The Zero-Latency Threshold

Before you read this as theory, read it as a debt. Coombs gave us Situational Crisis Communication Theory in 1995. Benoit gave us Image Repair Theory in 1997. Jin, Austin, and Liu gave us the Social-Mediated Crisis Communication Model in 2010. Three frameworks, three decades of empirical validation, and not one of them was built for a country where the government can turn off the internet on a Tuesday, where opinion is manufactured at industrial scale before a single citizen opens their app, and where half the population learned to use a smartphone faster than they learned to distrust one. This is not a rejection of that work. It is what happens when you take it seriously enough to notice where it stops.

Why Crisis Communication Theory Was Built for a World That No Longer Exists

Let me start with a number that has nothing to do with public relations.

10 milliseconds.

That is roughly how long it takes a signal to travel 1,000 kilometers through fiber optic cable. Not a metaphor. A physical constant, set by the refractive index of glass, unmovable by budget, regulation, or good intentions. New York to London, a round trip is about 59 milliseconds. Jakarta to anywhere that matters - a similar order of magnitude.

Here is the number that matters more. The average institution takes hours, sometimes days, to issue a verified, board-approved public response to a crisis.

Between those two numbers sits every crisis communication theory ever written, and all of them are losing ground, every year, to a gap that is not closing. It is not closing because we do not want it to close. We are building the infrastructure that widens it on purpose, because faster internet is genuinely good for almost everything else humanity does. Nobody is going to slow down 5G rollout, submarine cable expansion, or low-orbit satellite coverage to give corporate comms teams more breathing room. Nor should they.

So the real question was never "how do we respond faster." It was "what do we do about a race we were never designed to win."

More from Rochman Maarif →

What the Old Models Got Right, and Where They Stop

Three frameworks currently anchor crisis communication as a discipline, and each of them deserves to be named honestly before I argue past them.

Theory 1, by W. Timothy Coombs

Situational Crisis Communication Theory, developed by W. Timothy Coombs, argues that an organization's response strategy should match the level of responsibility the public attributes to it. Simple, durable, still correct as far as it goes.

Theory 2, by William Benoit

Image Repair Theory, from William Benoit, catalogues the rhetorical strategies organizations use to defend their image under attack, denial, evasion, bolstering, mortification. Also correct. Also still taught in nearly every graduate communications program on earth.

Theory 3, by Yan Jin

The Social-Mediated Crisis Communication Model, built by Yan Jin, Lucinda Austin, and Brooke Fisher Liu starting in 2010, extended Coombs's work to account for social media, splitting the public into creators, followers, and observers of crisis information.

All three share one silent assumption that a crisis signal, once it exists, is eventually visible and eventually traceable. That assumption was reasonable in 2010. It stopped being reasonable the moment WhatsApp, Facebook Messanger, and Telegram became the primary channels for information distribution in countries like Indonesia, and it stopped being reasonable a second time the moment large language models began treating search results as permanent training material.

Neither gap is a criticism of the original authors. Nobody in 2010 was supposed to predict retrieval-augmented generation. But a theory that cannot see two of the three phases a modern crisis actually moves through is not a complete theory anymore. It is a very good description of the middle third.

The Three Velocities

Here is the extension, stated plainly.

Dark Velocity. The phase before anyone with a sentiment dashboard notices anything. A screenshot moves through a closed WhatsApp group. A forward chain runs through Telegram. No hashtag, no public API, no metadata trail. By the time a crisis "starts" on an open platform, it has usually already been incubating, invisibly, for hours or days, in a space no monitoring tool was built to see.

Open Velocity. The phase the SMCC model actually describes well, the moment a story moves onto X, Threads, TikTok, and becomes measurable. The difference from 2010: the feedback loop is now algorithmic, not manual. A surge of angry comments is not just people sharing an opinion, it is an engagement signal the recommendation engine reads as "push this further." What used to accelerate linearly now accelerates exponentially, and the response window that used to be measured in days is now measured in hours, sometimes minutes.

Permanent Velocity. The phase none of the three classical frameworks anticipated, because all three predate the idea that a crisis could become training data. Once the story fades from the timeline, it does not end. It changes form, from a live event into a permanent entry in a search index and in the corpus a language model has learned from. Unlike the first two velocities, this one barely decays. It resurfaces every time someone searches the organization's name, for years, unless someone deliberately architects the content record to prevent it.

The sequencing matters more than the existence of each layer. A textbook-perfect apology fired at Open Velocity is wasted if the credibility problem actually originated in a Dark Velocity blind spot, and it becomes permanently indexed evidence of the response itself at Permanent Velocity, oddities included.

The Layer Beneath the Layers

This is where the theory needed to go somewhere Coombs, Benoit, and even Jin, Austin, and Liu never had to go, because none of them were writing about Indonesia.

Ask a communications theorist in Helsinki what happens when a crisis breaks, and the assumptions built into their mental model include a state that does not switch off the internet, a citizenry with among the highest media literacy scores on earth, and a media ecosystem where anonymity is the exception, not the operating norm. None of those assumptions hold in Indonesia, and pretending otherwise is not caution, it is bad modeling.

Five structural conditions sit underneath every crisis in Indonesia, conditions that do not appear in a single Western crisis communication textbook because they were not written by anyone who had to live inside them.

One, state asymmetry

The government does not just respond to crises, it can change the battlefield mid-crisis. SAFEnet has documented a pattern going back years, internet and social media access cut during unrest around Bawaslu, in Papua, and in Wamena in 2019, a pattern a SAFEnet director has called a routine tool for managing social conflict rather than an emergency exception. The pattern is current, not historical. In February 2026, Komdigi unilaterally blocked the domain auth.wikimedia.org, crippling login access for thousands of Indonesian Wikipedia contributors, and the block remained in place weeks later. A Jakarta administrative court has already ruled a prior government internet blackout in Papua unlawful. The ruling changed nothing about the pattern.

Two, an industrialized opinion economy

Academic research on Indonesia's domestic influence operations describes a structure that goes well beyond bots, coordinators, content creators, and paid influencers working in concert, a pattern consistent with the Oxford Internet Institute's finding that government agencies in dozens of countries now use computational propaganda as standard practice. Indonesian buzzer operations trace back to 2009 as commercial marketing tools before pivoting to political weaponization around the 2012 Jakarta gubernatorial race. By the time a crisis becomes visible on the open web, the sentiment reflecting it back may already be manufactured, not organic.

Three, uneven digital literacy

Indonesia's national digital literacy index has moved from roughly 3.46 in 2020 to 3.54 in 2022, on a five-point scale, with the rate of improvement slowing rather than accelerating. The weakest sub-index, consistently, is information literacy and critical thinking, not technical skill. Correlation data shows literacy tracks with younger age, higher education, and residence outside Java, meaning the gap is layered across several axes simultaneously, not a single line dividing rural from urban.

Four, a civility deficit paired with mass anonymity

Microsoft's Digital Civility Index places Indonesia 29th of 32 countries surveyed, the lowest ranking in Southeast Asia, and research on Indonesian online behavior directly links anonymity to reduced accountability and a corresponding drop in communicative ethics. Adults, not teenagers, are the largest contributors to the decline in online civility, according to Microsoft's own data, a detail most institutional crisis playbooks quietly assume the opposite of.

Five, ungated access

Cheap data plans mean anyone can participate the moment they go online, no verification step, no readiness threshold, sometimes no adult supervision at all.

None of these five conditions is a fourth velocity layer. They are the terrain the three velocities move across, and Indonesia's terrain is structurally worse than the terrain the original theories assumed.

What the Rest of the World Does Instead

Indonesia is not uniquely afflicted, and pretending otherwise would be its own kind of provincialism. The honest comparison sharpens the theory rather than excusing it.

India is the clearest developing-world parallel, a large digital democracy with a literacy gap of its own, but its state response is more extreme than Indonesia's. India has led the world in internet shutdowns for six consecutive years, 116 in 2023 alone, with Kashmir alone recording 226 documented shutdowns since 2012. India's own Supreme Court has ruled indefinite shutdowns illegal. The shutdowns continue anyway.

The European Union takes the opposite approach, regulation of platforms rather than control of infrastructure. The Digital Services Act requires large platforms to run risk assessments and maintain crisis response mechanisms for rapid disinformation spread. Even this model is contested internally, critics argue the Commission's power to unilaterally declare a "crisis" and direct platform behavior risks becoming a backdoor for the exact kind of state overreach the regulation was meant to prevent. The instinct to control is not absent in developed democracies. It is simply channeled through courts and codified law instead of a switch at the network layer.

Finland takes a third path entirely, no crisis mechanism, because the goal is immunity before the crisis exists. Finland is consistently ranked the most resilient country in Europe against disinformation, and the mechanism is education, media literacy woven into the national curriculum from early childhood through upper secondary school, built on the finding that critical thinking skill correlates directly with resistance to false information. The less obvious ingredient is trust. Finns report high institutional trust in government, police, and media, which means an official statement during a crisis actually lands as credible. That trust is the asset Indonesia's five structural conditions are actively eroding.

Four countries, four philosophies. Indonesia blocks selectively and after the fact. India shuts down wholesale. The EU regulates the platform layer. Finland builds the citizen's immune system decades in advance. Only one of the four treats the problem as something to prevent rather than something to manage once it arrives.

The Physical Constant Nobody Is Negotiating

Now for the correction I owe the framework's original premise, because precision matters more than a good line.

The speed of light has not been broken. Data in fiber optic cable moves at roughly two-thirds the speed of light, and every 1,000 kilometers still costs about 10 milliseconds of round-trip delay. That ceiling is not going anywhere, no amount of venture capital changes the refractive index of glass.

What has actually happened is worse in a quieter way. Institutional response time, measured in hours to days, has stayed roughly where it always was, while information propagation time, measured in tens of milliseconds intercontinentally, keeps compressing toward that physical floor. The ratio between the two is the variable that matters, and it is a ratio, not a race, which means no amount of "responding faster" solves it. You cannot out-sprint a limit defined by physics using tools built for a sprint defined by news cycles.

This is the piece that turns a country-specific observation into something closer to a general law: the gap is not a social media problem. It is an infrastructure problem that social media happens to sit on top of.

The Formal Model

Put the pieces together and a single ratio falls out.

CVD = V_info : [V_verify × W]

Where (V_info) is the speed information travels, now approaching the physical ceiling of the network itself. (V_verify) is the speed an institution or a public can confirm, correct, or officially respond, bound by human and organizational capacity, growing slowly if at all. (W) is a weighting factor between 0 and 1 representing how much a country's structural conditions, the five described above, erode effective verification capacity. (W) near 1 describes something close to the Finnish condition. W approaching 0 describes a country where verification capacity is functionally paralyzed by the terrain itself.

CVD is Crisis Velocity Differential, and it does not sit still. The numerator rises every year because we are deliberately building faster networks, a genuinely good thing for commerce, healthcare, education, almost everything. The denominator in Indonesia is not rising to match it. National digital literacy improved by roughly two-tenths of a point in two years while network speed compounded. The ratio is not stabilizing at a worse level. It is heading toward infinity, and it will keep heading there for as long as V_info keeps climbing while W stays flat.

The strategic implication is blunt. Any strategy built to chase the numerator, faster statements, faster clarifications, faster apologies, is fighting a variable that is mathematically unbeatable. The only lever that can actually be pulled sits in the denominator, and it has to be pulled before the crisis starts, not during it.

A theoretical ratio means nothing until it survives contact with structural friction. To evaluate whether the CVD model holds across asymmetric operating environments, the matrix below stress-tests the formula across six distinct institutional and corporate crises in Indonesia. The diagnostic output across all six archetypes remains structurally invariant: when $W$ collapses, reactive verification cannot outrun network physics, and reputational divergence becomes inevitable.

The detailed version and the conclusion -

Running It Against Bank Mandiri

Theory earns its place by surviving contact with a real case, so here is the test.

In August 2026, Rp80.9 million held in a customer's account at Bank Mandiri became inaccessible, triggering a five-day sequence that pulled the bank to the top of Indonesian social media, drew commentary from parliament and police, and pressured its share price. I have written about this case in detail elsewhere. Here, I want to run it strictly through the framework above.

The Layer Zero weighting factor (W) was depressed across nearly every structural axis simultaneously. The account holder was based in Pati, a region with below-average digital literacy relative to Java's metropolitan centers. This accelerated unverified spread across closed channels during Dark Velocity before the narrative ever crossed into public view. Once algorithmic distribution ignited at Open Velocity, Indonesia's civility deficit - a documented score of 96 for uncivil political interaction specifically - explains why the bank's comment section was overrun disproportionately to the underlying operational dispute. Anyone, regardless of verification capacity, could amplify instantly, because network access carries no gate.

Dark Velocity was invisible to the bank by design, not by negligence. No monitoring framework built on open-platform APIs was ever going to see a WhatsApp forward chain in Pati.

Open Velocity moved at close to the network's physical ceiling. Once the story crossed into X, recommendation engines did in hours what would have taken print media days a decade earlier. Concurrently, three separate institutionsthe bank, the police, and PPATK issued three unaligned accounts of who ordered what in the same public square. That is not a messaging failure. It is an architecture failure: three disparate narrators operating without a shared protocol.

Permanent Velocity absorbed all of it. The apology, the institutional confusion, and the retroactive shift in terminology from "block" to "transaction hold" are now permanently etched into search indices and LLM training corpora, surfacing every time an automated system or user queries Bank Mandiri's governance record.

The diagnosis that falls out is not that "the bank apologized badly." It is that the bank possessed no protocol for Dark Velocity, operated on a W too low to dampen Open Velocity, and had no pre-built reputation architecture to control what settled into Permanent Velocity. All three failures are structural, and all three were solvable only in the calm before the crisis, not in the five days after it.

Is This Ready to Publish

Yes, with one condition attached, and I want to state the condition as plainly as I state the theory itself.

What sits above is an original synthesis, grounded in verifiable data across five countries, tested against a live case, and mathematically explicit about the mechanism it claims to describe. That is a legitimate contribution. It is not, however, a peer-reviewed academic theory yet, and calling it one before it has been through that process would be the exact kind of unearned authority this framework argues institutions should stop manufacturing for themselves. Publish it as what it is: a working framework, offered for scrutiny, built to be argued with.

That distinction is not a weakness. Coombs's SCCT started as a working paper too. Every framework that eventually became a textbook chapter began as someone willing to be wrong in public first.

The Only Lever Left

A watchmaker does not wait for the hands to drift before measuring tolerance. The tolerance gets measured the moment the first gear is cut.

Reputation for any institution operating inside a shrinking-toward-zero response window works the same way. It is not an emergency protocol activated once a brand begins trending. It is deterministic infrastructure: you either engineered Layer Zero resilience and Permanent Velocity indexing before the network caught fire, or you did not. No statement drafted on day five alters that arithmetic.

The systemic antidote to a diverging CVD ratio is therefore not reactive PR, but the structural pre-engineering of the information layer itself - a methodology formalized in Corporate Digital PR & Reputation Architecture and Crisis Management.

— Rochman Maarif writes on search architecture, digital infrastructure, and crisis systems at YPYM, a venture ecosystem under PT ADI TJANDRA TEKNOLOGI.

Personal Note: August 26, 2026. I turn 34 today, but I prefer to see it as 17, with 17 years of compounding execution. I've spent that time watching Indonesia race against AI and infrastructure, still stumbling over asymmetries the developed world solved decades ago. This birthday isn't about aging; it's about recalibration. I'm still young, still ambitious, and still unafraid of mistakes. Every failure is just data completing the design. 26-08-2026 = 8-8-8. To the mystic, eternal prosperity. To the engineer, 1 byte. To me, empirical proof the universe rotated an 8 into ∞ just so the formula would compile without revisions.

Reference Table

I. Foundational Theory (extended by this framework)

II. Indonesia Structural Data (Layer Zero)

III. Comparative Country Data

IV. Physical Infrastructure Data

V. Applied Case