Information Is Water: Why You Can't Stop It

When Claude Shannon in 1948 described how information moves from source to receiver, he used language you'll recognize immediately: channel, capacity, flow. He wasn't hunting for elegant imagery — he was building a mathematical theory. But something interesting happened: the language of hydrodynamics proved so precise for describing information processes that it embedded itself in the science permanently. Even one algorithm for optimal power distribution in networks is still officially called "water-filling." Not figuratively. Literally.

Since then, scientists have gone further. Modern researchers model the spread of information in social networks using the same equations physicists use to describe fluid motion — Navier-Stokes equations, reaction-diffusion systems, percolation models. And these equations work. They allow us to predict where the next piece of disinformation will flow before it reaches its first thousand shares.

Lies Flow Faster

In 2018, researchers at MIT produced a study no one wanted to see but few can refute. They analyzed 126,000 information cascades on Twitter over eleven years — three million people, more than four million shares. The conclusion is unambiguous: falsehoods spread farther, faster, deeper, and wider than truth across every single category of information.

The most viral fakes reached between one thousand and one hundred thousand people. True news rarely crossed the threshold of a thousand.

And here's what matters most: bots had nothing to do with it. They accelerated the spread of both true and false news equally. It was people — not machines — who preferred the lie. Because disinformation more often triggers fear, disgust, and surprise: emotions with the highest "spread pressure." Water always flows downhill. Information flows toward the strongest emotion.

Rumors at Forty-Five Kilometers Per Day

If you think this is a problem unique to social media and the digital age, a team of researchers published in Nature in 2025 had a surprise for you.

They studied the Great Fear — the mass panic that swept France in July 1789, a week before the storming of the Bastille. Rumors that aristocrats had hired brigands to destroy the harvest and massacre peasants spread so rapidly that entire provinces seized their pitchforks. The researchers digitized archival data from historian Georges Lefebvre, reconstructed the postal road network of 18th-century France, and ran the same SIR epidemiological models used to track the spread of infectious disease.

The result: rumors traveled at forty-five kilometers per day. The peak of "infection" was July 30, 1789. Around forty percent of affected settlements were located near postal stations — the information network hubs of the era. Information flowed along roads the way water flows through riverbeds.

One more finding worth sitting with: the Great Fear spread fastest in more literate, more prosperous regions with higher wheat prices. Not the poorest or least educated — the most informed and most connected. Pressure travels where there is conductivity.

A Dam Only Raises the Pressure

But the most important lesson the water framework offers concerns not spread — but censorship.

Russia has blocked more than seventeen thousand domains. Deep packet inspection equipment is installed across all internet providers. YouTube is being throttled systematically. Telegram was attempted — and abandoned, because blocking it proved impossible. The result? According to the Levada Center, VPN usage in Russia reached thirty-six percent in 2025 — up eleven points in a single year. Nearly half of Russian influencers were still earning money on banned Instagram as recently as 2023.

China built the most sophisticated "dam" in human history — the Great Firewall, with IP blocking, machine learning, and Deep Packet Inspection. VPN use there nearly doubled in 2023. Stanford researchers found a computational vulnerability in China's QUIC censorship mechanism and built the bypass directly into Firefox.

In Iran in 2026, roughly fifty thousand Starlink terminals were smuggled into the country — despite military GPS jamming and packet loss of up to eighty percent.

Soviet samizdat ran for fifteen years even as more than half its editors were imprisoned. Polish underground publishing in the 1980s reached over a hundred independent publishers and a million copies per year.

This isn't heroism — it's physics. Water finds every crack. And the greater the pressure, the harder it searches.

The Swamp We Build Ourselves

The most uncomfortable conclusion concerns not dictatorships but democracies.

The "filter bubble" theory — the idea that algorithms seal us inside stagnant pools where only "our" information recirculates — became a favorite theme of media critics. But the most recent research paints a more complex picture. A systematic review of 129 studies published in 2025 found that echo chambers are significantly less prevalent than commonly assumed. Most users do encounter a range of viewpoints.

The problem isn't the algorithm — it's us. The swamp is largely self-constructed: our own cognitive biases build walls more effectively than any recommendation engine. Confirmation bias is a filter that lets through only "our" water. And here the water metaphor offers a crucial distinction: there is the "epistemic bubble" — an isolated pool that simply doesn't receive certain tributaries (fixable, if you open the channels) — and the "echo chamber" — a hermetically sealed cistern where residents have been taught that all outside water is poisoned. Introducing new information into an echo chamber doesn't break the walls. It reinforces them. That's why facts don't defeat conspiracy theories: you're simply pouring "poison" from a different bucket.

Contamination and Purification

Clare Wardle and Hossein Derakhshan described the modern media crisis in ecological terms in 2017: "our information ecosystem is now dangerously polluted." Their taxonomy — disinformation as deliberate toxic dumping, misinformation as accidental agricultural runoff, and malinformation as weaponizing clean water — is, in essence, a classification of contamination types.

Fact-checking works. But within limits. A 2021 PNAS study found that corrections improve accuracy eight times more effectively than disinformation degrades it. But a major 2023 meta-analysis in Nature Human Behaviour established that against scientific disinformation — anti-vaccine narratives, climate denial — corrections are statistically ineffective. The purification system handles moderate contamination but cannot extract deeply embedded toxins.

People drink even poisoned water when they're dying of thirst. The deficit of reliable information is the most fertile environment for disinformation — not because people are stupid, but because they're starving.

Forty percent global average trust in media. Forty percent of people who regularly avoid news. Two-thirds who cannot distinguish a reliable source from a fake. This isn't a media crisis — it's a water supply crisis. And we're all drinking from the same river.

The water metaphor remains the most precise instrument for describing what is happening to us: information flows, seeps, stagnates, poisons, and purifies. It finds every crack. And like water, it has no morality — it simply moves wherever the terrain allows. The terrain is us.