01
Rumor Mutation Lab
Watch one message change as it moves through different actors.
A conversation does not end when the message is spoken. In CPS, information can become structured memory. That memory can create a signal, become socially relevant, find an opportunity to move, pass or fail privacy and relationship gates, and become a new memory in another actor.
CPS Studio makes the path inspectable: who knew it, where it came from, how it changed, and why it moved or stopped.
These are the strongest current CPS use cases because they depend directly on propagation, bounded access, partial memory, mutation, and lineage.
01
Watch one message change as it moves through different actors.
02
Study how trust, evidence, authority, and hearsay shift what agents believe.
03
Build interactive worlds where characters remember, retell, withhold, and react.
04
Test how messages move through teams, roles, and institutions.
05
Explore how disclosure order, trust, and sequencing shape conflict or agreement.
06
Inspect what spreads, what stays private, what becomes hearsay, and what gets blocked.
07
Let visitor interactions leave traces that later visitors encounter indirectly.
08
Track who knows what inside a fictional group.
These use cases extend the same mechanism into education, civic scenarios, player research, and future multi-agent systems.
09
Show that AI agents have partial memory, bounded access, hearsay, and uncertainty.
10
Observe how public messages are reframed inside a small simulated community.
11
Study how players disclose, correct, manipulate, or repair information inside social systems.
12
A future multi-agent assistant where advice comes with a visible decision trail.
This is a positioning map, not a benchmark. It compares what each approach is primarily designed to make visible.
| Feature | CPS | Concordia | Generative Agents | Inworld | Convai | LangGraph / Agent OS | NVIDIA ACE | Google Agent Stack | OpenAI / Anthropic |
|---|---|---|---|---|---|---|---|---|---|
| Inspectable social memory | |||||||||
| Information propagation between subjects | |||||||||
| Lineage / provenance / chain of custody | |||||||||
| Privacy-bounded propagation | |||||||||
| Relationship / trust topology | |||||||||
| Signal → Intention → Opportunity pipeline | |||||||||
| CPS Studio / proof observability | |||||||||
| Controlled belief / rumor experiments | |||||||||
| Game-engine integration | |||||||||
| Realtime voice / avatar / embodied NPCs | |||||||||
| General agent orchestration / tool use | |||||||||
| Production deployment ecosystem |
Across these use cases, CPS focuses on a narrow question: how information moves through an artificial group, what changes, what remains contained, and what later actions can be traced back to it.
Lelit Distrikt remains one bounded downstream client and demonstrator for this work, not the definition of the system.