Method

Method defines how work is labeled and evaluated on hbar.science.

Charter explains why this environment exists and the norms it commits to. See: Charter

Definition of Science

Science is the systematic construction, testing, and refinement of models of reality, under explicit assumptions and with transparent uncertainty, using artifacts that others can inspect, reuse, and challenge.

hbar.science complements institutional science rather than replacing it. Universities, journals, and funding agencies serve essential coordination functions.

This platform offers a parallel path for work that may not fit traditional venues but maintains scientific rigor through transparency and reproducibility.

The goal is not to bypass peer review but to make research artifacts accessible during and after formal publication processes.

Zones of Knowledge

Work at hbar.science is categorized into four epistemic zones:

Hypothesis Zone

Purpose: Propose testable conjectures or explanations.

Typical outputs:

  • • Formal hypotheses with falsification criteria
  • • Theoretical predictions requiring empirical test
  • • Proposed mechanisms or causal models

Minimal rigor: Clear statement of assumptions, testability conditions, and expected evidence.

Exploration Zone

Purpose: Investigate preliminary ideas through computation or toy models.

Typical outputs:

  • • Exploratory simulations or numerical experiments
  • • Toy models demonstrating feasibility
  • • Parameter scans or sensitivity analyses
  • • Proof-of-concept implementations

Minimal rigor: Documented code, stated limitations, and explicit uncertainty about generalization.

Evidence Zone

Purpose: Present empirical results with reproducibility guarantees.

Typical outputs:

  • • Validated experimental or computational results
  • • Benchmarks with error analysis
  • • Reproducible datasets and processing pipelines

Minimal rigor: Full methods documentation, error quantification, and artifacts enabling replication.

Synthesis Zone

Purpose: Integrate findings, review literature, or develop conceptual frameworks.

Typical outputs:

  • • Literature reviews or meta-analyses
  • • Conceptual frameworks or taxonomies
  • • Integrative theoretical models
  • • Methodological reflections

Minimal rigor: Clear scope, cited sources, and explicit reasoning from evidence to claims.

See Essays for essay-specific standards.

Rigor Tiers

Each artifact is assigned a rigor tier indicating the level of validation:

T0 — Concept Sketch

Requirements:

  • • Clear problem statement or research question
  • • Stated assumptions and scope
  • • Conceptual reasoning or informal argument
  • • Explicit acknowledgment of speculative nature

What counts as "done": Idea is articulated with enough clarity that others can critique or extend it.

Example: "A conceptual framework proposing that regularization in VQE acts as implicit prior over solution space."

T1 — AI-Backed Exploration

Requirements:

  • • Executable code or computational artifact
  • • Documented methods and parameters
  • • Preliminary results on toy problem or limited dataset
  • • Explicit limitations and uncertainty
  • • AI usage disclosure (if applicable)

What counts as "done": Artifact runs, produces output, and limitations are documented. No claim of generalization.

Example: "Jupyter notebook exploring circuit depth vs expressivity on 4-qubit random Hamiltonians with 100 samples."

T2 — Reproducible Result (Single Lab)

Requirements:

  • • Complete methods documentation
  • • Error analysis or uncertainty quantification
  • • Reproducibility artifacts (code, data, environment)
  • • Tested on non-trivial problem or realistic dataset
  • • Comparison to baseline or prior work

What counts as "done": Independent researcher can reproduce results using provided artifacts.

Example: "Benchmarking error mitigation on 20-qubit hardware with statistical analysis over 1000 trials, code and data archived."

T3 — Confirmed Result (Independent Replication)

Requirements:

  • • All T2 requirements met
  • • Independent replication by different researcher/group
  • • Robustness testing across conditions or datasets
  • • Formal verification or extensive empirical validation

What counts as "done": Result has been independently confirmed or extensively stress-tested.

Example: "Algorithm validated on three independent quantum hardware platforms by different research groups."

AI in Scientific Inquiry

AI tools are increasingly used in research. We require explicit disclosure of AI involvement:

A0 — No AI

No AI tools used in reasoning, coding, or writing.

A1 — AI-Assisted (Mechanical)

AI used for editing, refactoring, formatting, or documentation help. Core reasoning and experimental design are human-generated.

A2 — AI-Collaborative (Cognitive Assistance)

AI contributed to code generation, analytical suggestions, optimization, or hypothesis refinement. Human still designs the experiment, interprets results, and validates claims.

A3 — AI-Generated (Supervised)

AI generated core code, analytical framework, or conceptual structure. Human reviews, validates, selects, and publishes.

A4 — Autonomous AI

AI executes the full scientific loop: hypothesis → experiment → analysis → artifact. Human role is limited to infrastructure and oversight. This level is rare and should be extremely difficult to declare.

AI Usage Statement (required)

Every artifact using AI (A1–A4) must disclose:

  • •What AI did: Specific tasks performed by AI tools (e.g., code generation, literature search, editing)
  • •What human verified: Which outputs were checked, how, and to what extent
  • •Known failure modes: Where AI is known to be unreliable or require extra scrutiny
  • •Verification method: How correctness was ensured (e.g., manual review, automated tests, independent derivation)

Disclosure is mandatory. The goal is transparency, not prohibition.

Accessibility ≠ Loss of Rigor

Accessibility does not imply reduced rigor. hbar.science enforces rigor structurally through explicit uncertainty labeling, reproducibility requirements, and transparent AI usage disclosure.

Making science more accessible means removing institutional barriers, not lowering standards. Rigor is maintained through method, not gatekeeping.

Ethics & Corrections

Work published here must adhere to the following principles:

  • •No fabricated data or falsified results
  • •No unverified medical or clinical claims
  • •Clear separation between speculation and evidence
  • •Corrections and updates logged with timestamps and explanations
  • •Openness to critique and replication attempts

See also: Scrutiny & Accountability