Canonical Research Roadmap Β· Research status September 6, 2026

Where BenchEWS stands today.

From the released version 1.4.0 through professional release preparation for Studio 2.0 and a planned AI-assisted research interface to the open question of cross-domain mathematical invariance.

This roadmap separates current development, planned capabilities, and long-term research horizons. It is not a product promise. Every later stage remains contingent on implementation, verification, and independent evidence.

RELEASEDCURRENTPLANNEDRESEARCH HORIZONLONG-TERM HORIZON
← Back to homepage

Software Β· Development status

Release horizon

BenchEWS Studio 2.0

Planned release: September 2026

Planning target Β· contingent on final release checks
BenchEWS Studio 3.0 β€” ECHO

Planned for mid-2027

Development horizon Β· next stage of evidence-based system diagnostics

01 / Canonical development path

Five levels Β· five explicitly separated statuses

Tap a stage to inspect its scope, evidence boundary, and next gates.

  1. 01RELEASED
    RELEASED Β· VERSIONED Β· CITABLE

    BenchEWS Studio 1.4.0

    The released and citable software foundation

    Studio 1.4.0 is the currently released BenchEWS software. The versioned Zenodo record combines the application, documentation, and an installable macOS build for Apple Silicon.

    DOCUMENTED CAPABILITIES
    • 12 task-oriented workspaces
    • Executable variance and lag-1 autocorrelation
    • Reproducible reference benchmark
    • Deterministic research workflow
    • Documented configurations, runs, and provenance
    • Exportable Knowledge Package
    • User Manual and scientific companion documentation
    EXPLICIT BOUNDARIES
    • CRTI is not yet integrated as an executable module
    • No quantified measurement uncertainty
    • No automated scientific interpretation
    • No certain tipping-point prediction
    • No independent cross-domain validation
    Open Studio 1.4.0 on Zenodo β†—Inspect capabilities and documentation β†’
  2. 02TODAY
    CURRENT Β· IN FINAL RELEASE PREPARATION

    BenchEWS Studio 2.0

    The planned first fully professional BenchEWS research software

    The goal is a scientifically robust visual tool with CRTI. Studio 2.0 is in final release preparation, but version 2.0 has not yet been released.

    QUALITY STANDARD

    Release quality, reproducibility, transparency, intuitive operation, scientific interpretability, and a professional research identity.

    OPEN RELEASE GATES
    • CRTI and remaining production modules complete
    • Scientific Cockpit
    • All 12 legacy panels preserved
    • Full regression suite
    • Synchronized User Manual
    • Scientific Companion
    • Requirements specification
    • Final cloud readiness
    • Real-Mac test
    • From-zero test
    • Final DMG
    • Zenodo release
    Inspect the current software status β†’
  3. 03PLANNEDPLANNED Β· NOT IMPLEMENTED Β· NOT VERIFIEDStudio 3.0 ECHOA scientific standard tool for early-warning and system diagnosticsOpen details οΌ‹

    Studio 3.0 ECHO is a strategic development goal. Shared mathematics would remain in the BenchEWS core; Domain Packs would make context, data requirements, and interpretation boundaries explicit.

    02A / PLANNED SYSTEM CLASSES
    Climate & Environment↗Economy↗Financial / Capital Markets↗Society↗Homo sapiens↗Technical Systems↗Networks↗Generic / Custom Systems↗
    02B / PLANNED DOMAIN PACK CONTRACT
    • System Boundary
    • Relevant observables
    • Required data
    • Typical timescales
    • Applicable EWS
    • Mathematical prerequisites
    • Failure modes
    • Interpretation limits
    • Provenance
    02C / TWO PLANNED OPERATING MODES
    MODE A

    SCIENTIST MODE

    Full control over data, parameters, methods, models, evidence, and interpretation.

    MODE B Β· PLANNED

    GUIDED RESEARCH MODE

    AI would support less specialized users throughout the research process. It must not conceal scientific choices or replace deterministic computation.

    02D / PLANNED CONTROLLED RESEARCH PATH
    1. 01Research Question
    2. 02System Boundary
    3. 03Observables
    4. 04Data
    5. 05Data Acquisition
    6. 06Source Qualification
    7. 07Data Harmonization
    8. 08EWS Selection
    9. 09Analysis
    10. 10Evidence / Uncertainty
    11. 11Interpretation
    12. 12Reproducibility Record
    02E / PLANNED Β· NOT IMPLEMENTED Β· NOT VERIFIED

    AI DATA ACQUISITION & EVIDENCE LAYER

    The planned layer would acquire source-bound data from scientific databases, open-data portals, APIs, government sources, repositories, and credible web sources.

    • Never invent or hallucinate measurements
    • Record source, retrieval date, and available license
    • Record unit, measurement method, and data quality
    • Document missingness and transformations
    • Expose conflicting sources
    • When evidence is insufficient: EVIDENCE_MISSING
    • When quality is insufficient: DATA_QUALITY_INSUFFICIENT
    • Keep every AI-assisted analysis reproducible
    02F / HIGH Β· CORE STUDIO 3.0 CAPABILITY Β· PLANNED

    EVIDENCE-BASED EWS RECOMMENDATION LAYER

    Studio 3.0 / ECHO is planned to compare available indicators under user-visible criteria such as warning lead time, detection latency, robustness, noise sensitivity, parameter stability, false-alarm behavior, temporal consistency, uncertainty, indicator agreement, and - where available - ground truth.

    STUDIO 2.0

    Calculation and structured evidence

    Data β†’ selected available EWS β†’ calculation and visualization β†’ researcher interpretation. The fixed Reference Benchmark remains an installation and reproducibility check.

    STUDIO 3.0 / ECHO Β· PLANNED

    Assessment without a forced winner

    One preferred EWS Β· complementary multi-EWS evidence Β· NO_RELIABLE_RECOMMENDATION / INDETERMINATE. Scientific responsibility remains with the researcher.

    Open the documented 3.0 outlook β†’
  4. 04LONG-TERM RESEARCHLONG-TERM SCIENTIFIC HORIZONLong-Term HorizonA law-like relationship β€” only if the evidence supports itOpen details οΌ‹

    Beyond Studio 3.0, the program would investigate whether loss of effective degrees of freedom, structural compression, increasing persistence, loss of observability, and critical transitions are connected by a general mathematical relationship.

    1. OBSERVATION
    2. INDICATOR
    3. MECHANISM
    4. INVARIANCE
    5. LAW

    A law-like or potentially natural-law relationship is a possible long-term objective, never a conclusion assumed in advance.

    Not by intention, but only if mathematical derivation, cross-domain invariance, and independent evidence support it.
    Examine the theoretical horizon β†’

02 / Self-correction implementation path

Formalize first. Implement second. Validate independently.

CURRENT / NEAR TERM

Formalize the construct, align existing modules, publish the positioning paper, and benchmark candidate loss indicators.

STUDIO 3.0 Β· PLANNED

Multidimensional diagnostics, domain adapters, network and real-time integration, and explicit recovery analysis.

RESEARCH HORIZON

Independent replication, domain validation, recovery benchmarking, and possible standards only if evidence supports them.

Open the canonical self-correction architecture β†’

03 / Scientific interpretability

Scientific interpretability is not an add-on.

Every future BenchEWS analysis should expose not only numbers, but also its reasoning path and boundaries.

  1. 01What was investigated?
  2. 02What was calculated?
  3. 03Which method was used?
  4. 04Which data were used?
  5. 05Why was this method selected?
  6. 06Is the input suitable?
  7. 07What evidence is available?
  8. 08What uncertainty remains?
  9. 09What warnings apply?
  10. 10What does the result mean?
  11. 11What alternative explanations exist?
  12. 12What must explicitly not be inferred?

No artificial certainty Β· no fake scores Β· no unsupported alarm thresholds

04 / Connected evidence

Move directly from the roadmap to the inspectable foundations

The roadmap also serves as navigation. Each area links to its existing primary page, avoiding unnecessary duplication.