From Ea-Nasir to the Digital Economy: Volatility and Its Undercurrents

— or, how a 4,000-year-old clay tablet explains why your gold dealer, magazine subscription, and Instagram feed are all part of the same story

Let’s start with a man named Nanni. Around 1750 BCE, in the ancient city of Ur (modern-day Iraq), Nanni did what any aggrieved customer does today: he filed a complaint.

His target was a copper merchant named Ea-Nasir. The issue? Sub-standard copper delivered after a long voyage, a servant treated with contempt, and a merchant who reportedly said: “If you want to take them, take them; if you do not want to take them, go away!”

Ea-Nasir, it turns out, wasn’t just a copper dealer—he was a meticulous record-keeper. Archaeologist Sir Leonard Woolley, excavating Ur about a century ago, unearthed not one but multiple complaint tablets in what was presumably Ea-Nasir’s own home. The man kept his bad reviews. Which means either he was a masochist, or—more likely—he was a prominent enough merchant that complaints were simply part of doing business.

Today, the tablet sits in the British Museum. It holds the Guinness World Record for the oldest written customer complaint. And on the internet, Ea-Nasir has become a meme—a 4,000-year-old symbol of the timeless tension between buyer and seller.

The more things change, the more they stay the same.

The Pattern, in Four Acts

Act I: The Commodity Dealer

Fast-forward nearly four millennia. In Sweden, a gold dealer named Guldbrev found itself in a strikingly similar position. The Swedish Consumer Ombudsman accused the company of misleading consumers: showing high gold prices online, while offering sellers significantly less when they actually sent in their gold.

The case went all the way to the EU Court of Justice and the Swedish Patent and Market Court of Appeal. After a five-year legal battle, the court ruled in Guldbrev’s favor. The information was available on the website, the court said. Consumers just had to look for it.

Here’s the irony: Guldbrev won the case. But the complaints kept coming. Because the gold price spiked after customers sold. And when the price spikes, the regret spikes. And when regret spikes, someone must be blamed.

The pattern is ancient: a commodity dealer, a volatile market, a customer who feels wronged, and a dispute that outlives the transaction.

Act II: The Magazine Subscription

In Finland, the same dynamic plays out in magazine subscriptions. The Finnish Competition and Consumer Authority (KKV) has received over 600 complaints about a single telemarketing operation, KT Kustannus, which used aggressive sales tactics and misleading statements to push magazine subscriptions.

Under Finnish law, telemarketing subscriptions come with a 14-day cooling-off period. The clock starts ticking when you receive the first magazine. You can cancel. No questions asked.

But here’s the rub: the 14-day period doesn’t start when you say “yes” on the phone. It starts when the magazine arrives. By the time the bill comes, the window has often closed. Consumers don’t remember agreeing. They feel trapped. They file complaints.

The system is legally compliant. The confirmation SMS is sent. The recording exists. The 14-day period is clearly stated. And yet, the complaints flood in. Because the customer has already bypassed reason—and entered the territory of emotion.

Act III: The Transparent Dealer

Now imagine a gold dealer so transparent that they operate on the ground floor of the Nasdaq Stockholm building—with dog-sized “NASDAQ” letters above the door and cow-sized letters on the floor. The spot price is quoted live from the exchange, which aggregates prices from the LBMA and CME. The gold is weighed in front of the customer. The spread is written in the offer. The recording is provided on request.

And the customer still files a complaint.

Because two months later, the gold price spiked. And the customer, who bought at the previous price, feels cheated. The spread that was reasonable at the time now looks like greed. The transaction that was transparent now looks like a conspiracy.

The dealer is honest. The evidence is clear. The complaint is baseless. And the customer is certain they’ve been wronged.

Act IV: The Algorithmic Pipeline

Now bring it all together. The same algorithmic logic that recommends content on social media also funnels users from legal “gray markets” into illegal “black markets”—a phenomenon researchers call “digital drift.”

One study found that 65% of simulated profiles that engaged with gray-market content (nicotine vapes and sex work) were algorithmically recommended illegal drug sellers within four days—despite never having searched for illegal content.

The algorithm doesn’t discriminate between adults and minors in any meaningful way once you’re over 18. It just optimizes for engagement. And engagement, as it turns out, is a pipeline.

Meanwhile, platforms like Meta have been estimated to earn roughly 10% of their 2024 revenue—around $16 billion—from ads promoting scams and banned products. The platform is full of holes. And it is marketed as a safe space for the vulnerable.

The Volatility Connection

This isn’t a coincidence. It’s a market volatility era phenomenon.

In 2024, the FBI received nearly 55,000 sextortion and extortion complaints—a 59% increase from the previous year. The FBI has reported a 700% increase in sextortion cases targeting teens since 2021. At least 20 victims of financial sextortion of minors have committed suicide; victims are typically males between 14 and 17.

In Finland, police recorded just 9 sextortion cases between 2019 and 2021. In the first nine months of 2025 alone, they recorded 101. Globally, the Internet Watch Foundation reported a 72% increase in sextortion cases in 2025 compared to the same period in 2024, with boys making up 97% of confirmed cases.

Globally, financial fraud losses reached an estimated $442 billion in 2025. According to INTERPOL, AI-enhanced fraud is 4.5 times more profitable than traditional methods. “Agentic AI” systems can autonomously plan and execute complete fraud campaigns—from reconnaissance to ransom demands.

Europol has described the cybercrime market as marked by “extreme volatility.” The OECD found that financial scams and frauds increased in 69% of responding jurisdictions between 2024 and 2025.

Volatility breeds vulnerability. Vulnerability breeds fraud. Fraud breeds complaints. Complaints breed regulation. Regulation breeds more sophisticated fraud. And the cycle continues.

Discerning Fraud from Non-Fraud: A Data-Driven Approach

The challenge of distinguishing between actual fraud and the perception of fraud is not merely academic—it has real economic consequences. The enforcement paradox is that aggressive measures to stop fraud often cost more than the fraud itself. In 2023, an estimated $157 billion in U.S. e-commerce sales were at risk due to false declines**, with **$81 billion projected to be permanently lost—significantly more than the $48 billion lost to actual fraud globally.

So how do we tell the difference between a legitimate complaint and a perception-driven one?

1. The Definitional Divide

Actual fraud involves a financially harmful action taken without the user’s consent. Perceived fraud, by contrast, occurs when a customer believes they are a victim of fraud but are in fact the victim of confusing financial information, market volatility, or their own regret.

2. The Complaint Signature

Actual fraud complaints and perception-driven complaints leave different digital traces. Researchers have developed LLM ensemble approaches to distinguish scam complaints from non-scam fraud complaints, with hybrid NLP frameworks using Knowledge Graphs and LLM embeddings achieving an F1-score of 0.80 with 100% recall.

3. The Feature Set

Feature	Actual Fraud Signal	Perception-Driven Signal
Timing	Complaint filed promptly after unauthorized transaction	Complaint filed after price movement, market correction, or bill arrival
Narrative	Specific, consistent, verifiable details	Vague, emotional, shifting account
Evidence	Willingness to provide documentation	Reluctance to engage with evidence; dismissal of recordings as “fake”
Response to evidence	Accepts resolution when fraud is disproven	Continues to insist on fraud despite clear evidence

Ensemble machine learning frameworks have achieved 96% F1-score on customer complaint datasets and 98% on seller complaint datasets, demonstrating that machine learning can reliably classify fraud incidents.

4. The False Positive Problem

Merchants reject an estimated 6% of e-commerce orders due to fraud suspicion, with false positive rates between 2% and 10%. The downstream impact is brutal: 41% of consumers globally say they’ll never shop with a brand again after a false decline.

5. The Volatility Amplifier

Market volatility acts as a powerful amplifier of perceived fraud. Research has found that financial fraud amounts are significantly correlated with market volatility, regulatory strength, and economic conditions. When prices spike and then correct, consumers who bought at the peak experience regret—and regret is easily transmuted into a belief that they were defrauded.

6. The Regulatory Signal

Regulatory bodies have documented a “regulatory sine curve”—oversight increases following a recession and wanes as the economy returns to normalcy. Fraud detection systems must be dynamic, with feedback loops where outcomes of enforcement influence the retraining of models.

7. The Practical Takeaway

For the consumer, the dealer, and the regulator, the data-driven approach to discerning fraud from non-fraud rests on three pillars:

Narrative consistency: Actual fraud leaves a clear, verifiable trail. Perceived fraud leaves an emotional, shifting account.
Timing: Fraud complaints cluster around unauthorized events. Perception-driven complaints cluster around market movements.
Evidence responsiveness: Victims of actual fraud cooperate with investigation. Victims of perceived fraud dismiss evidence that contradicts their belief.

Beyond Financial Fraud: The False Report Problem Across Crime Categories

The problem of distinguishing actual from perceived or fabricated harm extends far beyond financial fraud. Across the criminal justice system, false reports—whether driven by malice, mental illness, misidentification, or the weaponization of technology—create a persistent challenge.

Sextortion and Image-Based Abuse

Sextortion—the threat to share nude or sexual images to coerce compliance—occurs across a diverse range of contexts: intimate partner abuse, cyberbullying, online dating, sex trafficking, and organized crime. Prevalence studies focusing on youth report rates from 0.7% to 5.0% ; adult studies range from 4.0% to 18.7%. One in six mobile users report having been targeted by sextortion scams. Nearly 9 in 10 extortion victims reported emotional harm.

The reporting challenge is acute. Shame, fear, and negative perceptions of police and digital platforms keep help-seeking rates very low. False reports of sexual violence are under 2.5% of cases. Yet the perception of false reporting can itself be weaponized, as perpetrators use counter-accusations to discredit genuine victims.

Cyberstalking and False Accusations

Cyberstalking involves using technology in threatening ways: to surveil or harass, convey threats, make false accusations, or share embarrassing information. Perpetrators can falsely report their victims to law enforcement—as in documented cases where defendants admitted to multi-year cyberstalking campaigns that included gaining unauthorized access to victims’ accounts, creating fake accounts, obtaining restraining orders using fabricated evidence, and falsely reporting victims to law enforcement, resulting in the victims’ arrest.

AI-Generated Misidentification and False Accusations

The rise of generative AI has introduced a new vector for false accusation: the AI-generated “evidence” that fuels misidentification. In one high-profile case, an AI-generated Google summary falsely identified a Canadian musician as a sex offender, leading to cancelled concerts and reputational damage—based entirely on an AI system stitching together fragments of public information from a shared surname. AI tools have been found to wrongly identify images as real even when they are generated, and falsehoods spread rapidly through AI-generated images taken out of context.

Identity Theft and Criminal Misattribution

Identity theft presents another layer of the false report problem. Perpetrators can assume others’ identities to commit crimes, leaving an innocent person to bear the legal consequences. Synthetic identities were used in one in 10 fraud cases globally—an eightfold increase from other years. Overall identity fraud losses reached $27.2 billion in 2024.

The Investigative Framework

Across all these crime categories, the same three pillars apply:

Narrative consistency: Fabricated reports tend to shift over time; genuine accounts remain structurally coherent.
Timing and context: False reports often cluster around motivated contexts—just as perception-driven fraud complaints cluster around market volatility.
Evidence responsiveness: Those making false accusations tend to resist exculpatory evidence.

Profile Bias: Guilty by Fitting the Description

In discerning between actual and fabricated harm, there exists a deeper, more structural risk: profile bias. This bias does not arise from malice alone but from systemic failures in both human cognition and algorithmic design. An innocent person—simply because they fit a certain “criminal profile”—is at a higher risk of being suspected, accused, or even wrongfully convicted.

Profiling as Evidence Is Unreliable: Criminal profiling lacks sufficient scientific reliability to be used as evidence of a defendant’s guilt or innocence. Its potential to cause prejudice far outweighs its probative value.
The Real Cost of Racial and Ethnic Profiling: Black individuals are disproportionately represented in wrongful conviction cases, with racism being a significant factor. In Denmark, an analysis of over 2.5 million cases found that first-generation immigrants are 27% more likely to be wrongly charged by police than native-born citizens, while second-generation immigrants are 45% more likely.
Algorithmic Prediction and the Cementing of Bias: AI risk-profiling systems are often trained on historically biased data, systematically reinforcing discrimination. This can lead to wrongful criminal accusations against marginalized groups or their disproportionate labeling as “suspicious,” triggering consequences such as incarceration, homelessness, and deportation.
Cognitive Bias and False Accusations: Even without algorithms, human cognitive biases lead to false accusations. 45-50% of all errors of justice result from police or prosecutor misconduct. 22% stem from false confessions, with the rate soaring to 42% when the suspect is under 18. Eyewitness misidentification contributes to more than 60% of wrongful convictions that resulted in DNA exoneration.

Profile bias creates a closed loop: a person is more likely to be accused simply because they “look like” or are “predicted to be” a criminal—and once accused, the evidence is then skewed to prove the initial profile.

The Human Constant

Across all these cases—Ea-Nasir, Guldbrev, Finnish magazine subscriptions, the Nasdaq Stockholm gold dealer, the algorithmic pipeline, the false report problem, and the profile bias—one thing remains constant: the human response to volatility.

When prices spike, consumers regret. When they regret, they blame. When they blame, they file complaints. When complaints flood in, the system processes them. And when the system processes them, it finds—most of the time—that no fraud occurred.

The dealer is honest. The spread is reasonable. The recording is clear. The 14-day period was provided. The complaint is baseless.

But the customer still feels wronged. Because the market moved against them. And the human mind, confronted with loss, seeks a villain.

Ea-Nasir was that villain in 1750 BCE. Guldbrev was that villain in 2024. The magazine telemarketer is that villain today. The gold dealer on Nasdaq Stockholm’s ground floor is that villain tomorrow.

The algorithm doesn’t create the pattern. It just accelerates it.

Sources

Ancient History

British Museum. “Tablet: letter from Nanni to Ea-nasir.” Collection object W_1953-0411-71.
Guinness World Records. “Oldest written customer complaint.”
National Geographic. “Meet Ea-Nasir, a shady ancient merchant—and modern meme.” May 31, 2024.
Wikipedia. “Complaint tablet to Ea-nāṣir.”

Gold and Commodity Markets

Konsumentverket (Swedish Consumer Agency). “Guldbrev får rätt i domstol.” May 23, 2025.
Court of Justice of the European Union. Case C-379/23, Guldbrev AB v Konsumentombudsmannen.

Finnish Consumer Protection

Finnish Competition and Consumer Authority (KKV). “Newspaper and magazine subscriptions.”
Finnish Competition and Consumer Authority (KKV). “Kouluturvaa magazine marketer KT Kustannus continues to mislead consumers.” March 19, 2019.
YLE. “Finland preparing to tighten telemarketing laws.” March 7, 2022.

Sextortion and Online Exploitation

FBI. “FBI New Haven Issues Warning About Financial Sextortion.” August 22, 2024.
YLE. “Boys increasingly blackmailed with sexually explicit media, police say.” December 11, 2025.
Internet Watch Foundation. Annual Report 2025.
Malwarebytes. “New Malwarebytes Research Reveals that One in Six Mobile Users Targeted by Sextortion Scams.” October 14, 2025.

Fraud and Cybercrime

INTERPOL. Global Financial Fraud Threat Assessment 2026.
Europol. IOCTA 2026 report.
OECD. “Risks stemming from the operating environment: Consumer Finance Risk Monitor 2026.” March 2, 2026.
Thomson Reuters Institute. “Scams aren’t just fraud—they’re engineered to exploit human nature.” November 20, 2025.

Platform Governance and Digital Drift

University of Copenhagen. “From Gray to Black Markets–A Quasi-Experimental Study on Algorithmically Driven Digital Drift Opportunities on Social Media.” 2025.
European Parliament. Parliamentary question E-004574/2025: “Meta’s revenue from scam advertising and possible unfair commercial practices.” November 17, 2025.
Reuters / ABC News. “Meta projected 10% of 2024 sales came from scam, fraud ads.” November 2025.

Cyberstalking, AI Misidentification, and Identity Theft

ICPSR. “Cyberstalking: Research and Evaluation to Enhance Criminal Justice, United States, 2021-2023.”
CBC News. “Cape Breton fiddler Ashley MacIsaac says he may have been defamed by Google.” December 24, 2025.
The Mercury. “Musician loses work as AI wrongly names him as a sex offender.” January 13, 2026.
ACFE. “From billions lost to billions saved: A blueprint for beating synthetic IDs.” May 1, 2026.

Profile Bias and Wrongful Convictions

ACLU. “Fatal Flaws: Innocence, Race and Wrongful Convictions.” November 19, 2025.
Statewatch. “Police racism and criminalisation across Europe increasingly fuelled by digital ‘prediction’ and profiling systems.” June 30, 2025.
Rachlew, A. “Causes of Errors of Justice.” Norwegian Centre for Human Rights.
Florida Innocence Commission. “What Defines a Criminal?” February 24, 2025.
SMU Scholar. “Profile evidence” in wrongful convictions.

The robot recites. The lead shouts. The system processes. The complaint is filed. The dealer eats a fatayer.

We have been doing this since Ur. And we are not about to stop.