Rubble from the collapsed Morandi motorway bridge is strewn along the railway line in the northern port city of Genoa on Aug. 14, 2018.VALERY HACHE/AFP/Getty Images
Cosimo Pacciani is head of research and chief economist at Poste Italiane in Rome.
Laurence B. Mussio is chair of the Long Run Institute and fellow of the Royal Historical Society of the United Kingdom.
Giorgio de Chirico’s 1914 oil painting The Mystery and Melancholy of a Street presents a landscape of haunting precision: empty arcades casting long shadows, a solitary child running toward darkness, a distant statue observing. It captures not merely abandonment, but something more subtle: the persistence of structures after their animating intelligence has departed. This is precisely what occurs when institutions maintain their architectural presence while systematically eliminating their capacity for history and memory – and with it, their ability to comprehend risk.
From collapsed bridges to grounded fleets, from burned towns to frozen banking systems, the pattern is remarkably consistent. Three interlocking failures produce catastrophe: the fragmentation of knowledge across silos until no single authority possesses the complete picture; the absence of accountability for synthesis, ensuring that warnings are recorded but never compel action; and economic incentives that make proceeding more profitable than pausing. This trinity of institutional failure transforms memory from strategic asset into archived irrelevance.
In an age of cascading crises, the ability to remember is the ultimate competitive advantage; this essay is the second in a four-part series called Futures @ Risk on why that capacity is the most critical – and most endangered – asset for institutional survival.
Today, De Chirico’s empty arcades find their parallel in the silent server rooms of the global economy and the hollowed-out bureaucracies of the modern state. We have constructed systems of profound complexity – from high-frequency trading algorithms to transcontinental energy grids – that operate with what we might call ghost intelligence: the departed consciousness of their original architects persisting only as code, while the human understanding of why they work has been lost to turnover, outsourcing, and the attrition of time. Like De Chirico’s arcades, these systems retain their imposing form but have been emptied of the animating minds that created them. They function seamlessly until they don’t. And when they fracture, we discover that ghost intelligence cannot diagnose itself.
Cultural theorist Richard Terdiman diagnosed this condition as modernity’s “memory crisis”: we confuse storing information with actually remembering it. We build vast archives specifically so we can feel safe ignoring their contents. But the deeper pathology lies in what Mr. Terdiman calls the conflict between “Story” and “Body” – between our bureaucratic narratives (safety reports, economic models, political optics) and the stubborn physical reality they purport to describe. Gravity, geology and rust do not read reports. When the paperwork diverges from physical truth, reality eventually reasserts itself with violence.
Futures @ Risk: History teaches us how to manage risk
On Oct. 9, 1963, deep in the Dolomites, engineers monitoring the Vajont Dam possessed three years of data documenting accelerating slope movements on Monte Toc. When 260 million cubic metres of rock plunged into the reservoir, sending a 250-metre wave over the dam and killing 2,000 people, officials declared the disaster “unforeseen.”
The records suggesting otherwise were scattered across multiple agencies, with no single authority accountable for connecting them. The force of the air displacement alone stripped clothes from victims; debris was found kilometres away in the Piave riverbed.
The Vajont disaster exemplifies the trinity in its purest form. The engineers knew the mountain was unstable. Executives knew the engineers were concerned. The government knew about the reports. But rather than synthesizing these warnings into action, they actively suppressed them to protect the asset’s value during nationalization. And proceeding with the dam remained more profitable than pausing to address geological reality. What failed at Vajont was not memory, but the structural necessity to act on memory.
Fifty-five years later, Genoa’s Morandi Bridge collapsed, killing 43 people. The bridge’s designer had specified in 1979 that his creation would require constant maintenance. Decades of corrosion reports followed. Each warning was dutifully recorded, circulated, filed. The paper trail was perfect. The accountability was absent. Once again: knowledge fragmented across agencies, no one responsible for compelling action, and maintenance deferred because delay was cheaper than repair.
Why examine Italian infrastructure disasters in a Canadian newspaper? Because the pattern is universal.
The incineration of Jasper, Alta., in July, 2024 was a tragedy of jurisdictional paralysis. For years, researchers had explicitly warned that the mountain pine beetle infestation had transformed Jasper National Park into a tinderbox. Yet preventative action was paralyzed by risk aversion and political optics. Internal Parks Canada communications reveal that decision-makers may have feared “public and political perception” more than they prioritized the “actual prescription windows” for burning. When the catastrophe finally arrived, the response was further hampered by jurisdictional friction; an independent review commissioned by the town found that “jurisdictional overlap” between federal and provincial commands disrupted the focus of incident commanders while the town burned. While the fire raged, the institutional mechanism for unified command stuttered.
Wildfires ravaged the town of Jasper, Alta., on July 22, 2024, destroying hundreds of businesses and homes.THE CANADIAN PRESS/Amber Bracken
The result was the destruction of one-third of the town and more than $1.2-billion in damages. The risk was not unforeseen; it was structurally discounted.
The incinerated husk of Grenfell Tower in London stands as the grim architectural blueprint for this pathology. The 2024 public inquiry concluded that the 2017 disaster was the “culmination of decades of failure” driven by corporate “systemic dishonesty” and “institutional indifference.” Residents had explicitly warned on a blog months prior that “only a catastrophic event” would expose the ineptitude of their landlord, yet they were dismissed as “militant troublemakers.” The decision to wrap the tower in highly combustible cladding was driven by a “value engineering” exercise to save £293,368. As the flames climbed, the London Fire Brigade remained frozen in operational paralysis, clinging to a “stay put” protocol for a building that had become a vertical kiln. The 72 deaths were not an accident; they were the result of a system that prioritized profit over life – an act of “social murder.”
The pathology of risk blindness has also infected the digital nervous system of the global economy. In July, 2024, a single software update from cybersecurity firm CrowdStrike grounded thousands of flights, froze banking systems, and cancelled surgeries worldwide. The engineers involved had forgotten the foundational rule of resilience: never update the entire system at once. Efficiency demanded speed; resilience demanded a pause... The systems that failed were victims of efficiency over resiliency, relying on automated validators that failed to catch a critical logic error.
This pattern – meticulous documentation followed by catastrophic failure – reveals modern institutions’ most sophisticated pathology: strategic amnesia. This is not merely forgetting but the deliberate fragmentation of knowledge to diffuse accountability while preserving plausible deniability. When everyone knows something, but no one is responsible for that knowledge, institutional blindness becomes structural. Strategic amnesia achieves something even more devastating than avoiding blame. Without institutional memory, risk becomes unknowable. Risk assessment is fundamentally pattern recognition – identifying what has happened before to predict what might happen again. A bridge inspector who lacks access to how similar bridges have failed cannot properly assess deterioration patterns. A banker without synthesized knowledge of previous market crashes cannot price derivatives accurately. Strategic amnesia does not merely obscure accountability – it destroys the capacity to recognize risk itself.
The decision to wrap Grenfell tower in highly combustible cladding, which eventually contributed to the fire that burned it down, was driven by a 'value engineering' exercise to save £293,368.Hannah McKay/Reuters
That this destruction is a choice, not an inevitability, is demonstrated by institutions that have installed memory into their operating systems. Lloyd’s of London has survived 337 years of catastrophes – from the San Francisco earthquake to the Sept. 11 terrorist attacks – not despite its institutional memory, but because of it. The Lloyd’s Loss Book, maintained continuously since 1734, records every significant claim not merely as an accounting entry but as a narrative: what happened, why it happened, what warning signs preceded it. When underwriters assess novel risks, they consult the market’s living memory – historians and actuaries who serve as pattern librarians – surfacing precedents from centuries of documented experience. Lloyd’s syndicates that ignore these precedents do not merely make bad bets; they lose their market standing among peers who remember their failures. Memory at Lloyd’s is not archived data – it is living capital that shapes every decision.
The Lloyd’s model suggests solutions must address all three elements of institutional failure simultaneously. We propose two interlocking mechanisms. First, “resilience-linked bonds” would create market incentives for maintaining institutional memory. Organizations would receive an independently audited “risk memory score” based on measurable criteria: Can crisis teams access project postmortems from 15 years ago within one hour? What percentage of the risk committee has served through a full credit cycle? Does the institution document near-misses – disasters that almost happened – or only actual crashes? A city or corporation demonstrating robust pattern recognition might borrow at 2 per cent, while one suffering from institutional amnesia pays 5 per cent. This makes memory financially material, transforming an intangible asset into a quantifiable advantage.
Second, strategic historical impact assessments would require named officials to certify their review of relevant precedents before project approval. Liability would attach not to innovative decisions that fail despite proper consultation, but to failures repeating documented patterns without attempted mitigation. This encourages innovation while discouraging repetition of known failure modes. Pattern-recognition platforms could surface relevant historical precedents from distributed databases – but with a critical caveat regarding artificial intelligence: we cannot blindly entrust risk assessment to generative AI. In its current form, AI prioritizes probability over truth and remains prone to confident hallucination. More fundamentally, overreliance risks cognitive atrophy – a phenomenon where human critical thinking degrades because we assume the machine is remembering for us. Technology may surface the ghosts of the past, but only humans conscious of moral weight can choose to heed them. In an era of deepfakes and AI hallucinations, the human officer is no longer just a decision-maker; they are the auditor of reality.
The architecture of modern institutions can resemble De Chirico’s empty arcades: impressive structures drained of the intelligence that created them, sustained by ghost intelligence that cannot recognize its own fragility. We have designed institutions where memory lacks compelling force, where pattern recognition lacks reward, where risk blindness becomes more profitable than risk awareness. The bill for strategic amnesia compounds daily – in collapsed bridges, burned towns, grounded fleets and financial crises that surprise only those who have forgotten what to expect. We can continue to pay it, or we can finally build institutions where memory enables pattern recognition, where accountability compels action, where the past illuminates danger rather than obscuring it. Lloyd’s has demonstrated for three centuries that such architecture is possible. The question is whether we will follow in its steps.