Puppeteer

A distributed-systems substrate for experiences that happen together.

Developed at nCubo, it is designed to preserve causality and continuity across participants, devices and time.

One domain can be staged across different clients, runtimes and topologies without becoming a different system.

In continuous production since 2018.

The public V2 repository is a current open-source release; it should not be read as the unchanged production codebase from 2018.

Puppet

A Puppet keeps the domain unchanged while clients and deployments change.

With a Puppet, the domain can keep growing instead of being rebuilt around every new application.

The knowledge accumulated over time — routines, rules and operating decisions — stays with the domain instead of being recreated with every new application.

If you've built the same capability ten times, why should the eleventh start from zero?

See it measured in One Play, Many Stages

A Puppet's history is not every request it receives. An exercise becomes one of its acts only after admission; once admitted, success or failure does not change that.

stage 01stage 02stage 03same domain
The Architectural Center

State is local. Continuity is not.

Most systems preserve the resulting state of each interaction. Shared experiences need something more: a continuity that survives across sessions, services and clients.

Conventional
Session AScreen AState A
Session BScreen BState B
Session CScreen CState C

Continuity reconstructed across boundaries.

Puppeteer
  1. Participant actions
  2. Shared causal program
  3. Coordinated manifestations
  4. Continuous domain experience

One shared continuity, organized around causality.

How It Works

One flow, from action to shared experience.

  1. 01

    Participants act

    Each action is recorded as an operation, not just applied.

  2. 02

    Puppeteer interprets

    The action is evaluated against the rules, history and current condition.

  3. 03

    Consequences are coordinated

    Immediate or deferred — reaching one participant, many, or the shared environment.

  4. 04

    The experience continues

    Each participant receives its own manifestation of the same event.

ONE CAUSE, FOLLOWED THROUGHCONTINUOUS SHARED EXPERIENCE · advancesperceived → next actionpreservesinterpretsconditionsPARTICIPANT APARTICIPANT BPARTICIPANT CRECORDCausal JournalSHARED DOMAINrules · history · current conditionSYSTEMPuppeteerCoordinatedConsequencesManifestation AManifestation BShared Environment
The Model

Five parts, one coordinated system.

The parts do not run in a line. They form a cycle held together by continuity — the model comes first, the descriptions follow.

CONTINUITYCONDITIONS01PARTICIPANTS02ACTIONS03RULES04CAUSAL JOURNAL05MANIFESTATIONSCONTINUITYCONDITIONS03RULES01PARTICIPANTS02ACTIONS04CAUSAL JOURNAL05MANIFESTATIONS
  • 01Participants

    The people taking part, from any device or location.

  • 02Actions

    The decisions and inputs participants make.

  • 03Rules

    The logic that decides what an action means and what it produces.

  • 04Causal Journal

    A record of what happened, so the experience can be reasoned about and replayed.

  • 05Manifestations

    What each participant receives, produced from the journal.

The value of Puppeteer does not live in any single component. It lives in the contract between them.

Attention

A subject can remember to look again.

Some concerns matter precisely because nothing happened — the confirmation that never came, the approval nobody gave. A Puppet can set a condition under which it looks at them again. The Puppet owns what matters and why. The infrastructure only makes the return possible.

How attention works

DURABLE HISTORYTHE SAME SUBJECTNOTHING HAPPENSAN OPEN CONCERNATTENTION RETURNS

Memory preserves what happened.Attention preserves what still matters.

Experiences

One domain. Many ways to experience it.

The same identity can be staged for a community activity, a browser experience, an automated participant or a distributed environment.

Executable Worlds

Controlled environments where actions produce visible and persistent consequences.

One Play, Many Stages

Collective Systems

Familiar social structures in which many participants contribute to one evolving condition.

The Village Hall
The Asset

One architecture. A growing technical asset.

Each layer is reusable on its own. Together, they make Puppeteer inspectable, extensible and transferable.

Research & Executable ProofA growing research series, working prototypes, production infrastructure and continuity checks.
Staging & Client AdaptersEvery client reaches the domain through an adapter, so the domain travels without changing.
Domain SpecificationsInvariants, verbs and rules — one artifact that can be declared finished.
Architectural ModelOperations, participants, causality and replay.
The Research

One evolving architectural argument.

A series of published papers traces Puppeteer from its foundational distinction between operations and values to replay, continuity, participants and infrastructure.

  1. 09Identity precedes staging1 domain · 5 stagings · 6 clients · 0 domain edits
  2. 0AThe assembled verbOpen investigation: Does executing a sequence create a capability?
  3. ··10 papers so far
Proven

A production system with a research foundation.

The implementation is part of the argument. Inspect it.

View the source on GitHubPublic implementation · Apache 2.0

Inspect the architecture.

If you are evaluating an architecture for shared, continuous or multi-participant systems — or considering it as a strategic asset — we would like to understand what you are building.