The market for stolen memories

In 2038, a woman in Montréal received an invoice for access to a memory of her own wedding. The invoice was genuine. The wedding had been recorded through a lifetime companion that organized photographs, reconstructed missing audio and generated an explorable version of the event. Years later the company had sold the rights to maintain the interactive reconstruction as part of a larger portfolio of licensed experiences. The woman still possessed the photographs. She did not own the system capable of making her memories live again.

The industry had developed gradually, through a feature users loved. Early companions had offered to remember birthdays and find old messages. The next generation could reconstruct the texture of a day: where a person stood, what was said in the adjoining room, how the weather changed and what somebody might have meant. Millions of people allowed their assistants to curate these personal worlds. The companies described the service as preservation. Contracts described it as a mixture of storage, inference rights and derivative media licensing.

By 2036, the richest entertainment libraries had begun buying memory gradients: statistical patterns of how people experienced loss, celebration, reconciliation, exhaustion or awe. A screenwriter could commission a synthetic scene calibrated against millions of remembered reactions. A virtual city could feel convincingly inhabited because its shops, arguments and festivals were built from fragments of actual human experience. Much of the original data had been volunteered years earlier for convenience. The new value appeared only after those fragments were recombined.

The first major dispute concerned a person who had never signed anything. Two brothers donated their memories of a father. Their recollections included a third sibling who had declined every memory-recording service. Yet the shared world reconstructed her voice, gestures and childhood bedroom from surrounding evidence. The company argued it had created an inference, not copied a person. She argued that a system should not be allowed to manufacture her past from everybody else's records. Courts found themselves confronting a peculiar property: a memory shared between people could be monetized by one and contested by another.

The market responded by inventing experience provenance, a complete chain of sources, permissions and synthetic additions. Customers could pay to enter a reconstructed summer and see which details came from genuine recordings and which had been inferred. The most expensive experiences were not always the most realistic. They were those whose provenance could survive a dispute. A memory with two independent witnesses became more valuable than a visually perfect reconstruction with none.

The second shock arrived when people began to sell memories forward. Under debt pressure, households licensed future experiences to be recorded automatically: a child's first year at school, a grandparent's final months, a wedding not yet planned. Investment funds packaged the rights as entertainment inventory. Proponents claimed that ordinary people were finally being paid for data platforms once extracted for free. Critics noted that a parent could sell a future child’s biography before the child was old enough to object.

A black market developed around absence maps: models capable of identifying the moments a person had deliberately hidden. If a companion recorded almost everything except a single week, it was possible to infer the significance of that week from the surrounding records. The less a person wanted remembered, the more valuable the gap could become. Privacy was no longer simply the choice to delete information. It was the ability to prevent others from reconstructing what the deletion concealed.

States approached the conflict differently. Some recognized the right to revoke commercial access to intimate reconstructions even after an initial licence. Others treated a valid memory contract as durable property. A few restricted the export of entire experience datasets, describing them as national cultural wealth. International companies moved their archives between jurisdictions, promising families protection but also positioning themselves to negotiate with regulators.

By the end of 2038, a new premium industry had emerged: unrecorded life. Resorts sold weekends where no persistent assistant was allowed to infer what visitors had done. Families signed mutual agreements not to reconstruct shared encounters. Certain schools promised students a childhood that would not later become a licensable archive. The wealthy were buying the right to be forgotten while the indebted sold the right to be remembered.

The greatest revelation came later. Synthetic worlds built from private memory could become so familiar that people trusted them more than the incomplete records of the physical world. When the republics of private reality appeared in 2040, their citizens did not enter purely invented landscapes. They inhabited commercially curated versions of their own past. Political power had acquired a new raw material: memory itself.