Parallax separates generative imagination from critical authority. A Cartographer proposes distant structural candidates; a Skeptic searches for breaks; a Historian looks for prior art; an Experimentalist converts surviving relations into falsifiable tests; and a Boundary Keeper reveals ethical or categorical danger. Graph neural networks and multimodal embeddings can surface candidates, but formal role alignment, causal models and human judgment determine whether a bridge deserves to survive. The system rewards calibrated uncertainty and productive refutation rather than confident novelty.
PROJECT 023 · Live playable experiment
Parallax
“Can a game discover an idea that no discipline could produce alone?”
MOVE TO EXPLORETHE IDEA IN ONE LINEbranch
01 / PREMISE
The beginning
Parallax treats knowledge not as a collection of answers but as a space that can be folded. Players bring two distant fields together—an immune system and cybersecurity, a forest and a city, music and a market—and search for structures that remain invariant when vocabulary changes. The goal is not to notice a poetic resemblance. It is to construct an analogy strong enough to generate a prediction, expose its own limits and become an experiment. Knowledge connects what exists; imagination connects what has never met; Parallax tests what can survive between them.
Knowledge graphs organise entities we already recognise. Educational games reward recall. Generative models produce plausible connections faster than people can judge whether those connections carry explanatory power. The missing experience is one in which analogy becomes playable, criticism becomes a mechanic and disciplinary difference becomes an advantage. A superficial metaphor should collapse visibly; a useful structural transfer should create a question that could not be formulated inside either field alone.
02 / THE PRODUCT
What it could become
A visual multiplayer knowledge universe in which fields behave as deformable territories. Players select distant domains, align functional roles, fold their graphs together and transfer a hypothesis across the intersection. An AI Skeptic attacks category errors, false equivalences, scale mismatches and missing causal mechanisms. The player does not win by defending an analogy as truth, but by using its failure boundary to specify what must be tested. Surviving bridges become evolving public objects that can progress from coincidence to analogy, hypothesis, prediction, experiment, evidence and eventual refutation.
For whom
- Curious people who enjoy discovering patterns
- Scientists, artists and technologists working across disciplines
- Students learning how knowledge is constructed
- Research teams seeking unfamiliar hypotheses
- Communities capable of contesting how their knowledge is represented
Core capabilities
- Dynamic multimodal knowledge hypergraphs
- Structural analogy and graph-matching engines
- Interactive graph folding and spatial knowledge physics
- Adversarial AI critics with distinct epistemic roles
- Provenance and prior-art tracing
- Prediction transfer and experiment generation
- Multiplayer perspective-dependent views
- Persistent idea lineage from analogy to evidence
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.
03 / FOUNDATIONS
Intelligence and mathematics
Each field is represented as a typed temporal hypergraph. The game searches for partial graph homomorphisms and approximate category-theoretic functors that preserve relations while allowing vocabulary and scale to change. Optimal transport measures the cost of aligning structures; information geometry estimates how much meaning is distorted; causal abstraction tests whether interventions transfer rather than correlations alone; and minimum-description-length penalties resist elaborate analogies that explain nothing. A bridge's value is a vector of structural fidelity, domain distance, predictive novelty, falsifiability, robustness and ethical reversibility—not a single popularity score.
04 / VENTURE LOGIC
How it might live
The public game should remain freely playable as an idea-discovery commons. A professional layer could support private research expeditions for universities, R&D teams, venture studios and organisations confronting problems that cross established categories. Curated challenges, evidence rooms and experiment portfolios could generate revenue without enclosing the public graph. The durable asset would be a provenance-rich record of how cross-disciplinary hypotheses emerged, failed, changed and occasionally became useful—not a database of fashionable connections.
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.
05 / DESIGN PRINCIPLES
Rules for making it real
- 01
A resemblance is an invitation to test, never evidence.
- 02
Reward the discovery of limits as highly as successful connections.
- 03
Keep generation, criticism and verification institutionally separate.
- 04
Show whose knowledge made every bridge possible.
- 05
Never flatten Indigenous, lived or local knowledge into decorative analogy.
- 06
Preserve incompatible perspectives instead of averaging them.
- 07
Make every surviving idea falsifiable or explicitly speculative.
- 08
Popularity cannot substitute for explanatory power.
- 09
Let players remove a discipline and see what thought disappears.
- 10
The best outcome is a question nobody could ask before.
06 / ROADMAP
From question to company
- 01Frame
Launch the first playable two-field graph-folding experience.
- 02Prototype
Add a library of deeply modelled domains and structural roles.
- 03Prove
Allow players to author and challenge new field maps.
- 04Build
Introduce Cartographer, Historian and Boundary Keeper agents.
- 05Advance
Persist public bridges with provenance and revision history.
- 06Advance
Connect surviving predictions to real experimental protocols.
- 07Advance
Test multiplayer expeditions with mixed disciplines.
- 08Advance
Publish an open format for playable analogies and their failure boundaries.
07 / HONEST QUESTIONS
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:
- Turning superficial metaphor into apparent scientific discovery.
- Rewarding novelty while ignoring whether a connection is true.
- Encoding disciplinary bias in the structure of each graph.
- Extracting community knowledge without attribution or governance.
- AI critics converging on conventional ideas and suppressing strange insight.
- Players gaming scores through semantic distance alone.
- Public bridges spreading misinformation after their context disappears.
- Competitive play discouraging uncertainty and collaborative revision.
- Treating mathematical elegance as proof of real-world transfer.
- Building an addictive interface around knowledge rather than a useful instrument.