The AI would not be a content factory. Multimodal models would trace the migration of symbols, claims and visual grammars across media and communities; dynamic graph models would identify emerging coalitions and contested meanings; causal models would separate exposure from selection and estimate bounded effects where experimental or natural-experimental evidence exists. Multi-agent world models would generate plausible responses from institutions, competitors and communities, explicitly as hypotheses rather than synthetic people. Creative models would search a space of interventions while critic agents test stereotyping, coercion, cultural appropriation, dark patterns and the quiet disappearance of alternatives. The system would preserve minority readings instead of averaging them into one sentiment score. Every simulated effect would carry calibration, provenance and a domain beyond which it must not be trusted.
Consensus Field
“What if advertising agencies became accountable for the realities they make possible?”
Test the world the message produces.
Not a prediction. An inspectable hypothesis about how an intervention might alter a cultural field.
The beginning
Consensus Field begins with a provocation: advertising does not merely represent culture; it changes the geometry of what a culture can imagine, desire and coordinate around. A campaign can make an unfamiliar behaviour feel normal, pull distant concepts together, erase alternatives from attention and give a possible future enough shared language to become real. I imagine a new kind of agency that treats every intervention as a measurable transformation of a collective possibility space. Before releasing a story, it would model the cultural field it enters, simulate several plausible trajectories and declare the world it intends to make more likely. Its creative work would still need surprise, emotion and beauty—but it would arrive with causal hypotheses, uncertainty, affected perspectives and responsibility for second-order effects.
The advertising industry measures exposure, recall, clicks and conversion because these are observable near the campaign. It rarely measures how a message changes categories, norms, identities, language or the futures people consider reachable. Attribution systems overclaim causality; audience models compress people into segments; generative systems make infinite content without a corresponding increase in judgment; and optimisation learns to exploit whatever produces a response, including vulnerability. Brand purpose often remains a story placed above the same extraction machinery. The deeper problem is ontological: agencies already participate in constructing social reality, but their tools describe them as distributors of messages. They optimise local behavioural effects while leaving the resulting world outside the brief.
What it could become
Consensus Field would be a cultural intervention studio and simulation instrument. A living cultural graph maps narratives, symbols, communities, tensions, institutions and behaviours through time without reducing individuals to psychological dossiers. A field model estimates where an idea may resonate, interfere, polarise or disappear. Creative teams introduce competing propositions and explore three explicit futures: no intervention, conversion-maximising intervention and shared-value intervention. An adversarial layer gives voice to missing groups, searches for manipulation and reveals which assumptions create the apparent result. The final deliverable is not only a campaign but a signed reality brief: intended cultural transition, commercial hypothesis, uncertainty, groups carrying risk, prohibited tactics, early warning signals and a plan for measuring consequences after attention has moved on.
For whom
- Brands whose products require cultural or behavioural change
- Creative agencies seeking an alternative to content-scale competition
- Public-interest organisations designing social interventions
- Communities affected by the norms advertising helps produce
- Researchers studying culture, persuasion and collective behaviour
Core capabilities
- Temporal cultural knowledge graphs and narrative lineage
- Counterfactual campaign and category simulations
- Generative creative systems constrained by declared intent
- Resonance, interference and polarisation mapping
- Adversarial representation of absent perspectives
- Longitudinal measurement of cultural and behavioural change
- Reality briefs with auditable assumptions and prohibitions
- Reality Delta measurement beyond impressions and conversion
The project becomes meaningful only when a new technical possibility is translated into a clear human advantage, an experience people can understand, and a system capable of earning trust over time.
Intelligence and mathematics
The cultural field can be represented as a time-varying manifold whose local coordinates are meanings, practices and possible actions. Narratives act as transformations of that space: optimal transport measures how attention and meaning move; graph diffusion models propagation through communities; information geometry distinguishes genuine category change from repetition; and topological methods detect the birth or collapse of cultural possibilities. Statistical mechanics offers a careful analogy for collective transitions: local interactions can produce macroscopic order, metastable norms and sudden phase changes without a central controller. Renormalisation asks which patterns survive as we move from individuals to communities to institutions. Causal inference and interference-aware experiments estimate how treatments spill across networks. The central measure, Reality Delta, is not one scalar but a vector across semantic distance, behavioural adoption, plurality, persistence, distribution of benefit and reversibility. A Pareto surface prevents commercial return, cultural value and harm from disappearing into one convenient score.
How it might live
Consensus Field could begin as a high-value strategic and creative practice for category creation, behaviour change and major brand transitions. Its first product would be a pre-launch cultural field map and reality brief tested alongside conventional creative development. Repeated engagements would build a privacy-respecting library of interventions, hypotheses and longitudinal outcomes, enabling software for agencies and internal brand teams. The durable advantage would not be faster asset generation; it would be a causal memory of how narratives interact with culture, a formal language for intended change and a governance protocol clients can audit. In the long term it could become an independent standard for declaring and evaluating the realities commercial communication attempts to produce.
For me, a venture is more than an interesting technology. It needs a narrow first user, a repeated problem, a distribution path, a credible advantage and a reason to improve as more people use it. I would test those conditions before deciding whether this idea should become a company, a product, an open technology or an ongoing research programme.
Rules for making it real
- 01
Every campaign must state the world it makes more likely.
- 02
Model communities as participants in culture, never as vulnerabilities to exploit.
- 03
Preserve incompatible interpretations instead of collapsing them into sentiment.
- 04
Separate commercial effect, cultural effect and public value.
- 05
Treat simulation as disciplined imagination, not prediction.
- 06
Make prohibited optimisation targets explicit before creative production begins.
- 07
Measure consequences after attention and attribution windows close.
- 08
Give affected communities a route to contest the model and the intervention.
- 09
Prefer reversible cultural experiments before irreversible category shifts.
- 10
Creativity remains humanly accountable even when machines generate the material.
From question to company
- 01Frame
Choose one behaviour-change brief and map its narratives, actors, tensions and plausible futures.
- 02Prototype
Define a first Reality Delta vector and measure the cultural baseline before intervention.
- 03Prove
Prototype the interactive field model with competing creative propositions and explicit uncertainty.
- 04Build
Run adversarial workshops with creators, researchers and communities represented by the campaign.
- 05Advance
Launch one bounded intervention with prohibited tactics and pre-registered learning questions.
- 06Advance
Measure semantic, behavioural and distributional effects beyond the normal attribution window.
- 07Advance
Compare the simulated failure modes with observed consequences and recalibrate the model.
- 08Advance
Publish an auditable reality brief format and independent review protocol.
- 09Advance
Build software for agencies to compose, challenge and monitor cultural interventions.
- 10Advance
Develop a public standard for accountable reality-shaping communication.
What could go wrong
Serious imagination includes the possibility that an idea should change radically—or should not exist. These are the tensions the project would need to resolve:
- Becoming a more sophisticated machine for propaganda.
- Presenting cultural simulation with scientific authority it has not earned.
- Turning communities into reductive computational archetypes.
- Allowing powerful clients to define shared value on everyone else's behalf.
- Optimising social norms while hiding political choices inside mathematics.
- Using ethical language as premium positioning without changing incentives.
- Measuring culture so aggressively that observation becomes surveillance.
- Suppressing strange creative leaps because the model cannot predict them.
- Confusing resonance with truth or adoption with social benefit.
- Creating cultural changes that cannot be reversed when harms emerge.