Observational layer. Tracks where and how a leader's ideas appear across twelve verified channels in three tiers. Tier 3 is mathematically capped regardless of follower volume.
Quality gate. An NLP filter applied to every leader's content corpus. Penalises AI-generated boilerplate; rewards data-dense, structurally original output.
Describes the shape of influence, not its magnitude. A supplementary diagnostic across five dimensions; it does not feed the Final Score.
A global forum invites or a sovereign body appoints. Cannot be purchased, manufactured, or gamed.
Requires public articulation and unscripted defence of ideas before domain-credible audiences.
Near-zero barrier to entry. Two million followers cannot break into the Authority tier without Tier 1 substance.
Evidence-anchored; hand-verified for the top cohort, rules-based below. The method's limits are stated here, not obscured — they are the instrument's structural integrity, not its weaknesses.
The Index begins from a broad candidate longlist — more than 3,000 names in the 2026 cycle, drawn from public directories, institutional rolls, media and conference records, and AI-assisted discovery. That longlist performs one function: recall. It confers no standing. No score, rank, or tier is inherited from the source; every published entry is independently verified for identity, role, and organisation, then scored from scratch under the fixed protocol. The discovery layer is deliberately wide and openly imperfect — its failure mode is omission, which the bi-annual review catches, not inclusion of the unqualified, which the engine screens out.
Propagation is measured through documented, largely public, largely English-language channels. The consequence is directional: authority exercised in public is easier to evidence than authority exercised privately or in Nigerian languages. The Index therefore measures propagated authority, not latent authority. This is the same mechanism behind the propagation gap reported for women — documented channel access, not underlying rigour — and it extends to private principals and vernacular thought leaders. We disclose it rather than obscure it.
The Index excludes partisan political actors; for serving officials the test is independent intellectual standing, not portfolio. Excluded: any active partisan politician or candidate for elective office, and any office-holder whose prominence exists only by virtue of the office. Included — scored on ideas, never on office — are officials and agency heads who carry an independently documented body of propagated thought: leaders who would rank on their published record alone. A minister with an internationally recognised body of policy scholarship is in on that record; an office-holder with no record beyond the office is not.
As with every Podhium instrument, the GREY Capital Index carries a stated right of reply for rated entities and a confidence grade on the underlying evidence — and its scores are comparable across the index family: a 74 here reads as a 74 in any Podhium index.
The discipline that keeps it current, contestable, and honest.
Each cycle re-verifies roles and organisations as of the edition date, re-runs the evidence pass on the ranked cohort, and publishes a Movers Report tracking whose standing shifted and why. Roles, titles, and affiliations are point-in-time statements — accurate as of the edition date and re-confirmed every cycle. From the June 2027 cycle, candidate discovery and evidence submission move to a self-reported, platform-native model.
Every rated entity may review the evidence base behind their entry, submit corrections or additional documentation, and receive a reasoned response before the next cycle closes.
Confidence grades. Each entry carries an internal grade on the depth of its underlying evidence — hand-verified entries grade highest — disclosed to the rated entity on request.
Probity rule. An entry may be suspended or removed where credible legal, regulatory, or probity concerns arise, and reinstated on resolution.