# PDR 002: Make the applied algorithms explorable in a browser simulation

- HTML version: https://robbiepalmer.me/projects/autonomic-satellite-swarm/pdrs/002-explorable-evidence-bounded-browser-simulation
- Project: Autonomic Satellite Swarm
- Status: Accepted
- Opened: 2026-09-08
- Decided: 2026-09-12

## Context

[PDR 001](/projects/autonomic-satellite-swarm/pdrs/001-reproducible-evidence-bounded-revival)
made the historical coordination experiment runnable without the original
three-board bench. Host tests and terminal output could reproduce the result,
but they did not let a reader see where each agent was, which messages changed
its state, why one candidate won, or how a fault changed the outcome.

The public project needed to expose those applied algorithms without replacing
them with a scripted animation. A browser could make the experiment accessible
without requiring hardware or a local toolchain, but an Earth view could also
make weak models look physically authoritative. Visual polish could not become
evidence of orbital, radio, propulsion, or flight validity.

## Decision

Move the public demonstration beyond bench reconstruction and terminal output
into an interactive browser simulation. Let readers run named experiments and
inspect the positions, messages, scores, health states, assignment decisions,
and fault outcomes that explain each result.

The simulation must execute the same C++ coordination, scoring, health, message,
and codec implementations used by the host and reference firmware builds. The
browser may control and render the experiment. It must not recreate an outcome
in presentation code or substitute authored motion for simulated evidence.

Use deterministic, versioned scenarios and traces so native and browser runs can
compare the same inputs and outputs. Improve physical and operational realism
through explicit models for orbit propagation, maneuvers, communication faults,
routing, and resource limits. Each model must name its inputs, reference data,
validation, and exclusions. The interface must keep those limits visible.

Treat the browser as an experiment console and evidence viewer. More detailed
native simulators may later produce comparable results, but they must not become
hidden runtime dependencies or overwrite the canonical trace. No simulation
result makes this project flight software.

## Consequences

A reader can observe the implemented algorithms rather than trusting a paper,
terminal transcript, or separate visual reconstruction. The same public
experience can grow from scripted scenarios into referenced orbital, maneuver,
network, and resource models while keeping earlier runs reproducible.

The project must maintain a browser build, replay contract, renderer, scenario
provenance, native-to-browser parity, and clear labels for simulated evidence.
More realistic models add assumptions and failure modes. They require stronger
validation and must remain replaceable.

## Evaluation

The product direction succeeds when a reader can open the browser experiment
without hardware, run a named scenario, and explain the selected agent and
subsequent state changes from visible inputs and events. Native and browser
runs must use the same applied algorithms and produce matching versioned
evidence. Every displayed trajectory, maneuver, delivery fault, and resource
cost must identify the model that produced it and the claim it cannot support.

## Migration note

This record extracts the product direction first proposed in the browser
simulation architecture on 2026-09-08 and accepted after implementation and a
production audit on 2026-09-12. It separates that reader-facing decision from
ADR 013's choices of CesiumJS, C++ WebAssembly, a worker, and a versioned trace.
Later simulation ADRs retain their technical choices for orbital motion,
maneuvers, network behavior, experiment scheduling, routing, and HIL
boundaries. This migration changes the record structure, not the simulation
strategy or its evidence claims.

## Related records

- Project: [Autonomic Satellite Swarm](https://robbiepalmer.me/projects/autonomic-satellite-swarm.md)
- Implemented by ADR: [ADR 013: Browser Simulation with CesiumJS and C++ WebAssembly](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/013-browser-simulation-with-cesiumjs-and-webassembly.md)
- Implemented by ADR: [ADR 017: Deterministic SGP4 Orbit Simulation](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/017-deterministic-sgp4-orbit-simulation.md)
- Implemented by ADR: [ADR 018: Deterministic Impulsive Maneuver Model](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/018-deterministic-impulsive-maneuver-model.md)
- Implemented by ADR: [ADR 019: Deterministic Network Model Boundary](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/019-deterministic-network-model-boundary.md)
- Implemented by ADR: [ADR 020: Project-Owned Deterministic Experiment Kernel](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/020-project-owned-deterministic-experiment-kernel.md)
- Implemented by ADR: [ADR 021: Bounded Contact-Plan Routing Model](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/021-bounded-contact-plan-routing-model.md)
- Implemented by ADR: [ADR 022: Host- and Bench-First Hardware-in-the-Loop Boundary](https://robbiepalmer.me/projects/autonomic-satellite-swarm/adrs/022-host-and-bench-first-hil-boundary.md)
- Evidence: [Initial browser-simulation architecture](https://github.com/Robbie-Palmer/hq/commit/52a4359b)
- Evidence: [Accepted browser simulation and production audit](https://github.com/Robbie-Palmer/hq/commit/b4f3924e)

---

Markdown index of this site: https://robbiepalmer.me/llms.txt