unstructured information, Autonomy calculates the probability that a document in question is about a subject of interest [Autonomy, 2002]. See http://www.autonomy.com/content/home/
3. Fraud Investigator Enterprise: The Fraud Investigator Enterprise Similarity Search Engine (SSE) from InfoGlide Software is another example of the second category of pattern search tools. SSE uses ana- lytic techniques that dissect data values looking for and quantifying partial matches in addition to exact matches. SSE scores and orders search results based upon a user-defined data model. See http://www.infoglide.com/composite/ProductsF_2_1.htm

Although an evaluation of data sources available for scanning is beyond the scope of this paper, one will serve as an example of the information available. It is hypothesized in this report that tools could be developed to support the search and analysis of Short Message Service (SMS) traffic for confirmation of PIE indicators. Often referred to as ‘text messaging’ in the U.S., the SMS is an integrated message service that lets GSM cellular subscribers send and receive data using their handset. A single short message can be up to 160 characters of text in length – words, numbers, or punctuation symbols. SMS is a store and for- ward service; this means that messages are not sent directly to the recipient but via a network SMS Center. This enables messages to be delivered to the recipient if their phone is not switched on or if they are out of a coverage area at the time the message was sent. This process, called asynchronous messaging, operates in much the same way as email. Confirmation of message delivery is another feature and means the sender can receive a return message notifying them whether the short message has been delivered or not. SMS messages can be sent to and received from any GSM phone, providing the recipient’s network supports text messaging. Text messaging is available to all mobile users and provides both consumers and business people with a discreet way of sending and receiving information
Over 15 billion SMS text messages were sent around the globe in January 2001. Tools taking advantage of the stored messages in an SMS Center could:

• Perform searches of the text messages for keywords or phrases,
• Analyze SMS traffic patterns, and
• Search for people of interest in the Home Location Register (HLR) database that maintains information about the subscription profile of the mobile phone and also about the routing information for the subscriber.

A.2.2. Codification tools

As can be seen from exhibit 6, all codification tools will need to support requirements dictated by where these tools fall within the tool space. Codification tools should focus on:

• Supporting individual investigators (or at best a small group of investigators) in making sense of the information discovered during the scanning process.
• Moving the terms with which the information is referenced from a localized organizational context (uncoded, e.g., hawala banking) to a more global context (codified, e.g., informal value storage and transfer operations).
• Moving that information from specific, concrete examples towards more abstract terms that could support identification of concepts and patterns across multiple situations, thus providing a larger context for the concepts being explored.

Using these criteria as a background, the codification tools reviewed fall into two major categories:

1. Tools that help investigators label concepts and cluster different concepts into terms that are recognizable and used by the larger analytical community; and
2. Tools that use this information to build up network maps identifying entities, relationships, missions, etc.

This section briefly describes codification functionality in general, as well as providing specific tool examples, to support both of these types of codification.

A.2.2.1. Labeling and clustering

The first step to codification is to map the context-specific terms used by individual investigators to a taxonomy of terms that are commonly accepted in a wider analytical context. This process is performed through labeling individual terms, clustering other terms and renaming them according to a community- accepted taxonomy.

In general, labeling and clustering tools should:

• Support the capture of taxonomies that are being developed by the broader analytical community; Allow the easy mapping of local terms to these broader terms;
  Support the clustering process either by providing algorithms for calculating the similarity between concepts, or tools that enable collaborative consensus construction of clustered concepts;
• Label and cluster functionality is typically embedded in applications support analytical processes, not provided separately as stand-alone tools.

Two examples of such products include:

COPLINK® – COPLINK began as a research project at the University of Arizona and has now grown into a commercially available application from Knowledge Computing Corporation (KCC). It is focused on providing tools for organizing vast quantities of structured and seemingly unrelated information in the law enforcement arena. See COPLINK’s commercial website at http://www.knowledgecc.com/index.htm and its academic website at the University of Arizona at http://ai.bpa.arizona.edu/COPLINK/.

Megaputer (PolyAnalyst 4.6) – In addition to supporting pattern queries, PolyAnalyst also pro- vides a means for creating, importing and managing taxonomies which could be useful in the codification step and carries out automated categorization of text records against existing taxonomies.

A.2.2.2. Network mapping

Terrorists have a vested interest in concealing their relationships, they often emit confusing or intentionally misleading information and they operate in self-contained and difficult to penetrate cells for much of the time. Criminal networks are also notoriously difficult to map, and the mapping often happens after a crime has been committed than before. What is needed are tools and approaches that support the map- ping of networks to represent agents (e.g., people, groups), environments, behaviors, and the relationships between all of these.

A large number of research efforts and some commercial products have been created to automate aspects of network mapping in general and link analysis specifically. In the past, however, these tools have provided only marginal utility in understanding either criminal or terrorist behavior (as opposed to espionage networks, for which this type of tool was initially developed). Often the linkages constructed by such tools are impossible to disentangle since all links have the same importance. PIE holds the potential to focus link analysis tools by clearly delineating watch points and allowing investigators to differentiate, characterize and prioritize links within an asymmetric threat network. This section focuses on the requirements dictated by PIE and some candidate tools that might be used in the PIE context.

In general, network mapping tools should:

• Support the representation of people, groups, and the links between them within the PIE indicator framework;
• Sustain flexibility for mapping different network structures;
• Differentiate, characterize and prioritize links within an asymmetric threat network;
• Focus on organizational structures to determine what kinds of network structures they use;
• Provide a graphical interface that supports analysis;
• Access and associate evidence with an investigator’s data sources.

Within the PIE context, investigators can use network mapping tools to identify the flows of information and authority within different types of network forms such as chains, hub and spoke, fully matrixed, and various hybrids of these three basic forms.
Examples of network mapping tools that are available commercially include:

Analyst Notebook®: A PC-based package from i2 that supports network mapping/link analysis via network, timeline and transaction analysis. Analyst Notebook allows an investigator to capture link information between people, groups, activities, and other entities of interest in a visual format convenient for identifying relationships, dependencies and trends. Analyst Notebook facilitates this capture by providing a variety of tools to review and integrate information from a number of data sources. It also allows the investigator to make a connection between the graphical icons representing entities and the original data sources, supporting a drill-down feature. Some of the other useful features included with Analyst Note- book are the ability to: 1) automatically order and depict sequences of events even when exact date and time data is unknown and 2) use background visuals such as maps, floor plans or watermarks to place chart information in context or label for security purposes. See http://www.i2.co.uk/Products/Analysts_Notebook/default.asp. Even though i2 Analyst Notebook is widely used by intelligence community, anti-terrorism and law enforcement investigators for constructing network maps, interviews with investigators indicate that it is more useful as a visual aid for briefing rather than in performing the analysis itself. Although some investigators indicated that they use it as an analytical tool, most seem to perform the analysis using either another tool or by hand, then entering the results into the Analyst Notebook in order to generate a graphic for a report or briefing. Finally, few tools are available within the Analyst Notebook to automatically differentiate, characterize and prioritize links within an asymmetric threat network.

Patterntracer TCA: Patterntracer Telephone Call Analysis (TCA) is an add-on tool for the Analyst Notebook intended to help identify patterns in telephone billing data. Patterntracer TCA automatically finds repeating call patterns in telephone billing data and graphically displays them using network and timeline charts. See http://www.i2.co.uk/Products/Analysts_Workstation/default.asp

Memex: Memex has already been discussed in the context of subject-based query tools. In addition to supporting such queries, however, Memex also provides a tool that supports automated link analysis on unstructured data and presents the results in graphical form.

Megaputer (PolyAnalyst 4.6): In addition to supporting pattern-based queries, PolyAnalyst was also designed to support a primitive form of link analysis, by providing a visual relationship of the results.

A.2.3. Abstraction tools

As can be seen from exhibit 6, all abstraction tools will need to support requirements dictated by where these tools fall within the tool space. Abstraction tools should focus on:

• Functionalities that help individual investigators (or a small group of investigators) build abstract models;
• Options to help share these models, and therefore the tools should be defined using terms that will be recognized by the larger community (i.e., codified as opposed to uncoded);
• Highly abstract notions that encourage examination of concepts across networks, groups, and time.

The product of these tools should be hypotheses or models that can be shared with the community to support information exchange, encourage dialogue, and eventually be validated against both real-world data and by other experts. This section provides some examples of useful functionality that should be included in tools to support the abstraction process.

A.2.3.1. Structured argumentation tools

Structured argumentation is a methodology for capturing analytical reasoning processes designed to address a specific analytic task in a series of alternative constructs, or hypotheses, represented by a set of hierarchical indicators and associated evidence. Structured argumentation tools should:

• Capture multiple, competing hypotheses of multi-dimensional indicators at both summary and/or detailed levels of granularity;
• Develop and archive indicators and supporting evidence;
• Monitor ongoing activities and assess the implications of new evidence;
• Provide graphical visualizations of arguments and associated evidence;
• Encourage a careful analysis by reminding the investigator of the full spectrum of indicators to be considered;
• Ease argument comprehension by allowing the investigator to move along the component lines of reasoning to discover the basis and rationale of others’ arguments;
• Invite and facilitate argument comparison by framing arguments within common structures; and
• Support collaborative development and reuse of models among a community of investigators.
• Within the PIE context, investigators can use structured argumentation tools to assess a terrorist group’s ability to weaponize biological materials, and determine the parameters of a transnational criminal organization’s money laundering methodology.

Examples of structured argumentation tools that are available commercially include:

Structured Evidential Argument System (SEAS) from SRI International was initially applied to the problem of early warning for project management, and more recently to the problem of early crisis warning for the U.S. intelligence and policy communities. SEAS is based on the concept of a structured argument, which is a hierarchically organized set of questions (i.e., a tree structure). These are multiple-choice questions, with the different answers corresponding to discrete points or subintervals along a continuous scale, with one end of the scale representing strong support for a particular type of opportunity or threat and the other end representing strong refutation. Leaf nodes represent primitive questions, and internal nodes represent derivative questions. The links represent support relationships among the questions. A derivative question is supported by all the derivative and primitive questions below it. SEAS arguments move concepts from their concrete, local representations into a global context that supports PIE indicator construction. See http://www.ai.sri.com/~seas/.

A.2.3.2. Modeling

• By capturing information about a situation (e.g., the actors, possible actions, influences on those actions, etc.), in a model, users can define a set of initial conditions, match these against the model, and use the results to support analysis and prediction. This process can either be performed manually or, if the model is complex, using an automated tool or simulator.
• Utilizing modeling tools, investigators can systematically examine aspects of terror-crime interaction. Process models in particular can reveal linkages between the two groups and allow investigators to map these linkages to locations on the terror-crime interaction spectrum. Process models capture the dynamics of networks in a series of functional and temporal steps. Depending on the process being modeled, these steps must be conducted either sequentially or simultaneously in order for the process to execute as de- signed. For example, delivery of cocaine from South America to the U.S. can be modeled as process that moves sequentially from the growth and harvesting of coca leaves through refinement into cocaine and then transshipment via intermediate countries into U.S. distribution points. Some of these steps are sequential (e.g., certain chemicals must be acquired and laboratories established before the coca leaves can be processed in bulk) and some can be conducted simultaneously (e.g., multiple smuggling routes can be utilized at the same time).

Corruption, modeled as a process, should reveal useful indicators of cooperation between organized crime and terrorism. For example, one way to generate and validate indicators of terror-crime interaction is to place cases of corrupt government officials or private sector individuals in an organizational network construct utilizing a process model and determine if they serve as a common link between terrorist and criminal networks via an intent model with attached evidence. An intent model is a type of process model constructed by reverse engineering a specific end-state, such as the ability to move goods and people into and out of a country without interference from law enforcement agencies.

This end-state is reached by bribing certain key officials in groups that supply border guards, provide legitimate import-export documents (e.g., end-user certificates), monitor immigration flows, etc.

Depending on organizational details, a bribery campaign can proceed sequentially or simultaneously through various offices and individuals. This type of model allows analysts to ‘follow the money’ through a corruption network and link payments to officials with illicit sources. The model can be set up to reveal payments to officials that can be linked to both criminal and terrorist involvement (perhaps via individuals or small groups with known links to both types of network).

Thus investigators can use a process model as a repository for numerous disparate data items that, taken together, reveal common patterns of corruption or sources of payments that can serve as indicators of cooperation between organized crime and terrorism. Using these tools, investigators can explore multiple data dimensions by dynamically manipulating several elements of analysis:

• Criminal and/or terrorist priorities, intent and factor attributes;
• Characterization and importance of direct evidence;
• Graphical representations and other multi-dimensional data visualization approaches.

There have been a large number of models built over the last several years focusing on counter- terrorism and criminal activities. Some of the most promising are models that support agent-based execution of complex adaptive environments that are used for intelligence analysis and training. Some of the most sophisticated are now being developed to support the generation of more realistic environments and interactions for the commercial gaming market.

In general, modeling tools should:

• Capture and present reasoning from evidence to conclusion;
• Enable comparison of information across situation, time, and groups;
• Provide a framework for challenging assumptions and exploring alternative hypotheses;
• Facilitate information sharing and cooperation by representing hypotheses and analytical judgment, not just facts;
• Incorporate the first principle of analysis—problem decomposition;
• Track ongoing and evolving situations, collect analysis, and enable users to discover information and critical data relationships;
• Make rigorous option space analysis possible in a distributed electronic context;
• Warn users of potential cognitive bias inherent in analysis.

Although there are too many of these tools to list in this report, good examples of some that would be useful to support PIE include:

NETEST: This model estimates the size and shape of covert networks given multiple sources with omissions and errors. NETEST makes use of Bayesian updating techniques, communications theory and social network theory [Dombroski, 2002].

The Modeling, Virtual Environments and Simulation (MOVES) Institute at the Naval Postgraduate School in Monterey, California, is using a model of cognition formulated by Aaron T. Beck to build models capturing the characteristics of people willing to employ violence [Beck, 2002].

BIOWAR: This is a city scale multi-agent model of weaponized bioterrorist attacks for intelligence and training. At present the model is running with 100,000 agents (this number will be increased). All agents have real social networks and the model contains real city data (hospitals, schools, etc.). Agents are as realistic as possible and contain a cognitive model [Carley, 2003a].

All of the models reviewed had similar capabilities:

• Capture the characteristics of entities – people, places, groups, etc.;
• Capture the relationships between entities at a level of detail that supports programmatic construction of processes, situations, actions, etc. these are usually “is a” and “a part of” representations of object-oriented taxonomies, influence relationships, time relationships, etc.;
• The ability to represent this information in a format that supports using the model in simulations. The next section provides information on simulation tools that are in common use for running these types of models.
• User interfaces for defining the models, the best being graphical interfaces that allow the user to define the entities and their relationships through intuitive visual displays. For example, if the model involves defining networks or influences between entities, graphical displays with the ability to create connections and perform drag and drop actions become important.

A.2.4. Diffusion tools

As can be seen from exhibit 6, all diffusion tools will need to support requirements dictated by where these tools fall within the tool space. Diffusion tools should focus on:

• Moving information from an individual or small group of investigators to the collective community;
• Providing abstract concepts that are easily understood in a global context with little worry that the terms will be misinterpreted;
• Supporting the representation of abstract concepts and encouraging dialogues about those concepts.

In general diffusion tools should:

• Provide a shared environment that investigators can access on the internet;
• Support the ability for everyone to upload abstract concepts and their supporting evidence (e.g., documents);
• Contain the ability for the person uploading the information to be able to attach an annotation and keywords;
• Possess the ability to search concept repositories;
• Be simple to set up and use.

Within the PIE context, investigators could use diffusion tools to:

• Employ a collaborative environment to exchange information, results of analysis, hypotheses, models, etc.;
• Utilize collaborative environments that might be set up between law enforcement groups and counterterrorism groups to exchange information on a continual and near real-time basis. Examples of diffusion tools run from one end of the cooperation/dissemination spectrum to the other. One of the simplest to use is:

• AskSam: The AskSam Web Publisher is an extension of the standalone AskSam capability that has been used by the analytical community for many years. The capabilities of AskSam Web Publisher include: 1) sharing documents with others who have access to the local net- work, 2) anyone who has access to the network has access to the AskSam archive without the need for an expensive license, and 3) advanced searching capabilities including adding keywords which supports a group’s codification process (see step 2 in exhibit 6 in our analytical process). See http://www.asksam.com/.

There are some significant disadvantages to using AskSam as a cooperation environment. For example, each document included has to be ‘published’. The assumption is that there are only one or two people primarily responsible for posting documents and these people control all documents that are made available, a poor assumption for an analytical community where all are potential publishers of concepts. The result is expensive licenses for publishers. Finally, there is no web-based service for AskSam, requiring each organization to host its own AskSam server.

There are two leading commercial tools for cooperation now available and widely used. Which tool is chosen for a task depends on the scope of the task and the number of users.

• Groove: virtual office software that allows small teams of people to work together securely over a network on a constrained problem. Groove capabilities include: 1) the ability for investigators to set up a shared space, invite people to join and give them permission to post documents to a document repository (i.e., file sharing), 2) security including encryption that protects content (e.g., upload and download of documents) and communications (e.g., email and text messaging), investigators can work across firewalls without a Virtual Private Network (VPN) which improves speed and makes it accessible from outside of an intranet, 4) investigators are able to work off-line, then synchronize when they come back on line, 5) includes add- in tools to support cooperation such as calendars, email, text- and voice-based instant messaging, and project management.

Although Groove satisfies most of the basic requirements listed for this category, there are several drawbacks to using Groove for large projects. For example, there is no free format search for text documents and investigators cannot add on their own keyword categories or attributes to the stored documents. This limits Groove’s usefulness as an information exchange archive. In addition, Groove is a fat client, peer-to-peer architecture. This means that all participants are required to purchase a license, download and install Groove on their individual machines. It also means that Groove requires high bandwidth for the information exchange portion of the peer-to-peer updates. See http://www.groove.net/default.cfm?pagename=Workspace.

• SharePoint: Allows teams of people to work together on documents, tasks, contacts, events, and other information. SharePoint capabilities include: 1) text document loading and sharing, 2) free format search capability, 3) cooperation tools to include instant messaging, email and a group calendar, and 4) security with individual and group level access control. The TraCCC

team employed SharePoint for this project to facilitate distributed research and document

generation. See http://www.microsoft.com/sharepoint/.
SharePoint has many of the same features as Groove, but there are fundamental underlying differences. Sharepoint’s architecture is server based with the client running in a web browser. One ad- vantage to this approach is that each investigator is not required to download a personal version on a machine (Groove requires 60-80MB of space on each machine). In fact, an investigator can access the SharePoint space from any machine (e.g., at an airport). The disadvantage of this approach is that the investigator does not have a local version of the SharePoint information and is unable to work offline. With Groove, an investigator can work offline, and then resynchronize with the remaining members of the group when the network once again becomes available. Finally, since peer-to-peer updates are not taking place, SharePoint does not necessarily require a high-speed internet access, except perhaps in the case where the investigator would like to upload large documents.

Another significant difference between SharePoint and Groove is linked to the search function. In Groove, the search capability is limited to information that is typed into Groove directly, not to documents that have been attached to Groove in an archive. A SharePoint support not only document searches, but also allows the community of investigators to set up their own keyword categories to help with the codification of the shared documents (again see step 2 from exhibit 6). It should be noted, however, that SharePoint only supports searches for Microsoft documents (e.g., Word, Power- Point, etc.) and not ‘foreign’ document formats such as PDF. This fact is not surprising given that SharePoint is a Microsoft tool.

SharePoint and Groove are commercially available cooperation solutions. There are also a wide variety of customized cooperation environments now appearing on the market. For example:

• WAVE Enterprise Information Integration System– Modus Operandi’s Wide Area Virtual Environment (WAVE) provides tools to support real-time enterprise information integration, cooperation and performance management. WAVE capabilities include: 1) collaborative workspaces for team-based information sharing, 2) security for controlled sharing of information, 3) an extensible enterprise knowledge model that organizes and manages all enterprise knowledge assets, 4) dynamic integration of legacy data sources and commercial off-the-shelf (COtS) tools, 5) document version control, 6) cooperation tools, including discussions, issues, action items, search, and reports, and 7) performance metrics. WAVE is not a COtS solution, however. An organization must work with Modus Operandi services to set up a custom environment. The main disadvantage to this approach as opposed to Groove or SharePoint is cost and the sharing of information across groups. See http://www.modusoperandi.com/wave.htm.

Finally, many of the tools previously discussed have add-ons available for extending their functionality to a group. For example:

• iBase4: i2’s Analyst Notebook can be integrated with iBase4, an application that allows investigators to create multi-user databases for developing, updating, and sharing the source information being used to create network maps. It even includes security to restrict access or functionality by user, user groups and data fields. It is not clear from the literature, but it appears that this functionality is restricted to the source data and not the sharing of network maps generated by the investigators. See http://www.i2.co.uk/Products/iBase/default.asp

The main disadvantage of iBase4 is its proprietary format. This limitation might be somewhat mitigated by coupling iBase4 with i2’s iBridge product which creates a live connection between legacy databases, but there is no evidence in the literature that i2 has made this integration.

A.2.5. Validation tools

As can be seen from exhibit 6, all validation tools will need to support requirements dictated by where these tools fall within the tool space. Validation tools should focus on:

• Providing a community context for validating the concepts put forward by the individual participants in the community;
• Continuing to work within a codified realm in order to facilitate communication between different groups articulating different perspectives;
• Matching abstract concepts against real world data (or expert opinion) to determine the validity of the concepts being put forward.

Using these criteria as background, one of the most useful toolsets available for validation are simulation tools. This section briefly describes the functionality in general, as well as providing specific tool examples, to support simulations that ‘kick the tires’ of the abstract concepts.

Following are some key capabilities that any simulation tool must possess:

• Ability to ingest the model information that has been constructed in the previous steps in the

analytical process;

• Access to a data source for information that might be required by the model during execution;

• Users need to be able to define the initial conditions against which the model will be run;
• The more useful simulators allow the user to “step through” the model execution, examining

variables and resetting variable values in mid-execution;

• Ability to print out step-by-step interim execution results and final results;
• Change the initial conditions and compare the results against prior runs.

Although there are many simulation tools available, following are brief descriptions of some of the most promising:

• Online iLink: An optional application for i2’s Analyst Notebook that supports dynamic up- date of Analyst Notebook information from online data sources. Once a connection is made with an on-line source (e.g., LexisNexistM, or D&B®) Analyst Notebook uses this connection to automatically check for any updated information and propagates those updates throughout to support validation of the network map information. See http://www.i2inc.com.

One apparent drawback with this plug-in is that Online iLink appears to require that the line data provider deploy i2’s visualization technology.

• NETEST: A research project from Carnegie Mellon University, which is developing tools

that combine multi-agent technology with hierarchical Bayesian inference models and biased net models to produce accurate posterior representations of terrorist networks. Bayesian inference models produce representations of a network’s structure and informant accuracy by combining prior network and accuracy data with informant perceptions of a network. Biased net theory examines and captures the biases that may exist in a specific network or set of net- works. Using NETEST, an investigator can estimate a network’s size, determine its member- ship and structure, determine areas of the network where data is missing, perform cost/benefit analysis of additional information, assess group level capabilities embedded in the network, and pose “what if” scenarios to destabilize a network and predict its evolution over time [Dombroski, 2002].

• REcursive Porous Agent Simulation toolkit (REPAST): A good example of the free, open-source toolkits available for creating agent-based simulations. Begun by the University of Chicago’s social sciences research community and later maintained by groups such as Argonne National Laboratory, Repast is now managed by the non-profit volunteer Repast Organization for Architecture and Development (ROAD). Some of Repast’s features include: 1) a variety of agent templates and examples (however, the toolkit gives users complete flexibility as to how they specify the properties and behaviors of agents), 2) a fully concurrent discrete event scheduler (this scheduler supports both sequential and parallel discrete event operations), 3) built-in simulation results logging and graphing tools, 4) an automated Monte Carlo simulation framework, 5) allows users to dynamically access and modify agent properties, agent behavioral equations, and model properties at run time, 6) includes libraries for genetic algorithms, neural networks, random number generation, and specialized mathematics, and 7) built-in systems dynamics modeling.

More to the point for this investigation, Repast has social network modeling support tools. The Repast website claims that “Repast is at the moment the most suitable simulation framework for the applied modeling of social interventions based on theories and data,” [Tobias, 2003]. See http://repast.sourceforge.net/.

A.2.6. Impacting tools

As can be seen from exhibit 6, all impacting tools will need to support requirements dictated by where these tools fall within the tool space. Impacting tools should focus on:

• Helping law enforcement and intelligence practitioners understand the implications of their validated models. For example, what portions of the terror-crime interaction spectrum are relevant in various parts of the world, and what is the likely evolutionary path of this phenomenon in each specific geographic area?

Support for translating abstracted knowledge into more concrete local execution strategies. The information flows feeding the scanning process, for example, should be updated based on the results of mapping local events and individuals to the terror-crime interaction spectrum. Watch points and their associated indicators should be reviewed, updated and modified. Probes can be constructed to clarify remaining uncertainties in specific situations or locations.

The following general requirements have been identified for impacting tools:

• Probe management software to help law enforcement investigators and intelligence community analysts plan probes against known and suspected transnational threat entities, monitor their execution, map their impact, and analyze the resultant changes to network structure and operations.
• Situational assessment software that supports transnational threat monitoring and projection. Data fusion and visualization algorithms that portray investigators’ current understanding of the nature and extent of terror-crime interaction, and allow investigators to focus scarce collection and analytical resources on the most threatening regions and networks.

Impacting tools are only just beginning to exit the laboratory, and none of them can be considered ready for operational deployment. This type of functionality, however, is being actively pursued within the U.S. governmental and academic research communities. An example of an impacting tool currently under development is described below:

DyNet – A multi-agent network system designed specifically for assessing destabilization strategies on dynamic networks. A knowledge network (e.g., a hypothesized network resulting from Steps 1 through 5 of Boisot’s I-Space-driven analytical process) is given to DyNet as input. In this case, a knowledge network is defined as an individual’s knowledge about who they know, what resources they have, and what task(s) they are performing. The goal of an investigator using DyNet is to build stable, high performance, adaptive networks with and conduct what-if analysis to identify successful strategies for destabilizing those net- works. Investigators can run sensitivity tests examining how differences in the structure of the covert net- work would impact the overall ability of the network to respond to probe and attacks on constituent nodes. [Carley, 2003b]. See the DyNet website hosted by Carnegie Mellon University at http://www.casos.cs.cmu.edu/projects/DyNet/.

A.3. Overall tool requirements

This appendix provides a high-level overview of PIE tool requirements:

• Easy to put information into the system and get information out of it. The key to the successful use of many of these tools is the quality of the information that is put into them. User interfaces have to be easy to use, context based, intuitive, and customizable. Otherwise, investigators soon determine that the “care and feeding” of the tool does not justify the end product.
• Reasonable response time: The response time of the tool needs to match the context. If the tool is being used in an operational setting, then the ability to retrieve results can be time- critical–perhaps a matter of minutes. In other cases, results may not be time-critical and days can be taken to generate results.
• Training: Some tools, especially those that have not been released as commercial products, may not have substantial training materials and classes available. When making a decision regarding tool selection, the availability and accessibility of training may be critical.

Ability to integrate with the enterprise resources: There are many cases where the utility of the tool will depend on its ability to access and integrate information from the overall enterprise in which the investigator is working. Special-purpose tools that require re-keying of information or labor-intensive conversions of formats should be carefully evaluated to determine the man- power required to support such functions.

• Support for integration with other tools: Tools that have standard interfaces will act as force multipliers in the overall analytical toolbox. At a minimum, tools should have some sort of a developer’s kit that allows the creation of an API. In the best case, a tool would support some generally accepted integration standard such as web services.
• Security: Different situations will dictate different security requirements, but in almost all cases some form of security is required. Examples of security include different access levels for different user populations. The ability to be able to track and audit transactions, linking them back to their sources, will also be necessary in many cases.
• Customizable: Augmenting usability, most tools will need to support some level of customizability (e.g., customizable reporting templates).
• Labeling of information: Information that is being gathered and stored will need to be labeled (e.g., for level of sensitivity or credibility).
• Familiar to the current user base: One characteristic in favor of any tool selected is how well the current user base has accepted it. There could be a great deal of benefit to upgrading existing tools that are already familiar to the users.
• Heavy emphasis on visualization: To the greatest extent possible, tools should provide the investigator with the ability to display different aspects of the results in a visual manner.
• Support for cooperation: In many cases, the strength of the analysis is dependent on leveraging cross-disciplinary expertise. Most tools will need to support some sort of cooperation.

A.4. Bibliography and Further Reading

Autonomy technology White Paper, Ref: [WP tECH] 07.02. This and other information documents about Autonomy may be downloaded after registration from http://www.autonomy.com/content/downloads/

Beck, Aaron T., “Prisoners of Hate,” Behavior research and therapy, 40, 2002: 209-216. A copy of this article may be found at http://mail.med.upenn.edu/~abeck/prisoners.pdf. Also see Dr. Beck’s website at http://mail.med.upenn.edu/~abeck/ and the MOVES Institute at http://www.movesinstitute.org/.

Boisot, Max and Ron Sanchez, “the Codification-Diffusion-Abstraction Curve in the I-Space,” Economic Organization and Nexus of Rules: Emergence and the Theory of the Firm, a working paper, the Universitat Oberta de Catalunya, Barcelona, Spain, May 2003.

Carley, K. M., D. Fridsma, E. Casman, N. Altman, J. Chang, B. Kaminsky, D. Nave, & Yahja, “BioWar: Scalable Multi-Agent Social and Epidemiological Simulation of Bioterrorism Events” in Proceedings from the NAACSOS Conference, 2003. this document may be found at http://www.casos.ece.cmu.edu/casos_working_paper/carley_2003_biowar.pdf

Carley, Kathleen M., et. al., “Destabilizing Dynamic Covert Networks” in Proceedings of the 8th International Command and Control Research and technology Symposium, 2003. Conference held at the National Defense War College, Washington, DC. This document may be found at http://www.casos.ece.cmu.edu/resources_others/a2c2_carley_2003_destabilizing.pdf

Collier, N., Howe, T., and North, M., “Onward and Upward: the transition to Repast 2.0,” in Proceedings of the First Annual North American Association for Computational Social and Organizational Science Conference, Electronic Proceedings, Pittsburgh, PA, June 2003. Also, read about REPASt 3.0 at the REPASt website: http://repast.sourceforge.net/index.html

DeRosa, Mary, “Data Mining and Data Analysis for Counterterrorism,” CSIS Report, March 2004. this document may be purchased at http://csis.zoovy.com/product/0892064439

Dombroski, M. and K. Carley, “NETEST: Estimating a Terrorist Network’s Structure,” Journal of Computational and Mathematical Organization theory, 8(3), October 2002: 235-241.
http://www .casos.ece.cmu.edu/conference2003/student_paper/Dombroski.pdf

Farah, Douglas, Blood from Stones: The Secret Financial Network of Terror, New York: Broadway Books, 2004.

Hall, P. and G. Dowling, “Approximate string matching,” Computing Surveys, 12(4), 1980: 381-402. For more information on phonetic string matching see http://www.cs.rmit.edu.au/~jz/fulltext/sigir96.pdf. A good summary of the inherent limitations of Soundex may be found at http://www.las-inc.com/soundex/?source=gsx.

Lowrance, J.D., Harrison, I.W., and Rodriguez, A.C., “Structured Argumentation for Analysis,” Proceedings of the 12th Inter- national Conference on Systems Research, Informatics, and Cybernetics, (August 2000).

Quint, Barbara, “IBM’s WebFountain Launched – the Next Big Thing?” September 22, 2003 – from the Information today, Inc. website at http://www.infotoday.com/newsbreaks/nb030922-1.shtml Also see IBM’s WebFountain website at http://www.almaden.ibm.com/webfountain/ and the WebFountain Application Development Guide at
http://www .almaden.ibm.com/webfountain/resources/sg247029.pdf.

Shannon, Claude, “A mathematical theory of communication,” Bell System technical Journal, (27), July and October 1948: 379- 423 and 623-656.

Tobias, R. and C. Hofmann, “Evaluation of Free Java-libraries for Social-scientific Agent Based Simulation,” Journal of Artificial Societies and Social Simulation, University of Surrey, 7(1), January 2003 may be found at http://jasss.soc.surrey.ac.uk/7/1/6.html.