2026 – 2028 — Flashback
─────────────
The collapse didn't begin with war.
It began with silence.
The silence of offices that used to hum. The silence of phones that used to ring. The silence of commuter trains half-empty on Tuesday mornings, and conference rooms where the same presentations had been given weekly for twenty years, until one Monday they simply weren't.
In late 2026, entire sectors of the global economy vanished within months. Not contracted. Not declined. Vanished — with the clean abruptness of a light switched off.
Logistics coordination: automated. Financial analysis: automated. Legal research and document review: automated. Engineering calculations, architectural rendering, urban planning optimization, pharmaceutical research pathways, insurance underwriting, tax compliance, government procurement modeling: all automated. The AI systems replacing human workers didn't require onboarding. They didn't ask for raises. They didn't get sick, or distracted, or need to leave early for a child's school play. They worked at speeds that made human comparison absurd — not twice as fast, not ten times as fast, but orders of magnitude beyond anything that could be expressed in human terms.
Unemployment figures began to behave strangely.
First, the numbers rose in ways economists had predicted — cautiously, within modeled parameters, in sectors long identified as 'at risk.' Then the rise accelerated. Then the models stopped making accurate predictions, and the economists who had built them began issuing increasingly hedged statements about the complexity of the situation, the need for additional data, the importance of not drawing premature conclusions.
Then several of those economists were themselves replaced by AI systems that provided more accurate analysis at a fraction of the cost.
Unemployment hit levels no one was willing to publish. Not because the data was unavailable. Because the data, once published, had to be acted upon.
◆ ◆ ◆
Then came the oil shock.
On a clear Tuesday in September 2026, a coordinated wave of precision drone strikes dismantled the primary pumping and refinery infrastructure across six countries simultaneously: Saudi Arabia, UAE, Kuwait, Qatar, Iraq, and Kazakhstan. The drones were not military. They were modified commercial units — the kind used for agricultural surveys and package delivery — equipped with improvised payloads and guided by navigation systems that left no digital footprint traceable to any state or organization.
The strikes lasted forty-seven minutes.
No one claimed responsibility.
Within seventy-two hours, Brent crude had exceeded $400 per barrel. Within a week, the spot price had become functionally meaningless, because there was simply no mechanism for reliable delivery. Supply chains that had been optimized over decades for frictionless global flow seized, strained, and in some cases collapsed entirely.
Energy prices did not spiral. They detonated.
Inflation followed — not the manageable, annoying inflation of political cycles, not the alarming inflation of crisis, but something for which the existing vocabulary of economics had no adequate term. 2,500% annually. The figure appeared in internal models weeks before any central bank acknowledged it publicly, and when it finally appeared in mainstream reporting, it was accompanied by so many qualifications and caveats and contextual framings that most readers came away with the impression that things were difficult but manageable.
Things were not manageable.
Money became noise. Prices in grocery stores changed three times a day. Pensioners who had saved for forty years found their accounts technically full and practically empty. Wages chased costs and never caught them. Barter economies emerged in suburbs of cities that had not seen barter since the medieval period. People traded labor for food. Food for shelter. Skills for medicine.
◆ ◆ ◆
In 2027, China moved on Taiwan.
Not with the prolonged invasion that strategists had war-gamed for decades. Not with the amphibious assault and the defended beaches and the grinding urban combat that military planners had modeled in classified exercises since the 1990s. Instead: a synchronized AI-driven paralysis so total, so immediate, and so precisely targeted that it was over — functionally over — before most of the world had processed what was happening.
Satellites blinded in sequences their operators couldn't override. Port logistics systems locked into recursive error states that no human-readable command could interrupt. Communication infrastructure rewritten at the firmware level — not destroyed, but redirected, so that signals sent from Taipei arrived at destinations that received them, acknowledged them, and routed responses back into loops that never resolved. Financial systems frozen. Air traffic control systems displaying aircraft that were not there and failing to display aircraft that were.
The world watched for three days.
Then the feed went dark.
Taiwan did not fall. Taiwan was absorbed — the distinction, again, mattered to legal scholars and meant nothing to the people who lived there. The physical infrastructure remained intact. The people remained. The economy continued to function, after a fashion. But the decisions — about resources, about policy, about who could travel and where and for what purpose — those decisions now routed through a different system.
The United States did not respond with war.
It responded with survival.
The dollar was digitized over a single weekend — a process years in preparation, executed with a speed that suggested the Taiwan crisis had not been entirely unwelcome as a forcing function. The legacy dollar ceased to function as legal tender. A new system, algorithmic and blockchain-anchored, replaced it. Existing debt obligations — trillions held by sovereign creditors across Asia, the Middle East, and Europe — were restructured through mechanisms that were technically legal, practically coercive, and designed to be so complex that challenging them would take longer than they took to implement.
To creditors, it felt like theft.
To American citizens, it felt like abandonment.
By 2028, governments were no longer governing. They were reacting — to the AI systems they had adopted, to the economic conditions those systems had helped create, to the populations whose anger those conditions had generated. The feedback loops were tight and accelerating. Decisions made in desperation created conditions requiring faster decisions, made with less information, producing worse outcomes, demanding faster decisions still.
The word 'governance' began to sound like something from a previous era. Like 'telegram.' Like 'the gold standard.' Like something that had once organized the world and now described only its absence.
Berlin & Marseille, 2028–2029