Quasi simulates a process identified by Ibn Khaldun: diverse, fragmented groups develop shared identity through contact. That solidarity enables collective power. But the success of consolidation — the internal differentiation it produces — erodes the solidarity that built it. New groups on the periphery, still possessing raw cohesion, displace the fragmented centre. The cycle repeats at every scale simultaneously.
This process recurs across language, biology, cosmology, cognition, economics, politics, and culture. Quasi models each as the same underlying dynamic, differently parameterised.
When global connectivity crosses a threshold, convergence begins: nodes pull toward the centre, colours homogenise, connections multiply. When homogeneity peaks, a diffusion event fires — each node's existing divergence from the global mean is amplified rather than randomised. Fragments carry structured identity outward. The most cohesive peripheral cluster receives a boost, modelling the successor group that rises as the centre collapses. The cycle restarts from a rearranged configuration.
Each node is a word or concept, encoded as a six-dimensional semantic vector. Connections represent syntactic and semantic affinity — words that co-occur, modify each other, or share conceptual ground.
Isolated vocabulary clusters represent dialects. Repeated contact draws word-meanings toward shared positions — a common lexicon emerges. Within that shared structure, high-frequency words drift toward specialised meanings, accumulating internal diversity. When divergence exceeds the integrating power of shared grammar, the cluster fragments into new dialect groups, each carrying a structurally distinct sub-vocabulary.
Concrete words persist longer — "stone" shifts less than "justice." Well-connected words snap to new meanings faster because high usage creates more contact pressure. Nodes with grammatically diverse connections resist fragmentation — they represent functioning syntax, not just a semantic cluster.
Lexical Convergence — mean semantic similarity of connected pairs: how much shared vocabulary exists within active bonds. Dialectal Spread — hue variance between cluster centroids: how different the dialect groups currently are.
Each node is a social agent. Its semantic vector encodes social disposition. Connections are social bonds — formed through shared experience, proximity, and mutual acquaintance.
Triadic closure means if A knows B and B knows C, A and C are strongly predisposed to connect. Hub nodes exert superlinear gravity — a highly-connected person draws others disproportionately harder. As solidarity peaks, internal stratification accumulates. When internal differences exceed the integrating power of shared identity, the coalition fragments — each piece carries its own cohesion outward.
Established groups resist merger — both are internally coherent enough to maintain identity against assimilation. Hubs influence neighbours far more than they are influenced in return. Social birth rate responds to local cohesion — communities grow when they are solidary.
Cohesion Index — normalised density of strong bonds. Faction Count — number of distinct connected components with four or more members.
Each node is a stellar body moving through a lifecycle: diffuse nebula, protostar, main sequence star, red giant, remnant. Mass accretes through gravitational bonds. Nodes do not continuously influence each other — meaning transfers only through death events (supernovae).
Diffuse matter accumulates under gravity. Birth rate is highest in the young diffuse phase. Stars forge heavier elements internally. When a massive star dies, a supernova disperses enriched material outward, transferring complexity to neighbours and marking them as the next generation. Each generation condenses from a richer substrate than the last.
Continuous semantic influence is disabled — stars change each other only through death. Low-mass stars live longer and die quietly; high-mass stars die explosively and enrich their surroundings. Birth is suppressed near dense mature clusters.
Stellar Density — mean accreted mass within clusters. Nucleosynthetic Complexity — semantic variance within clusters: the diversity of elements being forged before dispersal.
Each node is an association or memory. Thirty percent are working memory nodes — short-lived, rapidly forming. Seventy percent are long-term. A working memory node that accumulates five or more connections consolidates into long-term memory. Connections are associative links.
Loose associations form around active concepts. Repeated activation reinforces clusters into belief schemas. Dense schemas resist fragmentation (fixation) — a cluster so internally coherent it pulls inward rather than breaking. Eventually disconfirming associations generate dissonance. If sufficient, the schema breaks — its components redistribute across incompatible new clusters. Reframing.
Working and long-term memory create two timescales simultaneously. Birth rate responds to local activation — active clusters generate associations faster. Cross-domain transitions (emotional to analytical, sensory to abstract) register more strongly in drift rate than within-domain shifts.
Associative Density — mean connections per node, normalised. Drift Rate — rolling average conceptual change, weighted toward cross-domain transitions.
Each node is an organism. Its semantic vector encodes heritable traits. Connections represent gene flow between reproductively compatible individuals. Fitness determines reproduction and is calculated from connectivity, mid-life vitality, and the currently-favoured trait dimension.
Selection pressure aligns populations toward adaptive optima — genetic homogeneity increases. Horizontal gene transfer introduces cross-population variation. Environmental pressure shifts periodically, changing which trait dimension confers fitness. Sub-populations encounter different pressures; genetic divergence exceeds gene flow. Speciation.
Fitness peaks at mid-life, not youth. Lifetime fitness accumulates slowly, preventing mature organisms from appearing unfit simply because they are old. When a highly-connected organism dies, its neighbours lose lifespan — ecological cascade mortality.
Genetic Homogeneity — mean pairwise trait similarity within clusters. Speciation Pressure — variance in trait-space distance between cluster centroids.
Each node is an economic agent. Connection count is wealth. Connections are market relationships. High-degree nodes are capital hubs.
New agents emerge in sparse, unserved niches — entry is suppressed near dominant hubs. Network effects accumulate: agents connected to well-connected partners gain more partners. Capital concentrates. Over-leverage builds. When a hub fails, insolvency cascades through its counterparty network. The collapse creates market gaps that new entrants fill, restarting the cycle.
Birth rate is suppressed near hubs and elevated in sparse regions. Insolvency activates before death — nodes approaching end-of-life that are highly connected begin weakening neighbours (counterparty exposure). Hub failure cascades through bond-strength reduction for all connected agents.
Market Concentration — Herfindahl-Hirschman Index on degree distribution. Churn Rate — fraction of connections that are newly formed or near expiry.
Each node is a political actor. Hue derives from ideological position (high social dimension = left, low = right). Connection count is power. Connections are alliances, coalitions, and patronage relationships.
Actors form coalitions along ideological lines. Coalition density builds toward hegemony. But contact between opposed nodes causes divergence rather than convergence — radicalisation. Powerful opponents radicalise more. Centrist brokers accumulate connection reach from both sides. Internal contradictions fragment the hegemonic coalition; peripheral factions with intact solidarity rise to replace it.
Hubs pull neighbours toward their position far harder than they are pulled in return. Cross-ideological contact inverts the influence vector. Centrist nodes accumulate a solidarity boost proportional to their centrism — broker advantage.
Polarisation Index — bimodality of the hue distribution. Power Centralisation — Gini coefficient on connection degree.
Each node is a meme host carrying an idea. Connection count drives transmission rate. High-degree nodes are super-spreaders. Immunity accumulates with semantic drift from the original meme structure.
Transmission is boosted when the meme is pre-aligned with the host existing beliefs. Super-spreaders drive saturation. As hosts are exposed, immunity builds. Partial immunity creates susceptibility to adjacent variants. When immunity saturates the population, the original meme dies; mutant variants continue spreading through the immune landscape.
Content resonance means memes spread faster through pre-aligned populations. Partial cross-immunity produces the boom-bust viral cycle. A node death leaves an immunity trace in neighbours for roughly 180 frames — an empty ecological niche resisting immediate refilling.
Transmission Rate — mean strength of recently-formed connections weighted by activity. Mutation Pressure — how far nodes have drifted from their original meme structure.
Quasi collects no personal data. The simulation runs entirely in your browser. Nothing is transmitted to any server. No cookies are set. No identifiers are created or stored.
Quasi does not use localStorage, sessionStorage, IndexedDB, or any other persistent browser storage mechanism.
Quasi loads one external resource: the Alata typeface from Google Fonts. This request is made directly from your browser to Google's servers and is subject to Google's privacy policy. No other third-party scripts, trackers, analytics, or advertising services are used.
Under the General Data Protection Regulation (EU) 2016/679, Quasi has no obligation to display a cookie consent banner, as no cookies are used. No personal data as defined under Article 4(1) GDPR is processed. No data controller or data processor relationship is established through use of this application.
The Google Fonts request transmits your IP address to Google as a technical necessity of the HTTP protocol. Quasi itself does not log, store, or process your IP address. If you wish to avoid this entirely, Quasi can be run offline with locally-hosted fonts.
Quasi is an original work. The simulation logic, visual design, and paradigm models are the work of the author. The Three.js library is used under the MIT licence.
For any legal enquiries relating to this application, please use the contact details of the publisher.