FDE BenchmarkPublic document reader
Available in sourcev0.1 · live public draft · official release pending

Canonical source view. This renders the bundled repository document as readable HTML. Source metadata remains in the repository, and relative document links remain inside this reader.

Project Charter

Status: v0.1 Public Draft
Founding steward: LockedIn Labs
Governance bodies: Proposed; see current status

This charter defines the mission, scope, and constitutional commitments of the LockedIn FDE Benchmark project. It is normative for project governance. The Independence and Institutional Firewall defines the fail-closed activation path for future human assessment and credentialing; its presence does not mean those controls are operational.

1. Mission

The project exists to develop trustworthy measurement infrastructure for demonstrated Forward Deployed Engineering capability.

It seeks to make evidence, uncertainty, verification, review, and limitations more legible without reducing a person to a title, résumé, test score, or popularity signal.

2. Founding proposition

Forward Deployed Engineering is demonstrated, not declared.

The benchmark therefore prioritizes relevant demonstrated work, controlled observation, production operation, corroboration, and accountable review over self-description alone.

3. Scope

The project may steward:

  • the FDE construct and versioned capability ontology;
  • evidence, verification, evaluation, and scoring specifications;
  • public schemas and safe reference implementations;
  • practice cases and controlled official-evaluation interfaces;
  • governance, conflict, appeal, research, and integrity processes;
  • disclosure-filtered profile and portable-credential specifications;
  • aggregate, privacy-preserving benchmark research.

The project does not operate as:

  • a training authority that awards credit for participation;
  • an employer or staffing decision maker;
  • a recruiting or staffing funnel embedded in, or presented as, neutral benchmark authority;
  • a vendor, sponsor, or platform ranking program;
  • a public human leaderboard;
  • a guarantee of hiring success or production performance;
  • an accredited personnel-certification scheme unless a later, separately governed program meets the applicable requirements.

4. Constitutional commitments

Every project release and operating program must preserve these commitments:

  1. Evidence over titles. Claims and prestige are not substitutes for relevant demonstration.
  2. Capability is not confidence. Capability, support confidence, coverage, maturity, verification, and result class remain separate.
  3. Unknown is not weak. Missing evidence produces an insufficient-evidence state.
  4. AI assists; people govern. AI may propose typed observations but cannot award a final score, verification, credential, or appeal outcome.
  5. Rules are versioned. Scores are deterministic projections from released rules and eligible adjudicated inputs.
  6. History is auditable. Material events are append-only; corrections and revocations create new events rather than silent overwrites.
  7. Privacy is a boundary. Public methodology does not require public evidence. Confidential material remains controlled and use-limited.
  8. Critical weakness is visible. Averaging cannot erase a consequential security, integrity, production, or forward-deployed deficiency.
  9. No preferential inputs. Employment by LockedIn, LockedIn training, use of any particular tool, payment, sponsorship, friendship, education prestige, employer prestige, and fame do not add score value.
  10. High-consequence use is constrained. Automated adverse employment decisions are prohibited; future employment use requires job relevance, human review, accommodation, auditability, correction, fairness monitoring, and legal review.
  11. Funding does not buy outcomes or governance. Sponsors receive no scores, thresholds, votes, hidden cases, appeals influence, or endorsements in exchange for support.
  12. The benchmark is benchmarked. Reliability, validity, fairness, accessibility, contamination, gaming, evaluator drift, and field relevance are measured and reported with limitations.
  13. Commercial roles cannot confer benchmark authority. Training, employment, staffing, recruiting, sales, customer, platform, and sponsor relationships are separated from affected verification, measurement, credential, appeal, and publication decisions. Any future official human output requires external-majority governance and the institutional firewall.

These commitments may be clarified but not weakened through an ordinary release vote.

5. Open and controlled boundaries

The public project should expose the construct, ontology, public evidence hierarchy, scoring principles, versioning, governance, limitations, schemas, sample cases, and safe reference interfaces.

It must control or keep private anything whose publication would materially compromise privacy, intellectual property, or evaluation integrity, including private evidence, identity records, hidden cases, holdouts, fraud heuristics, reviewer records, confidential prompts, customer artifacts, and security-sensitive operating details.

The existence of controlled assets must be disclosed at the category level. Their contents need not be public.

6. Stewardship and authority

LockedIn Labs is the founding steward and may fund, staff, publish, and operate the initial project. Stewardship does not confer authority to:

  • assign or edit an individual score;
  • exempt a founder, executive, employee, trainee, customer, sponsor, or partner from the released protocol;
  • represent an outside organization as an endorser without written authorization;
  • convert sponsor funding into governance votes or preferential treatment;
  • allow training, employment, staffing, recruiting, sales, engagement, or customer-service personnel to control an affected human result, credential, or appeal;
  • publish confidential evidence without the required rights and consent;
  • call a draft validated, independent, accredited, or standard-setting without evidence.

During the transitional v0.1 period, the founding steward records interim decisions publicly. Official human scoring, tiers, credentials, and verification authority cannot begin under transitional unilateral governance. Public materials may state that official scoring is disabled; they may not use “official” as an unqualified marketing adjective.

7. Independence and conflicts

Independence is demonstrated through process, not naming. The project must publish decision records, conflicts, recusals, release votes, limitations, and financial relationships at an appropriate level.

Verified future results require a non-conflicted review path. Founder and executive results require external review. Appeal decision makers must be separate from the original disputed decision.

Before any official human assessment, result, tier, or credential release:

  • an external-majority FDE Benchmark Commission with an external chair and independent quorum must hold protected decision authority;
  • role, data, compensation, funding, recusal, review, correction, and appeal firewalls must operate and be tested;
  • the Commission must publish a structural-independence decision evaluating a hosted program, separately governed affiliate, independent entity, and external certification-body path; and
  • the exact issuer, accountable legal entity, release, intended use, validity evidence, limitations, and lifecycle must be public.

External-majority governance is a constitutional safeguard selected for this project. Separate incorporation is a structural option, not a universal scientific or legal claim. A new entity does not by itself create independence, and a hosted program may not call itself organizationally independent merely because it has external advisors.

Requirements are specified in INDEPENDENCE_FIREWALL.md, CONFLICT_OF_INTEREST.md, and GOVERNANCE.md.

8. Project assets and intellectual property

Public code, content, and schema permissions must be stated through explicit license files. Open licenses do not grant trademark rights.

Private cases, evaluator assets, customer evidence, adjudication data, calibration data, and fraud controls may remain controlled where permitted by law and participant agreements. Controlled status cannot be used to hide material limitations or undisclosed score influence.

See TRADEMARK_POLICY.md and DATA_GOVERNANCE.md.

9. Amendment

Charter amendments require:

  1. a public RFC describing the change and its trust impact;
  2. at least 30 calendar days of public review, except a documented legal or security emergency;
  3. conflict review and recorded recusals;
  4. approval under the supermajority rule in GOVERNANCE.md;
  5. a changelog entry and new charter version.

During transitional v0.1 governance, the founding steward may propose and adopt an amendment only after the same public-review period and a written decision record. Such an interim amendment cannot activate official scoring or weaken the constitutional commitments.

10. Relationship to other documents

If an implementation conflicts with this charter, it cannot claim project authorization merely because it passes a schema test.