Roadmap
Stratos governance diagnostic. Version 0.1.
This roadmap describes the planned development phases for Stratos, with concrete milestones, success criteria, and kill-switch conditions at each phase. The kill-switch language is deliberate. Projects of this scope have well-documented failure modes that involve gradual scope creep, declining momentum, and eventual abandonment. The roadmap is designed to surface those signals early and force explicit decisions about continuation.
The roadmap is public so that contributors, reviewers, and users can see what is committed, what is planned, and what is conditional.
Phase 0: Foundation (weeks 1-2)
Goal: a working repository, working ingestion pipeline, and basic infrastructure that the rest of the project can build on.
Concrete deliverables:
- Public GitHub repository with README, LICENSE, CONTRIBUTING, CODE_OF_CONDUCT
- Vercel-hosted Next.js scaffold with placeholder routes
- Python ingestion pipeline running as GitHub Actions cron
- First ingestion run producing static JSON for World Bank Worldwide Governance Indicators, UNDP HDI, and V-Dem
- Methodology documentation published in the repository
- Open-source governance documentation published
- Issue tracker organised with appropriate labels and triage process
Success criteria for advancing to Phase 1:
- Repository is public and discoverable
- At least one ingestion source running end-to-end (data fetched, processed, served as JSON)
- Next.js scaffold deployed and accessible at a public URL
- Methodology documentation complete enough that an external reviewer could comment
Kill switch for Phase 0:
- If Phase 0 takes longer than four weeks, halt and reassess. Phase 0 is foundation work and should not require complex methodology decisions. If it is taking longer, the issue is likely scope confusion or tooling overhead, both of which compound in later phases.
Phase 1: Single-cluster MVP (weeks 3-8)
Goal: a working dashboard for Cluster A countries (high-capacity liberal democracies, approximately 25-30 nations) with the five-dimensional scoring, peer comparison, confidence intervals, and diagnostic narratives.
Why start with Cluster A: best data quality, most well-understood comparators (so the diagnostic outputs are checkable against expert intuition), lowest political risk for initial launch, and clearest demonstration of the methodology working as designed. The point of starting narrow is to ship something defensible and learn from real users before expanding scope.
Concrete deliverables:
- All five dimensions scored for Cluster A countries
- Within-cluster peer comparison view
- Structural-constraint conditioning operational for Cluster A
- Country profile pages with diagnostic narratives
- Methodology transparency views (every score traceable to inputs)
- 80 percent confidence intervals on all dimensional scores
- Five-year trajectory module
- Mobile-responsive frontend
- Open issue tracker with active triage
- Public methodology page
Success criteria for advancing to Phase 2:
- Live dashboard accessible to the public
- At least three external technical reviewers engaged (governance scholars, statisticians, or open data practitioners)
- At least one piece of external coverage (technical blog post, academic newsletter, etc.)
- Zero unresolved high-severity methodology issues
- Codebase is in a state that an external contributor could plausibly clone, run, and modify
Public launch happens at the end of Phase 1. Pre-launch, the repository is public but the dashboard is unannounced. Launch announcement goes through the public-writing pipeline (LinkedIn, technical blogs, governance-research communities) once the dashboard is stable.
Kill switches for Phase 1:
- If Phase 1 exceeds 12 weeks, halt and reassess scope. Six weeks is the target. Twelve weeks is the maximum acceptable.
- If after launch the dashboard receives substantial methodology criticism that cannot be addressed within four weeks, pause expansion and address criticism first.
- If after launch no external technical reviewers engage within six weeks, the methodology may not be sufficiently distinctive. Reassess the differentiation before expanding.
Phase 2: Regime expansion (months 3-5)
Goal: extend coverage from Cluster A to all six regime clusters, building the full regime-classification module and adapting the diagnostic engine for cross-regime appropriateness.
Concrete deliverables:
- Regime classification module operational for all countries with V-Dem coverage (180+)
- Five-dimensional scoring extended to all clusters with appropriate adjustments
- Cluster-specific diagnostic templates
- Cross-source validation for at least one dimension
- Data quality scoring visible on every country profile
- Known-falsification flagging operational
- First external contributor merge accepted
- Translated diagnostic narratives in at least three languages
Success criteria for advancing to Phase 3:
- Full country coverage with appropriate regime-cluster handling
- At least one external contributor with merged substantive contributions
- Methodology paper draft circulated to at least three external reviewers
- Active issue tracker with consistent maintainer response time under 14 days
Kill switches for Phase 2:
- If extending to autocratic regime clusters surfaces methodology problems that cannot be addressed within eight weeks, halt expansion. The cross-regime handling is theoretically nontrivial and a bad implementation is worse than a narrower-scope product.
- If translation quality cannot be verified for languages where the maintainer is not fluent, hold those translations until language-expert review is available. Do not ship low-quality translated diagnostics.
Phase 3: PCR integration and uncertainty quantification (months 5-8)
Goal: integrate Probability Cloud Regression into the conversion-efficiency dimension, propagate uncertainty through all scoring, and complete the methodology paper.
Concrete deliverables:
- PCR applied to conversion-efficiency input-output relationships
- Confidence bands visible across all dimensional scores
- Uncertainty propagation to diagnostic narratives (the narratives respond to wide bands with appropriate hedging)
- Methodology paper draft ready for submission
- Cross-source uncertainty handling operational
- Bayesian hierarchical conditioning model fully documented
Success criteria for advancing to Phase 4:
- All five dimensions carry quantified uncertainty
- Methodology paper draft has received substantive review from at least two domain experts (statistics, governance scholarship)
- PCR integration is robust to data updates
Kill switches for Phase 3:
- If PCR integration takes longer than four months, reassess scope. PCR is the methodology dependency that most clearly differentiates Stratos. If integration is fundamentally difficult, the project may not be ready for full methodological ambition.
- If methodology paper review surfaces substantive validity concerns, halt expansion and address the concerns before extending the dashboard.
Phase 4: Trajectory module and early warning (months 7-12)
Goal: complete the trajectory analysis module with early-warning signals for democratic backsliding and state-capacity decline.
Concrete deliverables:
- Five and ten-year trajectory analysis on all dimensions
- Early-warning module flagging concerning trajectories for human review
- Trajectory narratives integrated into country profiles
- Historical comparison view (how does this country today compare to itself a decade ago)
- API endpoints for programmatic access to dashboard data
Success criteria for declaring v1.0:
- All planned features operational and stable
- At least three regular maintainers beyond the founding author
- Cited in at least one external research output
- Used in at least one civil society or journalism context
- Sustainable maintainership pathway documented
Kill switches for Phase 4:
- If the early-warning module produces false positives at a rate that would damage credibility, halt deployment and improve specificity first.
- If the project has not attracted at least one regular contributor by month nine, the open-source community-building is not working and the maintenance burden may be unsustainable. Either find a pathway to a contributor base or scope down maintenance commitments.
Beyond v1.0: long-term vision
Several extensions are possible after v1.0 and are not committed:
- Subnational extension for federal countries with significant within-country variance
- Sectoral deep dives (health governance, education governance, fiscal governance) as separate but linked dashboards
- Forecasting layer using the trajectory module as input
- Comparative scenario tools for users to model the impact of hypothetical policy changes
- API and data products for academic and civil society users with deeper access needs
These are flagged as possibilities, not commitments. Any of them would be a substantial project in its own right.
Resource and time honest assessment
The roadmap assumes part-time work by an independent researcher with AI-assisted coding tools. Realistic time estimates:
- Phase 0: 30-50 hours of work, calendar time 2 weeks
- Phase 1: 80-150 hours of work, calendar time 6-8 weeks
- Phase 2: 100-200 hours of work, calendar time 8-12 weeks
- Phase 3: 150-250 hours of work, calendar time 12-16 weeks
- Phase 4: 80-150 hours of work, calendar time 8-12 weeks
These estimates assume the founder is the only consistent contributor. Each external maintainer joining reduces the founder time but adds coordination overhead.
The total project, from start to v1.0, is realistically 8-12 months of part-time work. This is honest. It is not a weekend project, a side project, or a quick win.
Decision points and review
The roadmap is reviewed quarterly. Decisions to advance phases, hold phases, scope down, or halt are documented in ROADMAP_DECISIONS.md with reasoning.
External contributors are invited to comment on roadmap decisions through the issue tracker. The founding author retains final decision authority through Phase 1 (the BDFL period). After Phase 1, roadmap decisions move to a documented governance process specified in OPEN_SOURCE_GOVERNANCE.md.
Dependencies and risk
The roadmap has three external dependencies that could materially affect timing.
Data source availability. Several data sources release on annual cycles. If a source delays its release or changes its licensing terms, the affected dimensions may need fallback handling. Tracked in DATA_SOURCES.md.
PCR development. Stratos depends on PCR as a methodology dependency. The PCR codebase is maintained separately. Changes to PCR that affect the integration point require coordination.
Contributor availability. The transition from BDFL to broader maintainership in Phase 2 depends on attracting and retaining contributors. This is genuinely uncertain. The kill-switch language in Phase 4 reflects that uncertainty.
What this roadmap is not
It is not a marketing document. It does not promise a specific launch date publicly. It is not a commercial project plan. It is a working artefact for the maintainer and contributors to track honest progress against honest expectations.
When the roadmap is wrong, it gets updated. The version history is preserved.