Silence has a story¶

On information voids, the narratives that fill them, and why the damage outlasts the failure

The clacks tower on Treacle Mine Road goes dark at three in the morning. The fault is a seized mechanism and one absent operator. Under normal circumstances, this is three hours of work: find the operator, send the engineer, turn the mechanism, restore the signal. The tower will be running by six. Nobody will notice anything happened.

This is not normal circumstances. By seven, the Watch patrol at Pseudopolis Yard cannot reach the duty officer on Filigree Street. By nine, the Guild of Merchants has sent four messages that have received no reply. By noon, the taverns are running their own investigations. The leading theory involves plague. The second most popular involves a coup, or possibly a counter-coup, accounts differ. The Unseen University appears in several versions, as it tends to when something is wrong and nobody has a better explanation. The actual fault is not in any of these theories. But the actual fault is also not what is being discussed, because the actual fault is unknown. The void left by the absent explanation has been filled by the explanations that were available, which are the explanations that fit what people already feared might be true.

By the time the tower is restored at mid-afternoon and an official account of the fault reaches the guild halls and watch posts, the narrative has been running for nine hours. The official account is accurate. It is also, at this point, one story among several, and it has the significant disadvantage of being less interesting than its competitors. The damage to trust in the clacks service, in the Watch coordination it supports, and in the Patrician’s administration that is responsible for both, is not proportional to a seized mechanism and a morning’s disruption. It is proportional to a nine-hour absence of official explanation during which the city imagined the worst.

The fundamentals¶

The information void amplifier is a reinforcing loop with a simple structure. A failure occurs. Official communication is absent, delayed, or insufficient. A void forms in public information about what has happened, why, and what is being done about it. Rumour fills the void. Rumour travels faster than official communication under most conditions, is not constrained by the need for accuracy, and is shaped by the fears and prior beliefs of the people generating and transmitting it. The narrative that hardens in the void is not random. It is the most resonant narrative available, which in most cases is the one most consistent with existing distrust of the institution responsible.

When the official explanation eventually arrives, it does not enter a neutral information environment. It enters an environment already structured by the narrative that formed in its absence. Correcting an established narrative is harder than establishing a new one. The official account must now compete with an account that has been circulating for hours, that has been repeated enough to feel familiar, and that has the additional advantage of explaining why the official account might not be the whole truth. The institution is now in the position of arguing against a story that its own silence helped to create.

The trust damage that results from this sequence is not proportional to the technical severity of the original failure. It is proportional to the length of the silence, the range of interpretations the silence permitted, and the degree to which those interpretations resonated with pre-existing concerns about the institution. A severe failure communicated immediately causes less long-term trust damage than a minor failure surrounded by hours of official absence.

Why voids fill badly¶

Information voids do not fill randomly. The narratives that form in the absence of official explanation follow a consistent pattern that makes the void more damaging than a random sample of possible explanations would be.

The first element is selection for resonance. People filling an information void draw on what they already know and fear. A city that has recent experience of disease outbreaks reaches for plague before it reaches for seized mechanisms. A population that distrusts its government reaches for cover-up before it reaches for technical fault. The explanations that fill the void are not a neutral survey of possibilities; they are weighted heavily towards the explanations that confirm existing anxieties.

The second element is selection for concealment. In the absence of information from the institution responsible, the silence itself becomes evidence. Why is there no statement? The most emotionally compelling answer is that there is something to conceal. An explanation that includes institutional concealment as a feature is more resistant to correction by the institution later, because any correction can be framed as further concealment. The void creates an interpretive frame in which official communication, when it eventually arrives, is already suspect.

The third element is competition asymmetry. Rumour travels through informal channels that are faster and more distributed than official communication. By the time an official statement is prepared, verified, approved, and distributed, the rumour has made several circuits of the taverns, the guild halls, and the street corners. The statement must achieve, in a single transmission, what the rumour has achieved through dozens of repetitions. Repetition is how information becomes felt truth. The official statement is always playing catch-up, and catching up requires more than a single accurate account.

Deutsche Telekom and the BSI contrast¶

In November 2016, a variant of the Mirai botnet exploited a vulnerability in Deutsche Telekom home routers and took approximately 900,000 DSL connections offline. The attack was active over several days. During this period, the information void around the outage filled across social media with a sequence of competing explanations: equipment fault, then cyberattack, then foreign state actor involvement, then back to cyberattack. Each version spread faster than the previous one, partly because each new explanation incorporated the failure of the previous one to gain authoritative confirmation.

Deutsche Telekom’s own communications were slow and technically opaque. The company confirmed an outage and indicated it was investigating, but provided neither a clear technical account nor a timeline for resolution for several days. The void remained open. The narrative that hardened was not the accurate one, a specific vulnerability in a specific router management protocol, exploited by a criminal botnet operator, but a generalised account of DT as an insecure provider unable to protect its customers or communicate honestly about what had happened to them.

The German Federal Office for Information Security, the BSI, communicated differently. Within 48 hours of the incident becoming publicly visible, the BSI had issued an advisory that named the specific vulnerability, the TR-069 remote management protocol, provided a workaround that any router owner could apply immediately, and gave a timeframe for a permanent fix from the router manufacturer. The BSI advisory did not require the full facts to be established before communication could begin. It required only the facts that were available: what was known, what was being done about it, and when more would be known.

The narrative around the BSI’s response stabilised. The narrative around Deutsche Telekom’s response did not. The outage lasted three days. The reputational damage from Deutsche Telekom’s communication handling ran for weeks. The technical fault was resolved; the information void damage was not, because by the time DT’s communication improved, the void had already produced its narrative.

• More than 900k routers of Deutsche Telekom German users went offline, Pierluigi Paganini, November 28, 2016

• Massive Störung bei der Telekom, 28.11.2016

• Global cyberattack caused Telekom Deutschland router outage, 29 NOV 2016

France and the mask communication¶

In the early months of 2020, French public health authorities communicated that face masks were not recommended for the public and were not effective at preventing transmission of respiratory illness in non-clinical settings. This communication was consistent with contemporaneous WHO guidance and reflected a genuine state of scientific uncertainty about masks in community settings.

It also coincided with a severe shortage of masks in French public health stockpiles, which had been depleted and not replaced in the years preceding the pandemic. Whether the mask guidance was driven primarily by the scientific evidence or primarily by the supply constraint was not communicated. The gap between what the authorities knew about supply and what they communicated was not addressed directly. An information void opened between the official message and the question that a significant portion of the population was asking: is this guidance about masks really about evidence, or is it about shortage management?

The void filled with the more resonant explanation: that the guidance was supply-driven, that the government knew masks worked and was discouraging their use to prevent panic-buying ahead of a controlled distribution to healthcare workers. This explanation was partially accurate, in the sense that supply considerations were part of the context, and was not officially confirmed or denied. When mask guidance reversed as supply improved, the earlier communication became retrospectively framed as deliberate misdirection rather than as updated guidance based on improved evidence and improved supply.

The trust damage was not primarily to the individual health officials who issued the guidance. It was to the credibility of official public health communication in France as a structural model. Subsequent public health messages, about booster doses, about ventilation guidance, about the epidemiology of variants, were received in an environment already shaped by the mask communication episode. The void that formed around a supply constraint in early 2020 contributed to a background distrust of official health guidance that persisted through subsequent phases of the pandemic and affected compliance with measures that had nothing to do with masks.

The void had been allowed to fill with an interpretation that was partially accurate and fully corrosive, because the accurate account, which would have acknowledged supply constraints directly alongside the evidence uncertainty, was not communicated in the window when it would have shaped the narrative rather than correcting it.

• Chronique d’un « mensonge ». Déclarations gouvernementales sur l’« inutilité » du port du masque en « population générale » et réactions indignées du public, Gérald Gaglio, Cedric Calvignac, Franck Cochoy

• Man or woman? Young or old? What is the profile of the average French ‘anti-masker’? Daniel Harries, 10-Sep-2020

The feedback that deepens the silence¶

The information void amplifier has a second loop within it that is less often discussed. Institutions that have experienced narrative attacks in previous crises, in which communication during the event produced hostile coverage that complicated the response, develop an instinct toward communication caution. Legal review, management sign-off, coordination with political offices, verification of all facts before any are released: these processes exist for legitimate reasons, and they add hours to the window between a failure and an official account. Those hours are the void. The void that forms in those hours produces the narrative damage. The narrative damage produces the lesson that communication is dangerous. The lesson produces more caution in the next crisis. The next void is wider.

This loop does not run through external forces. It runs through institutional self-protection. The organisation experiencing it is not ignoring the value of communication. It is prioritising protection against the specific risk that communication under uncertainty produces incorrect statements that compound the crisis. This is a real risk. The loop does not resolve it. It trades the risk of a communication error for the certainty of a narrative void. The narrative void is, in most documented cases, the more expensive outcome. But it is an expense that arrives diffusely, measured in trust metrics and political relationships, while the risk of a communication error is specific, measurable, and easily attributed. Institutional incentives protect against the specific, attributable risk. They do not protect against the diffuse, background cost.

The asymmetry of establishment and correction¶

There is a structural asymmetry between establishing a narrative and correcting one that makes the timing of communication critical in a way that the internal communication review process does not account for.

A narrative established in a void is established in a condition of low prior information. It only needs to be consistent with the observable facts, which are limited to: something went wrong, no one has explained it, and the institution responsible has not communicated. A narrative consistent with these facts has a wide design space. Any explanation that fits “something went wrong, and they’re not saying why” is a candidate.

A correction, issued hours later, must operate in a condition of high prior information: the correction must be consistent with the already- established narrative, already emotionally invested-in by the people who believe it, already repeated enough to feel confirmed rather than merely rumoured. The correction has a much smaller design space. It must be accurate, it must be specific, and it must explain not only what happened but why the institution failed to communicate it earlier, because the delay itself has become part of the narrative.

This asymmetry does not mean that communication at hour nine is useless. It means that communication at hour one is worth several times as much as communication at hour nine, and that the hours between one and nine are not neutral time during which the technical team can work undisturbed. They are the window in which the primary damage of the event is being done, through a mechanism that the technical team’s work will not address.

The tower and the tavern¶

When the Treacle Mine Road clacks tower is restored at mid-afternoon and the official account of the seized mechanism reaches the guild halls, it is accurate, it is detailed, and it is about nine hours too late. The taverns have had the morning to work with. The theories that formed in those nine hours did not need to be true; they only needed to be compelling, and they had the significant structural advantage of forming in a space where no competing account existed to challenge them.

The fault was three hours of work. The silence around the fault was nine hours of narrative. The three hours will be forgotten. The nine hours will be recalled the next time the clacks tower dims, as evidence that something more than a mechanical fault is worth considering.

The seized mechanism cost the city an engineer and a morning. The silence around the seized mechanism cost considerably more, in the currency that cities run on and that is hardest to recover once spent.

That currency is not gold. It is the assumption that when something goes wrong, the people responsible will tell you what it was.