A confrontation between rival supporters of two São Paulo football clubs recently resulted in the deaths of two young fans who were killed by shots fired during the disturbance. It is sad to report this type of occurrence is far from occasional or exceptional and it suffices to watch any sports match on a Sunday to witness the unfortunate practice of fighting between supporters of rival teams. These melees often result in significant losses and injuries that must be indemnified. But indemnified by whom? How does one substantiate causation of the harm suffered and impute liability to the individual alleged to be responsible?

The analysis presented here deals with situations where the harm is caused by an unidentifiable member of a known group. While it is possible to identify the social segment that generated the harmful conduct, it is impossible to individualize such conduct, “hidden” as it is amidst the mass of individuals who form the group. This raises questions of whether group liability can exist absent the proper identification of the agent who caused the damages, and how to attribute causation and damages to his behavior.

Before delving into an analysis of the circumstances surrounding this case and the elements needed to understand it, we must first address the theme of co-authorship as used to develop the theory of collective liability.

i. Co-authorship and the distribution of liability: complementary causation (concomitant causes) and cumulative causation (concurrent causes).

Investigating causation can lead to the realization that damages are actually the result of various conduct or activities that together led to that particular loss occurring. This arises either because each person’s conduct, in itself, is sufficient to cause the outcome (concurrent causation), or because each of the forms of conduct requires some other action or conduct to complement it in order to generate the harm in the manner it occured (complementary or concurrent causation).

Identifying a plurality of individuals potentially liable for the harm demonstrates the phenomenon of co-authorship or co-participation, which allows us to see the individualization of the conduct of all the actors who together contribute to producing the harm. Co-authorship will be delimited when we are confronted with situations of multiple causation, thus giving rise to the liability of all those whose conduct contributed to the harm.

Both complementary causation and cumulative causation, whether simultaneous or successive, present questions concerning the degree of liability of each of the tortfeasors. Should they all answer for the result as it occurred, or should there be an equitable apportioning of liability to the extent of each actor’s participation in causing the harm? In other words, will there be joint liability among all the tortfeasors, regardless of their degree of fault, or will each of them only be liable to the extent of their participation in resulting the harm?[iii]

The general rule provides for joint liability where simultaneous causation occurs (article 942, Civil Code[iv]). As Ignacio de Cuevillas Matozzi sees it, the application of this rule must always be considered in accordance with the principles of equity. Only then can indemnificatino be permitted where there are multiple causes. Requiring individualized proof of the causal link of each actor in performing the harmful act would compromise the very purpose of civil liability, which is to repair unjust harm. The demonstration of this proof would be render the system unworkable, especially on account of the principle of unity of damage.[v]

Conceptualizing harm as the result of a single item – seeing as the offense is single – suggests the use of joint fault. In the classic example of simultaneous complementary causes, if two people, unaware of each other’s actions, administer poison to a third party who dies as a result, both will held jointly liable for the death (thereby identified as a single outcome), regardless of whether they knew that each of their actions alone was insufficient to cause death. That is, even if one of the agents later proves that the poison they administered was not strong enough to cause a person’s death on its own, they would still be held liable, at civil law, on account of their causal contribution to the death.[vi] Likewise, if the two doses of poison were each independently capable of causing death, we would then have a situation of concurrent causation leading to the same result as individual liability but also allowing one of the parties to allege the irrelevance of their conduct, given that death would result anyway due to the conduct of the second party who administered an equivalent dose.[vii]

Investigating multiple causation becomes relevant to the study of collective liability where there are several actions that possibly caused the harm (theoretically, concurrent causation), and where it is impossible to identify the particular conduct specifically, and solely, responsible. This raises the question of whether an obligation exists on the part of the group to indemnify and whether the nature of this liability is based in a theory of co-authorship or cumulative causation, among others, as we shall see below.

ii. Alternative causation: theory and application

Typically identified with co-authorship – even though it conceptually does not refer to the same exact elements, as we shall see – “anonymous fault”[viii] as coined by Ignacio de Cuevillas Matozzi, or “alternative causation” or “collective liability”[ix] as referred to by Portuguese authors, is characterized by harm caused by an unidentifiable member of a determinable and known group.[x] In other words, any of the group’s members could have caused the harm, hence the label “alternative causation”.

A brief critique concerning the nomenclature used to explain this theory is in order. Adopting the phrase “anonymous fault”, which symbolically expresses the issue in question is, nonetheless, by its very nature, misleading. The concepts raised in this theory can refer to situations of both subjective and objective civil liability, which explains why it is wrong to speak of fault. Furthermore, the notion of fault are for this purpose factors that attribute responsibility rather than identifying the responsible party, inasmuch as they qualify the conduct or activity by establishing whether it creates an obligation to indemnify. It would be more appropriate to deal with this situation using the expression “anonymous conduct”, without classifying it as blameworthy or culpable, or even “anonymous cause”, since in reality causation – an element to identify the responsibility party – is anonymous.

Likewise, the idea of “collective liability” is exaggerated and overly broad in that not all members of the group are personally responsible for the damage that was brought about. Proving that their participation in the damaging event was impossible suffices to negate their liability.[xi] Moreover, the expression “collective” suggests the idea of joint conduct or a joint activity, whereas the truth is that this situation refers to individual conduct (or to multiple individuals’ conduct) that is not identifiable and confused with the purpose of grouping several actions together, even though it is the outcome of this.

The best expression to define this situation is, without a doubt, “alternative causation”. But even this expression can be the object of critique. Suffice it to think of the meaning of “alternative”, which conveys the idea of possible choices. Alternative causation, however, does not symbolize the substitution of responsibility, that is, it does not entail identifying individual and alternative causes. Instead, there exists one single causal connection that defies direct identification. This explains the theory’s presumption with respect to the group as a whole. However, it is not a matter of claiming that one individual or another is responsible, but only the group itself, since it is not possible to point to the individual whose conduct was the direct cause of the damage. This expression, “alternative causation”, being closest to the real function of the theory in question, will be preferred throughout this text.

In regards to the proclaimed similarity between co-authorship (multiple causation) and alternative causation, one point has to be cleared up to avoid theoretical distortions. The necessary reservation has to do with attributing liability. There are three possible consequences of identifying the existence of multiple causation, as seen above: 1) liability is jointly apportioned among those alleged responsible for the damage (article 942, CC); 2) liability is apportioned based on each agent’s causal contribution to the injury; and 3) liability is attributed to only one of the agents, if it is possible to identify a separation between the causal connection of the two successive acts. Theoretically, the liability derived from alternative causation does not fit any of the hypotheses of attribution of liability described above. This is because the harmful conduct can be unified, which does not characterize multiple causation. Furthermore, multiple causation refers to the capacity for causal contribution of several conducts or actions to the damaging events, which does not occur in the case of alternative causation, where the damage is caused by one sole conduct that is impossible to attribute to individual actors on account of the group being a cohesive unit. Consequently, the use of “co-authorship” to refer to the responsibility of the group and its members only occurs due to the need to compensate the victim, who would otherwise receive nothing. Therefore, strictly speaking there is no co-authorship, since the group answers as a whole and not as if all of its members had contributed to the resulting harm.

The critiques formulated above serve to introduce the issues surrounding alternative causation. The central question for this study is as follows: is it reasonable deny compensation for the harm suffered due to the inability of identifying the individual responsible; or, is it fair to hold the whole group responsible for the conduct of one of its members who, even though unidentified, was the actual cause of the harm? The choice, of course, is not easy.[xii]

There is no doubt that a tendency exists to treat the group as a whole when attributing civil liability as a way of facilitating reparation of the damages that were unjustly caused. At first, one might imagine this type of situation as rare and that it should therefore not be analyzed in such detail. But this is mistaken thinking. For example, one can easily think of those daily situations where damage is caused by a group of children playing with a ball that breaks a neighbor’s window, or the situation where someone is injured as a result of a car accident with multiple actors, a so-called “pile-up”, without anyone knowing which of the cars caused the accident.

In such cases we are confronted with damages caused by identified groups where the conduct of the person who directly caused the damage for which compensation is being sought is not capable of being isolated and identified. We can also identify the following examples of collective liability: 1) illegal street racing; 2) fighting between rival gangs[xiii]; 3) public demonstrations, such as protests; and 4) medical liability in surgery, among others.

This last example was given extensive jurisprudential recognition in Argentina,[xiv] which examined the situation where multiple medical professionals are present in an operating room when damage is caused to the patient undergoing surgery. In this situation, how is it possible to attribute liability for the damages to just one of the individuals present? Unless identification of causation is absolutely direct, the group will have to answer together as a whole, or jointly, given the impossibility of determining who should otherwise bear the burden of compensating for the damage.

The question that arises in this case refers precisely to identifying causation. When we address the theory of collective liability, the causal link is not delimited in an individualized or specific manner. In collective liability, the causal link connects the damage caused to the group to which the individual(s) that caused the damage belongs, even though it is not possible to identify him. This is a case of external causation, which only identifies the group from which the conduct or activity that caused the damage originated, without the perpetrator being specifically identified. A reasoning of exclusion of causation is what indicates: the group is held liable because the damage could only have been caused by one of its members, even though the individual actually responsible is unknown.

Some relevant insights are gained from the conceptualization of this special type of liability. The first has to do with the very principles and values that serve as a basis for the general theory of civil liability. The recognition of alternative causation by the lawmakers – whether as a general clause or in specific cases – and in the jurisprudence is a departure from the underlying principle of causation, namely that “for each injury, there is one cause”.

At the same time, accepting collective civil liability entails acknowledging the principle of social solidarity, the basis of the Democratic Rule of Law, as well as human dignity. The victim thereby occupies a prominent place in the administration of justice, with the reparation of damages suffered a priority, leaving aside another facet of civil liability, namely the “punishment” of those responsible for the harm. Regardless of who actually caused the injury, the group answers as a collective unit since the objective is to repair the damage suffered rather than punish those responsible.

The study of foreign legislation has contributed greatly to the understanding of this theory. German Law, for example, contains an express statement in its BGB, § 830 which addresses collective liability. According to this article, “if several people caused damage by an illicit action carried out together, each one of them is responsible for the damage. The same holds true when it is not possible to discover the sole person whose particular acts caused the damage”. Additionally, Dutch legislation (Civil Code, article 99), for example, contains a general clause admitting collective liability.[xv]

Although Brazil has not formulated a specific norm to deal with the question, alternative causation has been adopted in more limited cases. One example is article 938 of the Civil Code, which establishes civil liability for the residents of an apartment building for any objects that fall or are thrown from it. This is a situation of alternative causation, seeing that if it is not possible to identify from which apartment the object was thrown, the entire building will answer for the harm suffered. According to Maria Celina Bodin de Moraes,

The judicial decision to hold the condominium liable, by literally interpreting the norm contained in article 1.529 of the Civil Code [the author is referring to the Civil Code of 1916], creates additional burdens on the residents, but helps the victim, for whom any other solution, albeit more faithful to the letter of the law, would be difficult. Imagine if someone nowadays was hit by a projectile falling or thrown from a building with numerous floors and flats and had to commence judicial proceedings against all of the residents of the building, or even of one or more surrounding blocks from where the object presumably fell or was thrown: the scenario is unlikely. [xvi]

The theory of collective liability was therefore developed through caselaw and cannot be inferred from a direct reading of article 938, which provides that “an inhabitant of a building, or part of it, is responsible for the damage caused by the things that fall or are improperly thrown from it”. Judicial interpretation considers collective liability only when its necessary elements are present, the principal one being that the actual tortfeasor is unknown. [xvii]

Analysis of other Brazilian legislation, namely articles 3, 14 and 19 of the Football Supporters’ Bylaws (Law 10.671/03), leads to the conclusion that the legislature was primarily concerned with preserving the safety of the victim and fully compensating their losses. The bylaw states that the football clubs are liable for any damage caused by supporters inside the stadium, a responsibility defined depending on which club owns the stadium (that is to say, whether it is a home or away game).[xviii] In this sense, if any injury is sustained by a supporter inside the football stadium, the home team organizers[xix] (established by the organizers of the sports event before the championship begins) must compensate him for the harm, regardless of whether there exists direct causation between the promotion of the sports event and the resulting damage.[xx] Accordingly, compensation for the damages is sought from whomever is best-positioned to indemnify. If it is possible to identify the cause of the damage, in the sense of identifying the agent (or agents) whose conduct brought it about, the organizer may exercise a right of recourse against him. [xxi]

It can be claimed, nonetheless, that this creates an actual presumption of liability. In this situation, the group does not answer for the damages caused by an unidentifiable member because if that were true, the whole group of organized supporters, for example, would be the ones responsible for the damage caused by one of their members – which is not the case as set down by law. Instead, the current situation places liability on the organizer, as a provider of sports services, for the damages caused by those who benefit from the service or due to a security flaw. The legislature seems to have tried to guarantee compensation of the victim for the injury that occurs at sporting events by establishing liability of the event organizer for their direct actions but also their liability regardless of the presence of any causal link between the activity they engaged in and any harm that arose.

The fundamental basis of this liability is to ensure compensation of the victim by creating a presumption in regards to causation. [xxii] Alternative causation seeks to enable compensation for harm by shifting the burden of proof. Instead of the victim having to prove that the conduct of a certain individual caused his injury, he can rely on a presumption of causation; it is sufficient for him to prove that he suffered some harm and that the harm was the result of a certain activity performed by an identifiable group. [xxiii]

It should be noted that the smaller the group, the easier it is to understand the application of alternative causation. For example, if the group only has two members, there exists a 50% chance that each caused the damage. However, if the group is made up of four members, then the probability of any one of them having caused the harm is reduced to 25%, which can potentially lead to a result considered unfair inasmuch as the three members who did not cause the damage are also held liable, even though they only contributed one fourth of the possible risk of harm.

Still and all, for Hassen Aberkane, an author who produced an early study on alternative causation, the size of the group can play a puzzling role in analyzing collective liability, to the extent that it would be considered fairer to place responsibility on a group with two members, where only one will be unfairly held liable, than on a group with twenty members, where nineteen will be unfairly held liable in order to indemnify for a given injury. We dare to disagree with this position, on the grounds that the basis for liability under alternative causation is not the formation of the group, or its composition, but rather the impossibility of establishing causation between the conduct of the individual actually responsible for the harm – a member of the group – and the resulting injury. If the damage results directly and, by implication, necessarily from a collective group action, the obligation to indemnify must be attributed to that group if it is impossible – or practically impossible – to point to one individual as the actor behind the deed. The internal relations between the group members will be resolved a posteriori (through the right of recourse), but the fundamental objective of civil liability, compensating for harm suffered, must be given priority.

Faced with this dilemma and, consequently, the fairness of applying alternative causation in certain cases, we must now analyze the basis for alternative causation and its requirements.

iii. Basis of alternative causation.

In eschewing a determination of presumption of causation, magistrates have justified collective liability based on two principles: (i) collective fault in the activity and (ii) collective control of the object that caused the harm. In the case of children playing a game of football in the street where damage is caused by the ball being kicked, liability would fall on those responsible for the children, to the degree that the two factors mentioned above are both present: fault in the activity (inherent risk in the game) and collective control of the ball (by the group).

The idea of “group fault” was developed in reaction to numerous decisions in France, as applauded by the Mazeaud brothers,[xxiv] who held that there could be no possibility of compensation for the victim for the injury if there was no proof that the individual conduct of a member of the group caused the result. Individual liability and its requisites, were also fundamental concepts for civil liability. The notion of “faute collective”, as proclaimed by Planiol and Ripert[xxv], was the basis for developing collective liability in France. It helped establish fault where a group avoids taking the necessary precautions to prevent the harm from occurring.

The second ground used to support collective liability is the construct of collective control of the instrument that caused the harm. The group was in control of the arms used by the hunters, just as the stones thrown by young vandals or a football kicked by children in the street. In light of this, the group’s responsibility arose from their on-going obligation to remain vigilant.

We can also classify as a basis for liability what Aberkane perceives as personification of the group. The group is considered to be a “person in fact” for the purposes of attributing liability. The fault of the group, or its collective control, would form the basis for such personification. In order to bestow this “personality” on the group, it is first necessary to decide whether the group can indeed be considered as such, in the acceptance of a group organized for a specific purpose. It must be possible to envisage a collective interest that unites the group’s members. This collective interest, however, should not be mistaken with the purpose of the organization. In other words, there has to be someone who organizes the activities of the group for it to be considered as such.

At the same time, one cannot lump together any haphazard or accidental gathering of people and consider it a group, because in those cases it would be impossible to attribute a personification on the basis of collective interest. In the event of any damages caused by such groups, the victim would have to identify the individual who caused the harm. If this proves an impossible task, the affair would be characterized as merely fortuitous. Accordingly,

[T]he notion of a group requires a collective interest originating in a common will, even if this lacks an organized form. In practice, this requirement is manifested in a common activity that is deliberately pursued and cherished. Two elements: one material, which consists of the activity exercised; the other psychological and subjective, which transcends this activity by dictating the objective that is common to and cherished by all. [xxvi]

Finally, according to Ignacio de Cuevillas Matozzi, collective liability is also based on the inherent risk of the activity, that is, people gathering together to perform a potentially dangerous activity would create a stronger obligation on the part of the group to compensate any damage caused by one of its members. The reason being:

[T]hese collective activities generated damage that can be potentially more serious, seeing as a meeting of people multiplies its power and efficacy, as well as its dangerousness; furthermore, the anonymity that shelters them presents us with what has become known as “fleeing from responsibility”.[xxvii]

iv. Necessary elements for holding a group liable for the harmful conduct of one of its members.

The necessary conditions for group liability are obviously different from those that characterize personal liability, even though they contain the same elements (such as damage, harmful conduct or risky behavior, and causation – albeit in a different sense than that which is typically studied). The first of these elements is that the person who caused the damage belongs to an identifiable group but is not capable of being identified as the sole author of the damage. This is the basic element for applying diffuse causation, because if it were possible to identify the individual whose conduct or actions brought about the damage, then the liability of the group would be negated, along with its characterization as the “guarantor” of indemnification, with the victim now basing his claim in personal liability against the known tortfeasor. For this reason. it has to be shown that all available means used to identify the individual responsible for the damage were exhausted and that the search proved fruitless[xxviii]. It must be stressed that mere indications of who the author of the damage is are insufficient to provide a basis for fault, which is why collective liability is presented where there is some possibility – but no certainty - of establishing the identity of the individual responsible for the conduct in question.

The second element is that the group to which the tortfeasor belongs is participating in an activity that is inherently more dangerous (likely to produce harm). [xxix] Agreement between the various members of the group on carrying out the activity is not necessary. It suffices that the activity be considered a joint endeavour rather than an individual or occasional activity. To put it differently, the formation of the group need not be intentional in the sense of being organized by one of its members or by some third party, but the meeting has to be among people with the same objective and the members must be aware that they belong to the group. If there is an agglomeration, such a meeting, for the purpose of applying the theory, characterizes a group, provided the activity practiced by the group is potentially harmful. However, it should be emphasized that fortuitous agglomeration does not characterize the group for the effect of holding it liable to the extent that the members are unaware of their membership in the group.

The final stage of the analysis permits an identifiable member of the group to be excluded from liability where he can establish the impossibility of his having caused the harm at issue. One example is where the injury was caused by a bullet and the individual did not carry a firearm. Collective and joint liability of the members of an ascertainable group only affects those who fail to definitively establish that they made no causal contribution to the injury. If such proof is absent, the member of the group answers as if he were the individual author of the damage, jointly with the others. In respect of this element, Aguiar Dias very wisely justifies alternative causation as follows:

There is some doubt about whether to accept the conclusion in this concrete case in which the author of the damage remained unidentified. If this is what happened, then it came about because the author of the harm and the other participants in the game remained silent. Although it cannot be denied that the author of the damage is a single person and that he, and he alone, should be held liable by applying the rules of joint liability. It is no less true that the expedient could not predominate, precisely because the silence of the actual author and his companions, after the fact, led to the joint liability, which is employed here in its most crude and least controversial form. (...) Tacitly, it is true, but unquestionably all the players would have expressed their solidarity with the direct author of the damage. [xxx]

The other members of the group who did not cause the damage were left with the real task of identifying the individual to be held liable, insofar as this is presumably easier for them than for the victim. The victim is free to carry out the identification, but this will only be done if the other members of the group show any interest in acting regressively, which rarely occurs, given the notorious corporative spirit of group associations.

To summarize, in order for collective liability to be applied, based on the idea of alternative causation or presumed causation, it is necessary that the following requirements are met, as enumerated by Judith Martins-Costa:

a) anonymity of damage, so called because it is not possible identify the author; b) certainty that the damage was caused by a member or members of a group (for example, the dejected members of ‘organized supporters’ who leave a football stadium after their team has lost a game and decide to damage cars parked in the vicinity); c) the group appears perfectly homogeneous, and its members are known, although it cannot be determined who caused the damage; d) the group does not need to be legally ‘organized’ or ‘incorporated’: it is a ‘de facto group, and can be formed in various ways; e) it is not a matter of damage caused by the group as such, which would characterize plural and personal liability: there is only one, or several individual authors, we are not dealing with co-authorship between all the members of the group; f) the group does not have a ‘leader’ to be held liable, the members being all on equal terms; g) the group engages in activities entailing risk or danger; and h) the actual author of the damage remains unknown, shielded by the group activity making it impossible for the victim to identify him.[xxxi]

Based on these foregoing elements, we shall now proceed to analyze some decisions reached by Brazilian courts.

v. Some Brazilian cases.

Two leading decisions exist in Brazil in respect of alternative causation, which are noted for their contribution to Brazilian case law. Six additional decisions that deal with the topic will also be discussed (five from the Court of Rio Grande do Sul and one from the Superior Court of Justice), which both support and reject alternative causation.

The first of the groundbreaking decisions, dating from 1973, involves the following scenario: several timber companies were hired to cut down trees in a certain area, with a limit being set by contract on the number of trees to be felled. Some of the timber companies exceeded this limit, but it was impossible to identify which of them violated the terms of the contract, since all the firms worked in concert.

At the appellate level, the judge, faced with the impossible task of identifying which company had exceeded the limit, decided that “(...) all the firms must be held at fault. Until such time as it is possible to establish the individual liability of each firm, they must be held jointly liable as per the terms of article 904 of the Civil Code [dated 1916]”. [xxxii] This decision, however, was modified by the STF, which determined that there was insufficient proof regarding causation between the responsible company’s conduct and the damage: “(...) here it could not even be ascertained who ordered more pine trees to be cut down in breach of the established limit, or how many were actually felled. Each of the accused could possibly be liable for its own action, if there was no agreement between them to act jointly, joint liability does not apply due to the difficulty in identifying the party responsible for the breaching the contract”. [xxxiii]

The second appellate decision is better known as the case of “The Hunters’ Float”. [xxxiv] During a parade, one float, called “The Hunters”, carried a group dressed in costumes and carrying rifles to represent the hunters in the region. The guns were loaded with blanks and “BBs”. At a certain moment the hunters were to fire at pigeons that were thrown in the air. One of the BBs shots hit a spectator in the face, causing serious injury. Due to the impossibility of identifying which gun had fired the shot, all those in the float were held liable, because “(...) ‘the hunters’ had the duty of loading their arms or cartridges with chumbo, thereby preventing, in this risky venture, any undue or reckless use of the firearms or ammunition from causing injury to third parties”. [xxxv] According to Vasco Della Giustina, in this case “real causation is clearly replaced by the possibility of causation”. [xxxvi]

In regards to fighting between rival fans, the Superior Court of Justice held the following in regards to the liability of the group: