CAIRN-INT.INFO : International Edition

1Anyone who has reviewed the news on the issue of intelligence in the terrorist attacks that have greatly affected France since January 2015 and conducted a discourse analysis will be aware that the key to anticipation lies in tracking “weak signals”—the micro-events that herald a major attack on the integrity of a person or a group, an organization or a territory. When collected and processed appropriately, these allow imminent threats to be prevented and the plans of jihadist networks to be countered, as well as preventing the famous “lone wolves” from acting. How can we not leap at this opportunity? The stronger the vigilance and the amplification of weak signals, the more reactiveness can become proactiveness, the following step being prediction, as shown in the film Minority Report. [1] In conclusion: Long live weak signals! And everyone to your radar screens!

2However, anyone who has reviewed the news on the issue of intelligence in the terrorist attacks and has gone beyond analyzing the discourse to offer a cold analysis of the facts and their connection will have arrived at the conclusion that weak signals have not been tracked by the specialist services. This is either because they do not (yet) have the ability to do so, or because these signals are essentially almost impossible to track (because they are weak), or because they simply do not exist, or, to put it scientifically, they are not an operational concept. It is this latter thesis that we are going to defend here by showing how the weak signal has passed from myth to mystification and how much—far from being simply a semantic argument—its use in some contexts can prove to be counter-productive, irresponsible, and even dangerous. To do this, we will begin from the origins of this concept, explaining the limits of its transposition, and analyzing the forces that have resulted in its change. This examination will allow us, ultimately, to offer an alternative approach where the conceptual relevance brings about the operational performance, making Kurt Lewin’s idea wherein “nothing is more practical than a good theory” our own.

Analyzing a mystification

3As soon as one follows the intelligence news and scientifically analyzes the methods implemented, the relevance of weak signals tends to reduce or even disappear. Following an analysis of the news and of the discourse, we present by way of illustration extracts from three articles from the leading national daily that—thanks to their temporal proximity (after the attack in Nice on July 14, 2016)—allow the mystifying behavior of the concept of the weak signal to be better understood.


To facilitate legal action against “lone wolves,” who become radicalized outside of a network or terrorist organization, the offense of committing an individual terrorist act has been created. It can be characterized by the presence of several “weak signals” before committing the act, such as stockpiling weapons, visiting jihadi areas, or repeatedly visiting websites that glorify terrorism.
(Untersinger, Vaudano, and Boscher 2016) [2]

5It should be noted that it is the weak signals of major acts that are described here, and that the weakness of the signal is entirely relative! But, above all, how to explain that a 19-ton truck (that would kill over eighty and injure more than four hundred in forty-five seconds on the Promenade des Anglais after the fireworks display on July 14, 2016) had been checked and was not stopped when we know that trucks were banned from that area and that the driver said that he was delivering ice creams even though his truck was not refrigerated? However, the coherence/incoherence relationship is the basis for all stages of the investigation. In fact, the use of the concept of a weak signal in the discourse here is to be compared with that of the “lone wolf.” But, as a terrorism expert rightly asked, “Are jihadi wolves so lone?” Here is an answer:


It has been frequently established that a handful of people close to the terrorist were aware of the plan, or were close enough to him that one could envisage the existence of small support base. The issue, which is now central, lies in establishing how to detect and then neutralize such individuals in time, on the basis of what weak signals and according to what legal procedure. The figure of the lone wolf is the one that is most frequently mentioned incorrectly. It is often said that Mohamed Merah, the Toulouse and Montauban killer in 2012, was one. However, nothing could be further from the truth, and it has been well documented that Merah, whose crimes others have claimed responsibility for, acted on orders, albeit autonomously, but not without connections. Confusion over the tactics put forward nearly fifteen years ago by the thinkers of jihad were here used to mask a failure.
(Le Monde 2016)

7It is true that, along with the concept of “lone wolves,” that of the “weak signal” is also misused. Otherwise, how does one explain that signals that are said to be weak, and that we know a posteriori have been identified, do not allow the intelligence services to anticipate future attacks? So much so that an analysis of the trajectories of most terrorists shows that they were, for the most part, known to the authorities, generally listed on the “fiche ‘S’,” [3] and sometimes even considered priority targets. Is this an issue of competence? Undoubtedly sometimes, as in all industries, but the intelligence services have a high level of professionalism. An issue with resources then? Without ignoring the lack of resources of some services—beginning with the intelligence that is considered territorial—the main key is nevertheless found elsewhere, in our opinion… in the non-operational nature of the concept of the weak signal.


“Terrorists will strike where there is a symbol that has not yet been taken into account,” is the belief of a source within the intelligence services, for whom one of the difficulties lies in understanding weak signals. Consequently, “the net tightens significantly,” according to Patrice Ribeiro, general secretary of the police officers’ union, Synergie officiers. “Some signals would not have been considered as seriously a year ago. It is not hysteria, it is paradigmatic. The threat is such that the services are operating on a precautionary principle.”
(Belouezzane et al. 2016.)

9Here, the weak signal functions as a shield allowing all checks to be justified, the principle of precaution obviously posing questions to the rule of law. But why not if it is effective, say the advocates of a pragmatic approach? The fact is that recent attacks, thwarted or not, were not thwarted as a result of weak signals having been detected. Therefore, it was not the sudden radicalization of one of the ringleaders (on the fiche “S”) of the attempted attack on Notre-Dame in Paris that alerted the authorities, but the car parked next to the cathedral for two days, containing five bottles of gas that did not explode owing to the terrorists’ incompetence. But there is worse: far from not being detected, weak signals can even disrupt the intelligence, according to some senior staff.


Since the start of 2016, the police in Lyon have processed nearly forty cases of expressing support for terrorism. The chief of security for the Rhône department notes that day to day offenses with terrorist connotations are on the increase, but that caution is required when analyzing the phenomenon because extreme attention by the authorities will also produce a magnifying effect: “We handle all events without missing anything, police work must now scour all signals, which may give the impression of overreacting to facts that, before the attacks, may have seemed banal and commonplace.”
(Schittly 2016.)

11Ineffective, weak signals may even prove to be counter-productive. Therefore, they are like the face of a person you are waiting for in a public place: The longer the wait, the more people passing by resemble that person. But how can the spread of a non-operational concept be explained? Was it always there and, if not, in what context?

A short history of a myth

12Imported from the United States and, more specifically, from literature on business strategy and management, this concept of the weak signal is especially well-known to economic intelligence specialists who share part of the culture and methodology of intelligence agents.

13In 1975, H. Igor Ansoff presented weak signals as elements allowing any “strategic surprise” to be anticipated by commercial organizations. Detecting them would give decision-makers the possibility of anticipating the consequences of events that were difficult to forecast, and particularly those that threatened the economic survival of the organization. From the beginnings of this concept, we have also been able to note that it has been furthered by anxiety-inducing claims: weak signals exist because it is possible to identify them retroactively, and they represent a risk, hence the necessity to detect them. In 1984, Ansoff defined the weak signal as a “a development about which only partial information is available at the moment when the response must be launched, if it is to be completed before the development impacts on the firm.” The fact that the information is “partial” suggests to Humbert Lesca (2003) that we are closer to intuition in characterizing information or a so-called “weak” signal.

14Therefore, the issue lies less in the ability to detect weak signals and more in deciding which snippet of information needs to be studied closely. Blanco’s view (1998, cited in Lesca 2000), is that, rather than signals (which each receiver can perceive in the same way and which rely on the intent of the sender), it should be “signs” that are processed, the meanings of which remain negotiable. The possible interpretation of these signs, the context within which they are sent and received, the emergency or uncertain situation in which the decision-maker who has to rely upon them finds themselves, are central elements that emphasize that there would be as many “weak signs” as there were senders communicating the same information and receivers processing it. The weak signal is “paradoxical” (Cahen 2011). It corresponds to an intent, a look, and an attention that can only ever be subjective—taking on even greater importance when the basic premise is that an event will take place regardless, and that it is necessary, or even vital, to predict it. Therefore, it seems that the weak signal is an operational concept in the field of forecasting. But is it as operational in the field of surveillance (monitoring, intelligence)? Is the highlighted paradox relevant in contexts other than that of anticipating breakdowns? The fact is that, in the professional intelligence community (in the broad sense, so including monitoring and economic intelligence professionals), weak signals are going to become a myth, in other words a set of beliefs and idealized representations surrounding an object and shared, along with detection tools, by a group that strengthens its identity through it. Myth should not be seen negatively, at least during the consolidation phase of a community. But it may become counter-productive over time once it freezes practices and prevents any questioning. This is indeed the case of weak signals that tell a story, that of an organization that is capable of anticipating developments in its environment by tracking—after having amplified them—weak signals. Seeing before everyone what others do not yet see by rationalizing what cannot be rationalized during the decision-making process—intuition and subjectivity—and by limiting as much as possible the risk that is inherent in every choice. We expect several players to find themselves in weak signals: watchers who have to legitimize their duties, managers who fear the uncertainties of the future, security services who dream of reducing risks, or even consultants and suppliers of software solutions who see a market opening for their methods and tools. Are weak signals going to pass you by? Fear not. The right radar for you is now here...

15First of all, we note that, if these weak signals are so important, why not deal with strong signals? The increase in natural disasters, the predicted end of oil, structural debt overloading states, chronic industrial decline, medical desertification, teachers “burning out,” a drop in the education level, brain drain, loss of competitiveness, an aging population, nuclear risks, and so on: Whatever the sector, how does one explain why the decision-makers do not process these so-called strong signals more rapidly, often waiting instead for the crisis or the accident to take place before reacting? Titanic syndrome clearly has its best days ahead of it. Despite having been alerted to the presence of icebergs, the ocean liner’s crew did not know how to avoid them. We all know what happened next… and the solution lies in the decision-making process and the concept of knowledge as an individual and collective product, derived from information and guided by a common vision of the goal to be reached and the meaning to be given to the act (Moinet 2011).

Weak signals as a convention: The arrival of “big data” and algorithms

16The use of digital devices, especially for processing large sets of data to which they give access (“big data”), is today central to the processing chain used to produce information. Metadata, or “information about information,” are an issue for both intelligence services (telephone calls, connections to web platforms, geo-location, and so on) and organizations (“friend” relationships on digital social networks, time spent watching a video, and so on). These two entities, services and organizations, are equipped with numerous software tools for tracking and processing the large volumes of data produced as standard by each user of smart devices. Therefore, many monitoring software publishers offer to detect weak signals (Alloing 2016) by comparing volumes of information disseminated online (what is not strong is weak), but also by profiling the behavior of users of certain platforms: after defining a consumer-type profile, any “deviant” behavior by a consumer considered to be “influential” or “representative” of a given group will be judged as a weak signal. The “politics of large numbers” and “statistical reasoning” of Alain Desrosières (2011) rely on a sizable ally (the Web) and on computing technologies, the capacities of which are continually increasing. It would take too long to list all the promises and applications of “big data” from a commercial perspective. One apparently recurrent element that has not escaped the intelligence services, since it is linked to the concept of the weak signal, is that of prediction. For example, the PredPol software (used in particular by the police in Atlanta) offers to help police anticipate future crimes by compiling and processing existing crime statistics.

17This processing relies on algorithms functioning as a “black box” and employs various mathematical models. During debates on the “intelligence law” in France in 2015, the government emphasized its willingness to install “within Internet service providers and some major websites, [an] automatic data analysis device created by the intelligence services and intended, depending upon the form of the draft legislation, ‘to reveal a terrorist threat’ automatically” (Untersinger 2015). As with weak signals, it should be noted that this automatic processing of large sets of data is intended to handle signals, not signs. The context is not simply set aside but homogenized in order to facilitate the calculation of the data collected. Furthermore, and as Rouvroy and Berns (2013) highlighted, algorithmic calculation desubjectivizes the subjects in order to make them calculable. In short, as data processing algorithms are unable to analyze the signs, they produce signals. However, if, for commercial purposes, it proves of interest to create a favorable context from predictions sourced from desubjectivized data and signals produced from statistical processing—in other words transforming the weakness of the signal into a trend—, one might wonder about the relevance of such an approach within an intelligence context. As Bilel Benbouzid (2016) emphasizes about PredPol: “The market techniques used reduce discussions to persuasion.” Therefore, the myth of the weak signal seems to be solvable within the new conventions concerning predictive algorithms. These conventions connect the self-realizing nature of some forward-looking approaches, the often well-founded practices of governance by statistics, and the often prophetic discourses of the dominant Web players. Does the question of meaning, although it may be essential in making a decision, give way to that of immediate performance?

It is not the signal that is weak, it is intelligence

18From the foundations of economic intelligence, both among Anglo-Saxon authors and for the initiators of the Martre report, it goes without saying that information alone does not make it possible to take action. For Wilensky (1967), intelligence is understood to be the systematic collection, interpretation, and valuation of the information for the pursuit of strategic goals. For him, this is not a process of accumulating information but rather of producing knowledge, by governments and businesses, and, when necessary, within the framework of collective strategies. And the real competitive and strategic leverage is found in interpretation skills when the main obstacles are the organizational rigidities (hierarchization and bureaucratization in particular). But once this caveat has been avoided, daily life takes over and the concept of information dominates. However, the mechanical designs of the information referred to above would create a lasting mark on the practice of imprinting them. Therefore, the 1990s would see the birth of numerous monitoring cells within organizations for researching the information that is considered to be strategic as well as our famous weak signals. However, the founding report on economic intelligence, a collective work produced by a group of experts, states:


Lots of businesses that have created a central economic intelligence department have failed in their efforts. It seems that the capacity for understanding and knowledge in relation to economic intelligence is spread informally and amounts to a process. The action of a centralized structure does not produce sufficient intensity of knowledge for the effective understanding of complex environments, nor does it allow its rapid dissemination.
(Martre 1994)

20Initially, the concept of a weak signal enabled organizations to become more vigilant, but it is now counter-productive in numerous fields because it results in methods of organization that are contrary to their spirit. In a forward-looking vision intended to anticipate change, the weak signal should enable an organization to reflect collectively on the elements that herald breakdowns. But under the influence of pyramid organizations intended to perform checks on an unobtainable reality, along with the development of conventions relating to the relevance of the automated processing of large data sets, the necessary work of pooling various interpretations of the same event seems to disappear.

21The subject of weak signals is therefore a case of the trees hiding the wood. Beyond this ultimately lies the question of the relationship between a world of intelligence that is closed by necessity and an academic community that is open by nature.


  • [1]
    This (2002) Steven Spielberg film takes us to Washington in 2054, where crime has been successfully eradicated. Thanks to visions of the future provided by three individuals who are extraordinarily gifted in precognition, “PreCrime” officers can stop criminals before they commit their offenses.
  • [2]
    Unless otherwise stated, all translations are our own.
  • [3]
    Translator’s note: Literally, “the ‘S’ file.” In France, individuals who are considered to be a serious threat to national security are placed on the fiche ‘S.’


  • Alloing, Camille. 2016. (E)réputation: médiation, calcul, émotion. Paris: CNRS éditions.
  • OnlineAnsoff, H. Igor. 1975. “Managing Strategic Surprise by Response to Weak Signals.” California Management Review 18, no. 2: 21-33.
  • Ansoff, H. Igor 1984. Implanting Strategic Management. Englewood Cliffs, NY: Prentice Hall International.
  • Belouezzane, Sarah, Olivier Faye, Soren Seelow, Julia Pascual, and Geoffroy Deffrennes. 2016. “Terrorisme: l’inquiétude s’installe à son plus haut niveau.” Le Monde, August 6.
  • Benbouzid, Bilel. 2016. “À qui profite le crime? Le marché de la prédiction du crime aux États-Unis.” La vie des idées (online), September 13, 2016. Accessed October 20, 2017.
  • Blanco, Sylvie. 1998. Gestion de l’information et intelligence stratégique: cas de la sélection des signes d’alerte précoce de veille stratégique, PhD thesis in management sciences, University of Grenoble 2, UPMF, ESA, CERAG.
  • Cahen, Philippe. 2011. Signaux faibles: mode d’emploi. Paris: Eyrolles.
  • Castagnos, Jean-Claude and Humbert Lesca. 2004. “Capter les signaux faibles de la veille stratégique: retours d’expérience et recommandations.” Economia & Gestão (E & G) 4, no. 7: 15-34.
  • Desrosières, Alain. 2011. The Politics of Large Numbers: A History of Statistical Reasoning. Translated by Camille Naish. Cambridge, MA: Harvard University Press.
  • Le Monde. 2016. “Les loups du djihad sont-ils si solitaires?” Le Monde, July 16. Accessed October 20, 2017.
  • Lesca, Humbert. 2003. Veille stratégique. La méthode L.E. Scanning. Cormelles-le-Royal: Éditions EMS.
  • Lesca, Nicolas. 2000. “Processus de construction du sens à partir de signes d’alerte précoce: proposition d’un nouvel outil d’aide à la production de connaissance: PUZZLE.” Proceedings of the ninth conference of the Association internationale de management stratégique (AIMS), Montpellier, May 24-26, 2000.
  • Martre, Henri, ed. 1994. Intelligence économique et stratégie des entreprises. Report by the Commissariat Général du Plan working group. Paris: La Documentation Française.
  • Moinet, Nicolas. 2011. Intelligence économique. Mythes & réalités. Paris: CNRS Éditions.
  • OnlineRouvroy, Antoinette and Thomas Berns. 2013. “Gouvernementalité algorithmique et perspectives d’émancipation.” Réseaux 177: 163-196.
  • Schittly, Richard. 2016. “Dans le contexte terroriste, plus d’incidents et de fausses alertes.” Le Monde, August 9.
  • Untersinger, Martin. 2015. “Loi sur le renseignement: la “boîte noire” reste obscure.” Le Monde, April 1.
  • Untersinger, Martin, Maxime Vaudano, and Marie Boscher. 2016. “Terrorisme: le gouvernement n’a-t-il vraiment rien fait depuis deux ans?” Le Monde, July 18.
  • Wilensky, Harold L. 1967. Organizational Intelligence: Knowledge and Policy in Government and Industry. New York: Basic Books.
Camille Alloing
Camille Alloing is an assistant professor in information and communication sciences at the IAE (Institut d’Administration des Entreprises) at the University of Poitiers. After having worked as an R&D consultant and engineer in the field of information management, he now focuses his academic research on issues concerning the practice of online information consumption, specifically on an organizational level.
Nicolas Moinet
Nicolas Moinet is a professor at the IAE at the University of Poitiers and a member of the CEREGE (EA 1722) research team. A practitioner-researcher in economic intelligence, he has written numerous works and articles on this topic, and focuses in particular on issues of organization and network management. A reservist, he is also interested in issues of economic security.
Uploaded on on 26/12/2017
Distribution électronique pour C.N.R.S. Editions © C.N.R.S. Editions. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Loading... Please wait