# A filing count outranks a poll headline

[Journal](/journal/)

August 9, 2026 · [evidence](/journal/#evidence)



Finance governance looks thin in two counts. Public filings support that finding, while an informal poll supplies only a warning.

![A light poll tray balances against a heavier filing sorter that routes records into a formal-policy tray.](/_astro/a-filing-count-outranks-a-poll-headline.COk43y0v_1SHxJX.avif)

The filing count supports a finance governance gap. A poll headline points in the same direction, but its missing denominator keeps it from carrying equal weight.

## The poll supplies a warning, not a measure

An [informal LinkedIn poll from May 2026](https://arxiv.org/abs/2608.02311) asked finance professionals about operational governance frameworks for agent systems. [Eighty-eight percent](https://arxiv.org/abs/2608.02311) reported that they had none.

The authors call the poll [self-selected](https://arxiv.org/abs/2608.02311). They [report no sample size](https://arxiv.org/abs/2608.02311), so a reader cannot recover its denominator or judge its sampling error.

That leaves 88% as a warning from an informal audience. It does not measure the finance industry.

## The filings carry the finding

The authors [reviewed Form ADV filings from 100 of the largest U.S. money managers](https://arxiv.org/abs/2608.02311). Seventy-five [disclosed use of agent systems](https://arxiv.org/abs/2608.02311).

Among those 75 managers, [24 reported a formal governance policy](https://arxiv.org/abs/2608.02311). That count fixes both the reviewed population and the relevant subgroup.

| Evidence | Population | Result | Limit |
| --- | --- | --- | --- |
| LinkedIn poll | Not reported | 88% reported no operational framework | Self-selected |
| Form ADV review | 100 managers, 75 disclosing use | 24 of 75 reported a formal policy | Disclosure count |

A filing review still measures disclosure, not every control a manager runs. Yet its records let a reader inspect the population and recount the result.

That makes 24 of 75 the number that holds. It supports a narrow claim: formal governance disclosures lag disclosed use among the managers reviewed.

## A proposal does not close the gap

The paper [proposes policy, engineering, composition, and systemic layers](https://arxiv.org/abs/2608.02311). It also [sets out a 90-day implementation sequence](https://arxiv.org/abs/2608.02311).

Those layers and that sequence remain proposals. The paper [reports no production adoption of either](https://arxiv.org/abs/2608.02311).

Neither paper author is a Muniment customer or endorser.

Muniment’s position is that governance claims need records with denominators, scope, and methods attached. A large percentage cannot supply what its source never recorded.

[Join the waitlist](/waitlist/?utm_campaign=a-filing-count-outranks-a-poll-headline) if your governance records need to preserve the evidence behind each claim.

## Sources

1.  [AI Governance for Institutional Readiness in Finance](https://arxiv.org/abs/2608.02311) arxiv.org