Here's my allocation, with reasoning for each:

Business leaders. The bottom line audits them regularly — bad choices eventually show up in profits, so C matters a lot. But most of a CEO's job is persuading: employees, boards, investors, customers. B is high. A helps (money, pedigree, commanding presence) but isn't decisive.

Political leaders. Audiences can rarely evaluate policy substance directly, so they rely on cues: does this person look, sound, and carry themselves like a leader (A), and do they make me feel allied and heard (B)? C is low not because good policy is unimportant, but because it's rarely what gets you the role — voters can't easily tell good choices from bad ones, and many policy "choices" are delegated to staff anyway.

Cultural activist leaders. The most A/B-heavy category. Movements compete for attention, so fame, looks, charisma, and media savvy (A) and coalition-building, flattery, and power projection within fractious movements (B) dominate. The actual content — slogans, strategies, theories — is often interchangeable; several movements succeed with weak arguments if the leader is compelling.

Intellectual/academic leaders. The closest to a genuine meritocracy of ideas: papers and arguments are published, cited, and criticized publicly, so C dominates. But don't underestimate the rest — charisma, wit, and connections (A) and befriending the right editors, funders, and senior scholars (B) explain a lot of who becomes a "leading intellectual" versus an equally smart unknown.

Innovation leaders. Similar C-dominance to academics, but for a different reason: the product either works or it doesn't, and users don't care about the founder's pedigree. A is lowest here — look at how many top innovators were awkward, unconnected outsiders. B still matters a fair bit because raising money, recruiting talent, and selling the vision are social acts, and many technically brilliant people fail to lead because they can't do them.

Two general observations:

For short-run selection and influence — who gets chosen, followed, funded, and listened to now, regardless of how history judges them — the weights shift substantially away from substance and toward signaling:

Business. Boards and investors can't evaluate a CEO's actual decision quality in real time — results take years and are confounded by market conditions. So they select on track record signals, pedigree, presence (A) and how the candidate interviews and networks (B). The C that matters short-run is mostly sounding like you make good choices, which is partly a B skill.

Political. Elections and party contests are decided before anyone can audit policy outcomes, so selection runs almost entirely on cues of leader-likeness (A) and on making voters, donors, and party elites feel heard and allied (B). C shows up mainly as slogan choice — which is as much about what resonates socially as what's right.

Cultural activists. Highest A+B of all. Movements pick leaders fast, based on who commands attention and media (A) and who can hold coalitions of prickly allies together (B). Whether the movement's arguments are good is nearly irrelevant to who rises — followers can't evaluate that quickly, and often don't try.

Intellectual/academic. Even here, short-run influence is heavily social: whose talks get invited, whose papers get read, depends on name recognition, institutional affiliation, and charisma (A) plus networking at conferences and currying favor with gatekeepers (B). C still gets the largest share of any category, because ideas are at least partially inspectable immediately — a talk or paper can be visibly clever on first exposure, unlike a policy.

Innovation. Short-run selection happens in fundraising and recruiting pitches, where founders are judged on confidence, storytelling (B) and prior exits, degrees, connections (A) far more than on the actual design, which most investors can't evaluate. C stays relatively high only because a working demo or prototype is one of the few short-run-visible signals of substance in any of these domains.

The meta-point: in the short run, almost every domain selects on signals of C rather than C itself. A and B are what make signals legible and persuasive. C only retains weight where reality provides a fast, inspectable feedback loop — demos, proofs, visible artifacts. Everywhere else, short-run influence is a status and social-skills contest wearing the costume of a substance contest.

Let me look into the evidence on this.

Let me check one more angle — how much social capital matters inside science and innovation specifically.

The evidence pushed back on my assumption more than I expected. Here's what I found, and the revision it warrants:

Academia: The cleanest causal test of status in science — tracking citation changes to papers written before their authors received prestigious HHMI appointments — finds a real but modest status premium: about 12% more citations for identical-quality work . Crucially, though, the status effect was much larger under uncertainty: for younger researchers, lower-profile journals, and novel research areas . A separate decomposition study found only weak prestige effects on citations once paper characteristics are controlled . So for recognition of published work, substance genuinely dominates. But for career selection — who gets hired, funded, tenured, invited — the picture is different: extraversion and networking robustly predict academic career success through social capital, mentoring access, and collaboration formation , and people with strong communication skills get recognized as experts regardless of actual qualifications .

Innovation: My original numbers were clearly too kind to substance. VCs overwhelmingly prioritize founder pedigree first when evaluating opportunities, above business model or path to profitability ; one fund documented a Harvard founder with just an idea raising at a $15M valuation while a comparable startup with actual revenue but mid-tier-university founders raised far less at $6M . VCs rate proposals more highly when founders come through their networks , and classic interview studies found founder experience and managerial capability predicted VC-backed success better than technical/product criteria . Even credit for innovation is status-allocated: hierarchical position in research groups strongly affects who gets named as a patent inventor .

The key theoretical point I got wrong: I had argued C keeps weight "where reality provides fast, inspectable feedback." But the status literature shows status signals matter most where quality is uncertain — and early-stage startups and novel research are among the highest-uncertainty products in existence. By that logic, innovation and frontier academia should be more status-driven, not less. The inspectability advantage (papers, demos, citations) is real but partial, and mostly kicks in later than short-run selection.

Residual reasons C still gets more weight in these two than elsewhere: work is at least sometimes directly inspectable (a paper can be read, a demo can be run) ; citation markets give partial quality feedback that election and movement markets lack ; and the measured pure-status premium on fixed-quality work, while real, is modest — suggesting audiences in these domains do substantially anchor on the artifact itself once it exists . But short-run selection in both domains — funding, hiring, credit, invitations — happens before that anchoring, and the evidence says it runs heavily on pedigree, connections, and social skill, much like everywhere else.