Forecasting the Information Ecosystem: What the Water Model Predicts

Prediction 1. The Splinternet Is a Dam System — and the Water Will Find Its Way Around

The hydrodynamic law: water always finds a path through any obstacle. The only way to completely stop the flow is to drain the source entirely or hermetically seal the basin. Both options carry catastrophic costs.

Russia is preparing for a "sovereign internet" effective March 1, 2026 — Roskomnadzor will have the authority to isolate Runet from the global network. China has been building the Great Firewall for two decades. Iran reduced connectivity to 3–4% of normal levels in January 2026. The United Kingdom, through the Online Safety Act 2023 and the Investigatory Powers Act 2024, is constructing its own version of a "digital iron curtain."

Prediction: none of these projects will achieve hermetic isolation. The water model predicts not a binary "splinternet" (connected / disconnected) but stratified reservoirs with varying permeability: a surface layer (the official, controlled internet), groundwater (VPNs, Tor, mesh networks), and artesian wells (Starlink, LEO satellites). VPN usage in Russia rose to 36% in 2025; in China, it nearly doubled in 2023. VPN demand in the UK surged 1,400% after the OSA was implemented.

The water analogy yields a precise prediction: the higher the dam, the greater the pressure, the more powerful the underground flow. The more states build "dams," the faster circumvention technologies evolve. By 2028–2030, LEO satellite networks (Starlink, Amazon Kuiper, OneWeb) will make physical isolation technically impossible without military-grade GPS jamming — and even Iran has shown that jamming doesn't stop the smuggling of 50,000 terminals into the country.

But the water model also predicts: total openness is just as much a myth as total blockage. The global internet is evolving not toward a binary split but toward a system of interconnected reservoirs with varying contamination levels and water quality — much like how actual hydrological systems work.

Prediction 2. The AI Deluge: Eutrophication of the Information Reservoir

Eutrophication is the process by which an excess of nutrients (typically nitrogen and phosphorus) triggers explosive algal blooms that kill the rest of the ecosystem by consuming all available oxygen. The water turns green, murky, unfit for life.

Generative AI is doing precisely the same to the information reservoir. By some estimates, AI-generated content could soon account for 99% or more of all information on the internet. Researchers have already documented the phenomenon of model collapse — when AI models train on the output of other AI models, quality degrades with each iteration, much like water that recirculates without treatment accumulates toxins.

The Reuters Institute has identified a new phenomenon — AI slop (information sludge): low-quality, unedited, yet formally literate content flooding the internet. NewsGuard has found hundreds of "news" sites entirely generated by AI. Oxford researcher Sandra Wachter calls this "careless speech" — technically not disinformation, but delivered in a confident tone with zero accountability for accuracy.

Prediction: information eutrophication will become the defining phenomenon of 2026–2030. Not disinformation — but sludge. Not poison — but turbidity. Not lies — but an endless torrent of content where it becomes impossible to distinguish quality from waste. The water model predicts that the primary threat is not that "dirty water will poison" but that the sheer volume of murky water will render clean water virtually invisible.

Paradoxically, a study by Campante et al. (2025) in partnership with Süddeutsche Zeitung showed that when readers become aware of the threat of disinformation, demand for verified sources increases. This is the eutrophication effect in reverse: when the reservoir "blooms," the value of a clean spring rises.

Prediction 3. Echo Chambers Are Not Swamps but Meromictic Lakes

The conventional view of echo chambers as "stagnant swamps" oversimplifies reality. A systematic review of 129 studies (Hartmann et al., 2025) showed that the echo chamber effect is real but far less prevalent than commonly assumed. Most users encounter some degree of informational diversity.

A more precise water analogy is the meromictic lake: a body of water where the upper layer (mixolimnion) circulates freely while the lower layer (monimolimnion) remains isolated for years. Between them lies the chemocline — a zone of abrupt transition. Most people inhabit the upper layer, where information circulates (they encounter diverse viewpoints through work, friends, incidental contacts). But deeply polarized communities form the monimolimnion — a zone where informational exchange with the outside world is virtually nonexistent.

Prediction: algorithmic personalization will transform partial stratification into full stratification by 2028 through three mechanisms. First — AI content curation: Reuters Institute reports that 70% of media leaders are concerned that AI responses are diverting attention from publisher content. Second — the creator economy: 39% of publishers fear losing their best talent to creators who offer more control and better compensation. Third — personalized news: researchers already predict that some outlets will begin not just ranking but generating personalized versions of the same stories, tailoring emphasis to each reader. The result — even those reading "the same" story will be drinking different water from the same glass.

But the water model also predicts a counter-effect: artificial stratification is unstable. In natural water bodies, artificial stratification eventually breaks down — either through seasonal mixing or catastrophic collapse (limnic eruption). The informational equivalent of a limnic eruption is the moment when reality so sharply contradicts the narrative of a closed community that the "bubble" bursts — Bucha, Chernobyl, the FTX collapse, COVID.

Prediction 4. Provenance Will Matter More Than Content

In water supply systems, quality is determined not only by the water's composition but by tracking its entire journey from source to tap. The same applies to the information space.

C2PA (Coalition for Content Provenance and Authenticity) is essentially a "source-to-tap" tracking system for content. Currently, fewer than 1% of news images worldwide include C2PA metadata. But the Reuters Institute forecasts growing adoption: the BBC, AP, and AFP are already piloting the system.

Prediction: by 2028–2030, provenance will become the primary marker of trust, displacing brand and even content. Just as water labeling (source, bottling date, certification) became mandatory in developed countries — content labeling (author, editorial team, AI involvement, data source) will become standard.

But the water model also predicts a limitation: metadata can be "stripped" at any stage (like peeling a label off a bottle), and the emergence of labeling won't eliminate unlabeled "tap water" — it will simply create two parallel markets: certified and uncertified information.

Prediction 5. Journalism Will Become "Bottled Water" — Expensive, Niche, Premium

Reuters Digital News Report 2024–2025: overall trust in news stands at 40%; news avoidance is at 40%. Edelman Trust Barometer 2025: media is the least trusted institution. Nearly two-thirds of respondents cannot distinguish reliable news from disinformation.

Prediction: traditional journalism will undergo the same evolution as drinking water. When tap water "went bad" (loss of trust through commercialization, clickbait, propaganda), the bottled water market emerged — more expensive but with quality guarantees. By 2030, journalism will stratify into three tiers:

"Tap water" — free, AI-generated, algorithmically curated streams. Mass-produced, low-quality, with impurities. This already exists: AI news sites, Google AI Overviews, aggregated feeds.

"Bottled water" — subscription-based, verified, edited journalism. The Washington Post Ripple model, Süddeutsche Zeitung, The Economist. Research by Campante et al. confirms: awareness of the disinformation threat increases subscriptions to quality media.

"Artesian spring" — independent investigative journalism funded by grants, crowdfunding, or sheer dedication. Bellingcat, OCCRP, Schemes (Skhemy). Small volume, high quality, disproportionate impact. This is the "pure mountain spring" — it feeds the most valuable streams, but is far from universally accessible.

The water model's paradox: the dirtier the reservoir, the more expensive clean water becomes — and the less accessible. Information inequality will grow.

Prediction 6. The Streisand Effect Will Mutate — but the Pressure Principle Persists

The water model predicts: pressure building behind a dam inevitably finds release. The Streisand Effect is a classic illustration. But AI changes the dynamics.

Prediction: by 2027–2028, the Streisand Effect will weaken as a mechanism of truth — because the "water" breaking through the dam may itself be synthetic. Deepfake "leaks," fabricated "investigations," AI-generated "evidence" — all create an information flow indistinguishable from a genuine breach. UC Berkeley researcher Hany Farid warns: people are already equally likely to mistake the real for fake and the fake for real.

But the water model also predicts: total blockage becomes more expensive than total openness. The cost of maintaining "dams" grows exponentially — each generation of circumvention technology is cheaper than the last, while each generation of blocking is more expensive. This is a fundamental thermodynamic asymmetry favoring openness.

Prediction 7. Fragmentation Will Be Fractal, Not Binary

Conventional discourse about the "splinternet" envisions a split into large blocs: a "Chinese internet," a "Russian internet," a "Western internet." But hydrology offers a more precise model.

Real water systems don't divide into two or three large basins. They form a fractal network: major rivers with minor tributaries, underground streams connecting formally isolated reservoirs, seasonal floods that temporarily unite separate systems. The Internet Society already identifies not a binary split but an "archipelago" — a network of interconnected yet heterogeneous spaces with varying degrees of openness.

Prediction: by 2030, the information space will take the form of a hydrological basin: one "world ocean" (the open internet, which is shrinking), several "seas" (regional internets with limited interconnection — the EU, BRICS, the Anglosphere), dozens of "lakes" (national networks of varying isolation), and an infinite number of "underground streams" (VPNs, mesh networks, Tor, Starlink) connecting everything despite the surface geography.

The sabotage of undersea cables in the Baltic Sea (2024–2025) is no longer a metaphor but literal severing of information "waterways." Fragmentation is moving from the software layer to the physical.

Prediction 8. Cognitive "Water Quality" Will Become a Public Health Issue

Gerbner's cultivation theory describes media influence as "a slow, steady drip that carves a channel through solid rock." If information is the water we are "made of" (just as the human body is 60–80% water), then the quality of that "water" is a health issue.

Reuters reports: 40% of people avoid news, and the primary reason is its negative impact on mood. This is, essentially, "hydrophobia" — a reflexive avoidance of all water out of fear of contamination. 65% cannot distinguish quality news from disinformation — they have lost the ability to "test the water."

Prediction: by 2028–2030, "information hygiene" (Mike Caulfield's term) will transform from a metaphor into a literal policy framework. Just as society recognized the link between water quality and health in the 19th century (John Snow and the cholera pump), the 21st century will recognize the link between information environment quality and cognitive/mental health.

This means: media literacy in school curricula (like sanitation education in 19th-century schools), regulation of "information discharge" (like environmental legislation), certification of information "sources" (like drinking water certification), and — inevitably — abuse of these tools by authoritarian regimes, which will shut off unwanted sources under the guise of "purification."

Integrated Forecast: Where Does the Water Flow?

The water model produces a paradoxical yet consistent picture. On one hand — fragmentation, eutrophication, stratification: the information space is becoming murkier, more divided, more layered. On the other — hydraulic pressure doesn't disappear; it grows: the more blockages and contamination, the more powerful the underground streams, the more expensive clean water, the higher the demand for verification.

The water model's central conclusion: informational equilibrium is impossible. Water systems are always in motion: erosion, accumulation, evaporation, precipitation, filtration. The information ecosystem will likewise never "stabilize" — it will only transition from one dynamic regime to another. And the most dangerous state is not a storm but stagnation. Because stagnant water is the only water guaranteed to rot.