Digital Partner Pen Play in Parallel — a cyberphysical, temporal, active-inference, neuroergonomic, phenomenological, accessibility, and place-based successor to PPPiP (Mikhailova & Friedman, 2018).
A composable, fully-tested computational companion and interactive web instantiation of the DigiPPPiP framework, rendered as a governed research manuscript.
| Version DOI (this release) | 10.5281/zenodo.21815705 |
| Concept DOI (family) | 10.5281/zenodo.21815704 |
| Zenodo record | https://zenodo.org/records/21815705 |
| GitHub release | digi-pppip v1.0.0 |
| Rendered PDF | Digi-PPPiP_combined.pdf (repo root) and output/pdf/digi-pppip_combined.pdf |
| Cover art | manuscript/cover.png (symbolic) |
DigiPPPiP treats the "paper" in PPPiP as a variable cyberphysical substrate — shared tablets, web canvases, AR/VR spaces, and persistent whiteboards — and studies two partners coupling through a shared mark field under an active inference lens. It is not a claim of therapeutic efficacy, neural-synchrony causality, universal accessibility, or AI benefit: those remain empirical questions for the controlled studies the manuscript designs.
This repository is the governed source-of-truth package and manuscript. It
ships typed, tested Python primitives in src/, a modular Pandoc manuscript in
manuscript/, a live interactive web app in web-app/ (a working
instantiation of the design kernel), generated figures and build outputs in
output/, and a no-mocks test suite with a 90% coverage gate.
All computational illustrations and the app's live dashboard metrics are deterministic conceptual models / simulated illustrative values, not empirical fNIRS/EEG findings. See the manuscript's "Evidence Scope and Non-Claims" section.
web-app/ is a self-contained React (Vite) + Socket.IO demo that realizes the
minimum viable kernel: two partners, a shared low-latency drawing surface,
perceptible traces of agency, and consentful control over persistence. Its
live "Coupled Dynamics" dashboard shows simulated Variational Free Energy,
Inter-Brain Synchrony, and Narrative Entropy with an explicit "not measured
clinical data" disclaimer.
Run it:
cd web-app/server && npm install && node index.js # Socket.IO on :3001
cd web-app/client && npm install && npm run dev # Vite on :5173Open http://localhost:5173 in two windows to draw together. The four screenshots
under web-app/screenshots/ are also registered as governed figures in the
manuscript (see src/figure_catalog.py; 39 registered figures total).
src/ pure tested primitives (90% coverage gate)
metrics.py SINGLE numeric authority (covered, tested)
active_inference.py dyadic coupled variational free energy
hyperscanning.py inter-brain synchrony + Forman–Ricci curvature
narrative.py narrative information theory (entropy/surprisal)
... (taxonomy, aesthetics, neuroergonomics, session_events, outcomes,
accessibility, source_quality, claim_ledger, source_verification,
study_readiness, systems_governance, figure_methods,
figure_artifact_audit, figure_catalog, provenance, evidence,
figures, manuscript_variables)
scripts/ thin orchestrators
manuscript/ modular Pandoc sections + config/preamble/SYNTAX/bib
web-app/ React + Socket.IO interactive instantiation + screenshots
output/ generated figures, PDF, web, slides, data, reports
tests/ no-mocks pytest suite
docs/ factored technical docs (architecture/figures/testing/scholarship)
ISA.md, AGENTS.md, RENDERING.md
uv run python scripts/digippppip_figures.py # 39 figures → output/figures/
uv run python scripts/z_generate_manuscript_variables.py
uv run pytest tests --cov=src --cov-branch --cov-report=term-missing --rootdir . --cov-fail-under=90
uv run ruff check src tests scripts
uv run mypy src tests scriptsBaseline: 39 registered figures, 128 tests, ≥95% line+branch coverage (97.45% with the pinned dev toolchain), ruff + mypy clean, figure-artifact audit score 1.0, and a green template prerender. Coverage is enforced at 90%.
This repo has no renderer of its own; it renders as a sidecar of the
docxology/template research pipeline.
Place (or symlink) it at template/projects/working/digi-pppip, then from the
template root:
.venv/bin/python -m infrastructure.validation.cli prerender \
projects/working/digi-pppip/manuscript --repo-root .
.venv/bin/python scripts/pipeline/stage_03_render.py --project working/digi-pppipThe combined PDF is written to output/pdf/digi-pppip_combined.pdf; a copy is
also kept at the repo root as Digi-PPPiP_combined.pdf. Full two-repo
instructions: RENDERING.md.
A real DOI was minted for v1.0.0 on a Zenodo deposit:
- Version DOI:
10.5281/zenodo.21815705— resolves to the published v1.0.0 record. - Concept DOI:
10.5281/zenodo.21815704— the version-agnostic family identifier. - Record: https://zenodo.org/records/21815705
- GitHub release: https://github.com/docxology/Digi-PPPiP/releases/tag/v1.0.0
The DOI is written into the manuscript title page and
manuscript/config.yaml (publication.doi, publication.version_doi).
Suggested citation:
Shrivastava, S., Goh, E. C., Mikhailova, A., & Friedman, D. A. (2026). DigiPPPiP: Digital Partner Pen Play in Parallel. Zenodo. https://doi.org/10.5281/zenodo.21815704
Every number that reaches the manuscript is computed by src/metrics.py
(tested, coverage-enforced), serialized to
output/data/digippppip_metrics.json, and hydrated via
src/manuscript_variables.py. src/figures.py renders only. Quoted manuscript
{{TOKEN}} values are bound to this authority; tests (e.g.
tests/test_integration_consistency.py) enforce figure/token/citation/article
cross-artifact integrity.
The manuscript separates peer-reviewed sources, theory books, preprints,
reports, and official governance anchors through src/source_quality.py and
src/source_verification.py. Each citekey is verified (title, venue, year,
DOI or stable URL) before it is added; Perplexity/web results are discovery
leads only. Every figure is a small reproducible claim object (see the
caption contract in src/figure_methods.py and the artifact audit in
src/figure_artifact_audit.py).
MIT (see manuscript/config.yaml → metadata.license) unless a specific
subdirectory states otherwise.