Sequence 1: the assassination

I begin with Gaitán’s assassination because it is a miniature of the interpretive ambiguities surrounding the Bogotazo more generally. Why did the assassin, a young working-class Bogotan named Juan Roa Sierra, go after the leader of the Liberal Party? His intentions were never disclosed: he left no manifesto, told no one of his plans, and refused to answer the questions of those who had captured him (see testimonies in Alape Reference Alape1983, 239–40). Contemporary commentators explained this mystery away by downgrading Roa to the ancillary role of “material author” of the assassination, and by tentatively assigning its “intellectual authorship” to one of the many political organizations that could have (in theory) benefited from Gaitán’s elimination.Footnote ⁵

Roa’s political associations were vague enough to suit virtually every conceivable conspiracy theory: his relatives described Roa to the police as an unattached client who had offered his services to both the Liberal Gaitán and the Conservative President Mariano Ospina Pérez; as an ardent Liberal who voted for Gaitán in 1946 but became disenchanted with the latter after he lost his bid to the presidency; as an occasional Nazi sympathizer (he had worked with his brother Luis at the German legation during World War Two); and as someone who just didn’t like to talk about politics.Footnote ⁶ In short, Roa had virtually as many political attitudes as he had people to discuss them with. The result was a “shroud of multiple identities” (to steal a phrase from Padgett and Ansell Reference Padgett and Ansell1993: 1310) which neither the people who knew him nor, later, the police and the press could cut through.

Instead of looking for hypothetical co-conspirators, we can reconstruct Roa’s concrete network of relations on the basis of testimonies gathered by the police after the crime (see Figure 1 ). This network presents, at first glance, an unremarkable sample of working-class sociability. Roa’s relations (both direct and indirect) came predominantly from kinship, work, and patronage (from local elites promising and occasionally delivering jobs, money, and other forms of material support, in exchange of loyalty). These three clusters, however, were poorly integrated, only connecting to one another through Roa himself.

Figure 1. The assassin’s social relations.

Remarkably, Roa pulled off a high-profile assassination by turning to his advantage this precarious network structure. Financing, hardware, biographical availability (i.e. freedom from relational constraints, McAdam Reference McAdam1986), and strategic access were all obtained through distant acquaintances, which Roa persuaded his connections to unblock under false pretenses (see Figure 2 )—a deliberate enrollment of weak ties to break out of the involuteness of stronger attachments (Granovetter Reference Granovetter1973).Footnote ⁷ Because each set of relations had a different perception of who Roa was and what his aspirations were or ought to be, but no means of checking this perception against information from other sets, they could easily be fed different stories by and about Roa without their suspicions being aroused. While these stories were globally incompatible, and could thus not yield a durable and consistent social status, they were individually plausible to each cluster of relations, and could therefore yield the latter’s limited cooperation.

Figure 2. Logistics of the assassination.

Admittedly, the foregoing analysis fails to present a motive in the juridical sense. It tells us how, not why, the assassination occurred. But it shows us that this particular how required the elision of an explicit why. For Roa, a practical condition for bringing together the different partitions of his network was to withhold explicit claims to a self-consistent personal identity. How Roa understood this constraint subjectively is both beyond empirical investigation and beside the point. For just what made the latter feasible also made it unintelligible, thereby effectively shifting the responsibility for determining what it meant onto the public that would witness its accomplishment.

Sequence 2: initial gatherings

As he attempted to flee the scene, the assassin was captured by two patrolling policemen. He was hurriedly taken to, and briefly sheltered inside, a drugstore across the street, before a group of men managed to wrench him out, and unceremoniously proceeded to lynch him. His lifeless body was then unclothed, dragged through the streets, and eventually exposed in front of the Presidential Palace (see Braun Reference Braun1985: 135–36). Meanwhile, only a few blocks away, another gathering was forming outside the clinic where a moribund Gaitán had been swiftly translated by his acolytes (see Figure 3 ).

Figure 3. Initial Gatherings (Downtown Bogotá).

Subsequent historical accounts have generally assumed that the “people” of Bogotá were staunchly devoted to Gaitán and that they spontaneously acted out their moral indignation at his death, first against the assassin, and then, unappeased, against the political rival who had beaten him in the 1946 presidential elections (e.g. Medina Reference Medina1984: 73; Braun Reference Braun1985: 135; Pécaut Reference Pécaut1987: 478). Yet this interpretation takes for granted just what needs to be investigated: that Bogotans possessed a ready-made collective identity—as the “people”—and that they were naturally predisposed to acting in concert in response to the assassination because it had obliterated the complementary identity of “leader of the people” (caudillo del pueblo). To be sure, Gaitán was a charismatic leader in the style of Perón and Vargas in Argentina and in Brazil, with a large following amplified by mass media and new repertoires of mobilization (see below). But lest we naturalize his charisma and the devotion of his followers, we need to treat both as emergent properties of a social relationship.Footnote ⁸

A gathering of individuals attending to the same object is a considerable social accomplishment (McPhail Reference McPhail2017). In practice, it requires decoupling participants from ongoing activities that make competing demands for their continued involvement. The Séptima, a busy commercial street, certainly provided a demographic baseline for mobilization. But this datum can be misleading. Braun (Reference Braun1985: 138, 157–58), following a familiar trope, speculates that the mass anonymity of the urban center facilitated the sudden shift from pacific routines to collective violence through a suspension of personal accountability. Yet, while the participants Braun interviewed claimed that they did not recognize familiar faces in the crowd, his own evidence suggests that most of them had in fact arrived to the Séptima as part of small groups of acquaintances. I conclude that while participants did not necessarily identify individuals in the crowd other than those they had come with, few of them were milling alone when Gaitán was shot.Footnote ⁹

Even short of nominal recognition, participants easily identified one another and interacted according to conventional story-sets. We can parse four types of ties in our sample: kinship (connecting local residents to family members); friendship and friendly acquaintanceship (connecting patrons at cafés and restaurants to their friends’ and their friends’ friends); business (between customers and specialized services via storefronts with store names and uniformed employees); and civility (within and between social classes, membership to the latter being signaled by rigid styles of dress). Thus, rather than assembling free-floating individuals into a multitude homogeneously identifying as the “people”, the initial gatherings were produced by sampling from, and activating ties across, a congeries of local cliques (see Figure 4 ).

Figure 4. Communities among witnesses on the Séptima (Max modularity solution).

Further, while the idea of the “people” presupposes unanimity around a common focus, the first protesters were cleft in twain from the very start, precisely because none of them had direct access to Gaitán. On the one hand, a first group of bystanders detached itself when Gaitán was carried off for treatment, walking behind the taxi transporting him, and then staging a vigil outside the clinic where he had been taken. As they could not leave of their own accord without appearing to defect from their vigil, they repeatedly called out the Liberal leaders who had flocked to Gaitán’s side to come out and lead them to the Presidential Palace—a responsibility that Liberal leaders prudently shirked by staying inside. On the other hand, a second group had stayed behind and was pressing against the door of the drugstore where the assassin had been taken, eventually managing to flush him out. Once lynched, the latter became movable, and the lynching mob made itself mobile by the same token. They marched on the Presidential Palace.

Now, the decision to relocate outside the Presidential Palace rather than on the Plaza de Bolívar—the city’s largest public place and the traditional meeting point of all large political protests—is especially noteworthy, because it ostensibly went against habits of popular mobilization. Citizens of Bogotá possessed a long-established right to occupy the Plaza, going back to colonial times (Ortega Ricaurte Reference Ortega Ricaurte1990). There, they would have found a safe, capacious space for public displays of unity, numbers, and commitment (Tilly and Tarrow Reference Tilly and Tarrow2015: 153). Conversely, they would encroach on the narrow street between the Palace and the quarters of the Presidential Guard at their own risks. As most of them at this point were unarmed, they effectively hedged themselves in a physically vulnerable position that they were completely unprepared to hold.

This seemingly poor tactical choice is all the more remarkable that protesters had to pass through the Plaza to get to the Palace, and obstinately resist calls from some participants to occupy the former instead (see testimonies in Alape Reference Alape1983: 249). Yet a demonstration on the Plaza would have required protesters to articulate public claims. For what makes a public place a polyvalent medium for assemblies of all persuasions is the fact that it has no political meaning attached to it other than what its temporary occupants may bring with them (e.g. Bayat Reference Bayat2013: 175ss; Said Reference Said2015). In order to occupy the Plaza, protesters would have had to specify, at the very least, who they blamed for Gaitán’s assassination and what sort of political reshuffling the vacancy suddenly left by his death should trigger.

The obstacle, in the event, was not that protesters lacked answers to these questions: rumors had already begun to circulate, and were soon echoed on the radio, that Gaitán had been shot by a Conservative policeman (or a delivery boy from the Conservative newspaper El Siglo) and that President Ospina would most likely resign in response (or had already been hanged) (Alape Reference Alape1983: 256; Braun Reference Braun1985: 146; see also Williford Reference Williford, Augusto, Diago, Javier, Osorio, Alberto and Villalobos2009). Converting unsubstantiated rumors into public claims, however, would have laid the protesters open to the risk of discredit and the threat of governmental retaliation, should the rumors prove untrue—which they did. To paraphrase J. Markoff (Reference Markoff2010: 243–44), it was much safer to riot on behalf of rumors than to assembly in support of an openly expressed and rationally developed critique of government policy, because the first move redirected the threat of governmental repression onto those who sowed the rumors, while protecting the rioters themselves from charges of sedition.

By dragging the assassin’s cadaver to the doorsteps of the Presidential Palace, and then attempting to crucify him to the gates, the protesters forcefully transferred interpretive responsibility for Gaitán’s assassination and its political implications to public authorities. They dared the Conservative Government to reclaim the cadaver, and thus to reveal themselves to have been behind the assassination, at the same time as they addressed them a veiled threat of meeting a fate similar to the assassin’s if they did. But nobody ever came out of the Palace to remove or at least cover the naked corpse, which remained in the same position throughout the afternoon (Braun Reference Braun1985: 157).

Sequence 3: sustained collective violence

The standoff was eventually broken when one protester surged at a soldier, wrested his rifle, and took aim at the Palace’s windows. Returning fire, the Presidential Guard swiftly drove the protesters away. As they folded back on the Plaza de Bólivar, the protesters merged with new clusters of participants drawn primarily from neighboring blocks. Scattered attacks against governmental buildings and military barrages, accompanied by the ransacking of police stations and hardware stores for weapons, gradually gave way to the coordinated destruction of administrative buildings, Conservative centers, churches, and ecclesiastical institutions around the Plaza. The protest then seemingly devolved into a moral holiday, marked by the smashing of private cars, streetcars, and high-street stores, the liberation of prisoners, generalized looting, public sales of seized goods below market price, and festive drinking. Despite heavy rainfall, choking smoke, and the threat of stray bullets from ambushed snipers, street action continued throughout the night, until military reinforcements retook control of the city center sometime before dawn (for a movielike reconstruction, see Braun Reference Braun1985: 155–66).

I listed at the outset of this article the many disparate historical interpretations that the looting and destruction of the city center have elicited, and I noted that all these interpretations seem reasonably plausible. But that’s the rub: they all do. Which is the most plausible is a moot point, because plausibility is conditional on the type of targets that the observer choses to highlight: governmental buildings, ecclesiastical institutions, Conservative symbols, or commercial establishments and merchandizes. Conflicting interpretations naturally arise because the protesters aimed their attacks at a variety of targets without ever formally stating what they wanted from the political, religious, and economic elites who occupied these premises.

Taking advantage of Bogotá’s grid layout, we can economically visualize the Bogotazo as a network of targets. Each node in Figure 5 represents a city block containing at least one attacked (i.e. looted, torched, or both) building; an edge is drawn between every targeted block and any of its eight nearest neighbors if these also contained a target, and if the latter could be accessed by a straight walk (i.e. edges were omitted if a turn was required to go from one target to the other). While a map (e.g. Medina Reference Medina1984: 72) gives a misleading impression that the riots concentrated around key governmental and ecclesiastical buildings, because a planar representation tends to emphasize sheer physical distances, a network visualization allows us to take into account walkability. The result is that the most symbolic targets were in their majority isolated from the rest of the network, presumably because they were quickly shielded by military barrages, and because the largest share of participants arrived when collective action had already shifted to looting.

Figure 5. Looting and destruction during the Bogotazo.

Note: Governmental and ecclesiastical buildings have been shaded.

Since target selection responded so massively to dynamic constraints, the very project of supplying a post hoc interpretation of the Bogotazo on the basis of what this or that target might have meant to some imaginary protester is, arguably, wrongheaded. Looking for what the protesters did not say but could have, directs our attention away from what they actually did—which did not include, in the event, the sort of repertoire needed to broadcast definite claims to an attentive audience.

For contrast, consider the remarkably elaborate repertoire of mobilization that the Gaitanist movement had developed through countless rallies and marches during the recent electoral campaigns of 1945 and 1947 (Sharpless Reference Sharpless1978: 147–71; Braun Reference Braun1985: 91–129; Green Reference Green2003: 233ss). This repertoire had been most recently perfected with the Manifestación del Silencio of February 1948, called to denounce Conservative violence against Liberals in the countryside (see Table 1 ), often regarded as the most successful rally of Gaitán’s career (see Alape Reference Alape1983, 103–8). Events such as the Manifestación del Silencio were often planned weeks and sometimes months in advance. Drafting participants from distant neighborhoods, they carefully funneled them onto the Séptima, and then typically routed them to the Plaza de Bólivar. Strong limitations were placed on collective expression, either through mandatory silence or preselected banners, flags, and chants: Gaitán was meant to be the only official speaker. Crucially, the marches avoided direct contact with potential targets (in particular the Presidential Palace), lest unplanned surges should break marching order. At the Plaza, orderly collective occupation of public space was achieved by assigning each delegation to a specific location around the tribune where Gaitán would solemnly address protesters as his personal audience. Once the tribune had concluded his harangue, he would order the protesters to disperse peacefully.

Table 1. The Manifestación del Silencio and the Bogotazo

This configuration was turned on its head during the Bogotazo. Although some groups did initially parade in the streets with flags and banners from past mobilizations that did not quite match the occasion, most adopted a very different repertoire of contention. At such short notice, participants were chiefly drafted from locally available populations, with little immediate reinforcements from the rest of the city. Stochastic arrivals and departures caused considerable congestion. Locations in public space were constantly shifting, dictated by the tasks at hand, rather than fixed by categorical affiliation to a neighborhood committee. Along the way, almost all available targets were directly attacked, without explication de texte, because no banner or slogan had been preapproved. Instead, protesters shouted, cursed, and repeated wild rumors (for example, that the President and the leader of the Conservative party had been hanged). Although some Liberal leaders propped themselves up on streetcars and attempted to harangue the crowd, nobody paid them attention. No staged oratory performance captured the common focus, or produced an official interpretation on behalf of the gathering. Without ceremonial closure, dispersion was only brought about by a combination of endogenous decay and forced removal.

Yet despite initial appearances, the fundamental contrast between the two configurations is not between tidy organization and mayhem, but between two types of ordered sets. On the one hand, the linear ordering contrived by local Gaitanist committees and national party cadres is remarkable because it built on the simplest of order relations—with every participant occupying a structurally equivalent position of subordination to Gaitán. This pristine organization, however, required social organizers to remain in the background and to shunt distracting targets off the stage. On the other hand, and paradoxically, a thicket of confusion grew around the protesters on April 9 because they made no effort to conceal either their sources of supply or their targets. The resulting phenomenology was only dizzying because supplies and targets were chained together by looting and destroying establishments across several broad equivalence classes. Thus, supplies were drawn from “hardware stores” (makeshift weapons), “police stations” (firearms), and “high-street stores” (liquor, flags, and clothes), and targets were obtained from abandoned vehicles that could be attacked without weapons (cars and streetcars), protected buildings that could be stormed or fired at with weapons (e.g. “governmental buildings”), and undefended buildings that could just be destroyed (i.e. “governmental allies”, including the Church and the Conservative party).

We can easily represent these chains as forming a partially ordered set and contrast them with the set of actions during the Manifestación del Silencio (see Figure 6 ). The difference between the two sets, we may observe, is merely that the first only admits one total ordering, whereas the second is compatible with multiple possibilities. In short, contrary to what contemporary political observers hastily concluded (see Alape Reference Alape1983, 605–19), protesters did not run amok in the absence of competent organizational supervision. While we cannot tease out a unified political meaning of the street riots, we need not conclude that they lacked structure altogether. Protesters could not have sustained grueling actions for hours on end without inducing at least a few topological orderings. But these, in turn, merely followed from chaining together actions that might prompt a response from elites who could, and just might, authoritatively impose meaning on the situation.

Figure 6. Hasse diagrams for the Manifestación del Silencio and the Bogotazo.

Sequence 4: broadcasting the revolution

The protesters prompted a first, immediate response from a self-selected group of kibitzers. Small numbers of writers, journalists, academics, and students rushed to occupy all major radio stations in the city. They then went on air to announce that they were constituting themselves as “revolutionary juntas” and taking the reins of what they reframed as a revolution of the Colombian people (pueblo, revolución, and Colombia are the three most frequent words in our corpus, with 81, 51, and 40 occurrences respectively) (see Table 2 ). Yet, precisely because the protesters’ actions were not directed at them, the revolutionary juntas could only present themselves as the legitimate interpreters of the popular will by keeping up the fiction that these actions obeyed their explicit directives (e.g. the militaristic cry “A la carga!”, a trope borrowed from Gaitanist rhetoric, occurs 11 times in our corpus). This fiction, however, rapidly lost all plausibility as contradictions between street action and the narrative spewed on the radio cropped up.

Table 2. Most frequent nouns and adjectives in the Revolutionary Juntas corpus

First, though the narrative of a vanguard conducting the insurrection at a distance was nominally addressed to the protesters (repeatedly addressed as pueblo, movimiento, fuerzas [revolucionarias], compañeros, and calle), it was in practice heard by static listeners who were not engaged in street action. Thus, although the radios could not directly influence the behavior of actors on the ground, they could nevertheless generate expectations among listeners, which were at risk of being belied by the former. In a telling example, the radios announced that the army had taken sides with the insurgents and sent three tanks against the Presidential Palace. Listeners inside the Palace, including the President, initially succumbed to panic. Yet upon reaching their destination, soldiers inside the tanks revealed themselves to be loyalist troops (see the multiple testimonies collected in Alape Reference Alape1983: 278–80, 336–37, 340–48, 365–66). The radios also announced that the assassin was a Conservative policeman, that Laureano Gómez, President Ospina, and other prominent Conservatives had been lynched, that former Liberal president Eduardo Santos was returning from New York to assume the presidency, and that the Gaitanist Darío Echandía would assume power in the interim—all claims that would later be retracted.

Second, as the juntas had no means of enforcing the peremptory orders they issued on the air, the façade of authority that they attempted to present was vulnerable to popular actions that were formally inconsistent with the latter. Looting and festive drinking, in particular, alarmed the revolutionary juntas. The self-appointed president of the Comité Ejecutivo de la Junta Revolucionaria de Gobierno, Adán Andrade threatened the looters that they would be tried in a (made-up) Consejo Revolucionario (see testimony in Alape Reference Alape1983: 442). To be sure, these actions could have been narratively reconciled with other political interpretations of the situation. For example, looting would later be glossed as xenophobic reprisals against foreign economic interests (Sowell Reference Sowell1998) and festive drinking as a cross-class ritual of fraternization (Braun Reference Braun1985). But none of these alternatives was consistent with the rhetoric of disciplined masses obeying the direct orders of a revolutionary vanguard in pursuit of overarching political goals.

Interpretive incongruence was inevitable, third, because the revolutionary juntas could not move the narrative of a political revolution without the strategic cooperation of at least some state agents and political elites. These were needed to catalyze street action directly into state structures, force a removal of the President, and install a new government. The radios rushed to announce that the Liberal party, the army, and the police had joined the revolution against the Government (Liberal, gobierno, policia, and ejercito are all among the 10 most frequent words in our corpus). Yet despite some early individual defections,Footnote ¹⁰ most law enforcement and military forces in Bogotá adopted noncommittal lines of conduct throughout the afternoon, waiting for orders that cagey officials would not give. As a result, the narrative which was being looped on the radios of a triumphant offensive against and inside state structures appeared increasingly disconnected from the direction that street action was taking, ever further away from an assault on the Palace.

While it did little more than provide an increasingly out-of-touch commentary of the protests, the juntas’ precipitated claim-making on April 9 would retrospectively allow the Government to lay blame for the Bogotazo at the door of the Communist Party and the left-wing of the Liberal Party. Not unreasonably, historians (Aprile-Ginset Reference Aprile-Gniset1983; Sánchez Gómez Reference Sánchez Gómez1984; Braun Reference Braun1985; Martz Reference Martz2012 [1962]) have maintained that communists were opportunistically scapegoated by politicians who were eager to save face with international observers congregated in Bogotá for the Pan-American Conference. But regardless of the hidden intentions we may critically impute to the government, the fact is that when political observers searched for actors who had made explicit political claims during the Bogotazo, the only that could be found were the Leftist intellectuals and students who had repeatedly and emphatically called for an armed revolution on the radios.

Thus, although the board-game theory of revolutionary action advises to capture means of mass communication to establish a monopoly on the public interpretation of the situation, this can also be a liability, as actors who broadcast public claims become lastingly accountable for them. Conversely, the government quickly ordered a gag ruling on medias after Gaitán’s assassination. Even after the Army had regained control over most radio stations, a formal allocution by the President was not put out on the air until April 11, when the situation had definitely turned in favor of pro-governmental forces. More generally, we go on to see, the behavior of political elites in both the Government and the Opposition throughout the events was almost cautious to a fault.

Palace negotiations

The Liberal leaders who had gathered inside the Clínica Central delayed a public announcement of Gaitán’s death. Despite being called out by the crowd, they did not move from inside the clinic until they received an invitation from the President’s staff to meet with Ospina. They made a prolonged stop on their way to the Presidential Palace to ditch the cortege of protesters that accompanied them and to discuss strategy. Once at the Palace, they engaged in protracted negotiations with the Conservative President. When they emerged out of the Palace the next day, they had agreed to join a government of national unity led by Ospina, to back the President’s decision to declare a State of Emergency, and to let the military in charge of restoring public order.

Contemporary commentaries and retrospective history alike have often commended President Ospina for his ability to temporize a decision against mounting pressures, and criticized Liberal leaders for their unwillingness to pounce on the opportunity to seize power (e.g. Sharpless Reference Sharpless1978: 179; Sánchez Gómez Reference Sánchez Gómez1984: 22). The Government and the Opposition are thus respectively praised and blamed for adopting the exact same attitude—namely, deferring the articulation of an official position—during the Bogotazo. This seems to be a classic instance of asymmetrical explanation: the same variable is interpreted differently depending on whether we adopt the perspective of the winning or the losing side. Instead of siding with the victor, we should ask two questions: Why did both sides operate on the theory that they should stave off a formal claim to power? Why was one side able to outlast the other?

The eventual asymmetry of positions is all the more remarkable that the President and Liberal leaders met in an initially level playing field. Both sides faced equally intense levels of “environmental impatience” (Gibson Reference Gibson2011). Not only were external events impinging on the negotiations (sometimes directly, as when a stray bullet lodged itself into a wall of the conference room), but both sides were furthermore under threat of being overran by their extremes. On the one hand, the Joint Chiefs of Staff and the parliamentary wing of the Conservative Party (led by the ultraconservative Laureano Gómez) initially pushed for a military government.Footnote ¹¹ On the other hand, the revolutionary juntas called for an insurgent government. These two variables reinforced each other, for once inside the palace, both sides were physically cut out from the outside; both were accordingly dependent on extremely unreliable radio and phone communications with their impatient allies for information on the evolution of the situation on the street. A game of patience was therefore equally uncertain and perilous for both sides.

Second, and at the same time, face-to-face negotiations were inescapable because neither side had a free hand. The Opposition could not lawfully seize power as long as President Ospina was in place; they therefore needed him to tender his formal resignation to create a legal vacancy that would justify the designation of an interim President from the Liberal party. Similarly, President Ospina needed to secure the Opposition’s support because he could not lawfully declare a State of Emergency without the cooperation of the Legislative and the Judicial branch of government, which were controlled by Liberal majorities. Hence the minimalistic terms of the negotiations: each side merely pressed the adversary to redefine the situation in a way that would then allow them to make a strong claim to power. As any of these two options required that the other be formally rescinded, formal negotiations were at an impasse before even starting.

To understand how this deadlock was eventually broken, we must, first, call attention to the checkered pattern of the negotiations. Discussions between the President and leaders of the Opposition were constantly interrupted. Interruptions occurred because Ospina was simultaneously engaged in several other conversation networks—with military advisers, members of his government and his party, his personal staff and private entourage, and outside informants (including Laureano Gómez)—that made requests on him in a stochastic fashion, as fresh news arrived that urgently needed to be discussed. Liberals, too, were occasionally approached by individual participants in the President’s conversation networks, and they had their own contacts with outside informants through the telephone. Instead of using the Presidential palace to centralize all participants inside a unique “situation room”, President Ospina freely circulated between them, and no less significantly authorized others to interact both within and outside their respective circle without his being present to supervise their exchanges (see Figure 7 ).

Figure 7. Conversations inside the Palace.

Despite gaps in our timeline and imperfect data on the specifics of verbal behavior, we can identify recurring patterns within conversations by looking at the fairly limited set of topics broached by participants. Table 3 summarizes the timeline of events.Footnote ¹² Most strikingly, sustained attention to explicit proposals never occurred until the final round of negotiations. Prior to this point, whenever explicit proposals were hazarded by a participant, other participants invariably withdrew their support, and the topic was immediately dropped (e.g. 2→3, 16→17, 56→58, 63→64). Instead, each side repeatedly attempted to coax or extort indications of personal disposition and political position from the other. Participants thus attempted to prompt a response with threats to abandon the negotiations (e.g. 32, 44, 50, 73), pressures on external actors (e.g. 34, 40, 68), emphasis on time constraints and environmental pressures (e.g. 5, 7, 12, 18 and 4, 22→23→24, 52, 54), turn-transfer—i.e. asking somebody else to speak—and role-transfer—assessing hypothetical scenarios from somebody else’s point of view (e.g. 6, 8, 10, 19, 28, 50, 57 and 20, 25→26, 57, 72).

Table 3. Timeline of Meetings and Events Inside the Palace

These two patterns of interaction—conversational decentralization and indirections—allowed participants to engage in a continuous process of “Bayesian updating” (Mische and White Reference Mische and White1998) on both external events and internal interactions. First, circulation between conversational networks, in the long periods of intermission between talks, created frequent opportunities to gage the disposition of other participants and exchange news from the outside. Second, during talks, conversational indirections (for instance when an actor asked a direct question refuses to answer by keeping silent or by changing the topic) successively led to the revelation that Darío Echandía was reluctant to assume power (16→17), that President Ospina would not resign (25→26→27→28→29; 57), that the generals were unwilling to push for a military government without political support from at least one of the two main parties, which neither was willing to grant (58, 64), that the revolutionary juntas would not attempt an all-out assault on the Palace (40→41), that the Liberal delegates would not leave the Palace without a political solution (32, 44, 50, 73), and that a bipartisan cabinet might be acceptable to both Conservatives and Liberals (70→71→72). Only then did Ospina make his move (73).

The observed outcome therefore gradually emerged from, but required the temporal dilation of, the negotiations (compare with Gibson Reference Gibson2011: 404–5). Participants who were the most mobile between conversation groups throughout these protracted negotiations were more often exposed to fresher information, and thereby acquired frequent occasions to update their perception of shifting circumstances and dispositions. And none was more active than President Ospina. Yet, far from being the product of deliberate planning or political savviness, Ospina’s centrality in an otherwise decentralized conversational network merely followed from his being simultaneously the frequent addressee of urgent news, in his official capacity as President, and the (occasionally reluctant) addresser of repeated marks of attention to all his guests, in his courteous role as host.