Case Study XIV — The Investor Assembly

Placing capital allocation under deliberation: an institutional route toward Levinasian ethics

Abstract. Three asset managers now account for roughly three in ten of the votes cast at a median S&P 500 shareholder meeting. That power lacks a principled justification, and the remedies tried so far have not resolved the problem. This case study argues that the Investor Assembly proposed by Oliver Hart, Hélène Landemore and Luigi Zingales¹ is the strongest institutional answer now on the table, and that its deepest justification is ethical rather than economic: it is the first concrete design that subordinates the allocation of capital to a body that deliberates face to face on what markets cannot register. Section I states the problem. Section II presents the remedy. Section III gives the Levinasian analysis, including what Levinas requires the design to add. Section IV sets out the legal pathway and answers the strongest objections. The proposal is new, advisory in its first form, and exposed to specific legal hurdles. This paper treats those hurdles as engineering problems, not refutations.

Contents

1. The Problem: Capital Allocation Without Open Regard
2. The Remedy: An Assembly That Directs the Allocator
3. The Levinasian Analysis
4. Implementation, Legal Pathway, and Objections
5. Conclusion: The Hinge

Terms used in this paper

• Regard. The way one attends to something and holds it in esteem, as in “having regard for.” This paper distinguishes two kinds.
• Open regard. Attention to the relational whole that sustains the parts; the capacity to receive what is other as other. This is hospitality.
• Focal regard. Attention to the part: its task, its gain, its measure, within a fabric it did not make. In McGilchrist’s account this is the left hemisphere. It is necessary and legitimate within its bound; the pathology is its usurpation of the whole.
• Dispositions, not persons. Any individual or institution can hold either. The asymmetry is directional: open regard has ontological, not temporal, priority.²
• Value / values. Following Hart and Zingales: “value” is financial worth; “values” are ethical and political commitments. A value-values trade-off is a choice to give up some of the first for the second.
• Universal owner. An investor who holds the whole market, typically through index funds, and so owns a slice of nearly every company.
• Sortition. Selection of decision-makers by lot.
• Face, third party, justice (Levinas). The Other’s face commands before any calculation. When a third party appears alongside the Other, comparison and judgment become unavoidable; institutions of justice arise from that necessity and remain answerable to the ethical relation that produced them.

I. The Problem: Capital Allocation Without Open Regard

I.A — Concentrated voting power and its legitimacy deficit

Index funds held about 0.5% of S&P 500 capitalization in 1975 and about 19% by the end of 2024. Over a quarter-century Vanguard, BlackRock and State Street rose from 12% of the mutual fund industry to 36%. Their combined median stake in an S&P 500 company is now about 22%, and their share of the votes actually cast at a median meeting is about 29%.³ Power of this size behaves like regulation: when one of them campaigned for women on boards, hundreds of companies complied. In a democracy, regulation comes with procedures that justify it. Here there are none.

Hart, Landemore and Zingales (HLZ) distinguish internal legitimacy (toward the investors whose money is being voted) from external legitimacy (toward the public affected by those votes), and, within each, input (who decides), throughput (how) and output (with what results). The Big Three fail on both. The votes are cast by small professional teams under guidelines investors never saw, and the public never authorized them. The criticism arrives from opposite directions: some state officials call the power undemocratic, while others call its use inadequate.⁴ A power attacked in the same terms from both sides has no foundation of its own.

The universal owner owns a share of the whole economy, which is the structural position of one who ought to attend to the fabric, yet its working disposition is trained on measured return. The legitimacy crisis is what that mismatch looks like from outside.

I.B — The perceptual ceiling

The fiduciary defense collapses. The industry’s fallback is that fiduciary duty leaves no discretion: the only task is to maximize return. But return maximization is not an objective standard. On environmental and social resolutions at S&P 100 companies between 2021 and 2023, the Big Three voted the same way only 31% of the time; Vanguard opposed 72% of them, BlackRock 45%, State Street 28%. Separately, credit analysts’ forecasts of the economy flip with the party of the president, so even the “neutral” estimate of value carries values inside it. And Friedman’s separation of profit from social preference holds only when a firm’s money-making and its social effects are independent, as with a charitable donation. It fails when the profit activity produces the harm, as with pollution or monopoly pricing of a vaccine.⁵ Every vote, therefore, is already a value-values judgment. The only open question is whose.

The best utilitarian repair still hits a ceiling. The Shareholder Commons (TSC) is the most sophisticated attempt to repair the system from inside. Its logic is “beta stewardship”: a diversified owner’s return depends mostly on the whole market, so a company that raises its own profit by exporting costs to the economy can lower its owners’ portfolio return, and universal owners should vote to stop it.⁶ This is a real advance, and it is correct as far as it goes. But it extends the horizon (thirty years instead of three months) and the scope (the whole market instead of one stock) while keeping the same grammar: human and ecological goods enter only as inputs to a portfolio calculation. This is a quantitative change, not a qualitative one. A framework whose only moral vocabulary is expected return cannot, in principle, oppose an act that its model shows to be net-profitable for the portfolio. Scheler’s ordering of values ranks the useful and the pleasant beneath the vital, the spiritual and the sacred, and Heidegger’s “standing reserve” names what happens to persons and places when everything is rendered as a resource to be optimized.⁷ In McGilchrist’s terms, finance operates through a mode of attention that grasps, isolates and measures, and so cannot “presence” whatever is not measurable.

Untracked does not mean unreal. The decisive point is this: a value the index cannot register still acts on the index. Blindness belongs to the instrument, not to the world. Erosion of trust, withdrawn labor, political backlash and legislative intervention all reach the portfolio through channels the model has no variable for, until the moment they arrive as losses. HLZ’s own argument is an instance: the legitimacy deficit of the Big Three was never priced, and the political system is now pricing it for them through breakup proposals and antitrust suits. An allocation mechanism that cannot see these channels will misprice them; an institution that can deliberate about them is a corrective, not a distortion.

I.C — Why the standard remedies jam

The table compares the main remedies on a single question: what stops each one. The assessments are this paper’s own, drawn from the sources in the footnotes.

Remedy	How it was meant to work	Where it stops
Legislation and regulation	Democratically enacted rules set the limits.	Hard to pass in a polarized legislature; both camps now pressure the funds instead of legislating (HLZ).
Fiduciary-duty litigation	Courts read directors’ duties to cover the diversified portfolio.	McRitchie v. Zuckerberg was dismissed in 2024; Delaware treats duties as firm-specific.8
Beta stewardship (TSC)	Universal owners vote against externality-exporting strategies.	Depends on managers who have mostly declined; justifies ethics only as long-run return; exposed to the antitrust critique (IV.A).
Pass-through voting	Each investor directs the votes on his or her shares.	Rational apathy: only 2% of eligible Vanguard retail investors made a choice, and 40% of those delegated back; the manager still writes the menu.9
Exit / divestment	Sell what you reject.	Index funds cannot exit by construction, and most 401(k) savers have no alternative vehicle (HLZ).
Elected investor representatives	Delegates write the guidelines and face re-election.	Rational apathy again: few would read candidate statements or vote; accountability would be toothless (HLZ).
Investor Assembly	A paid, informed, randomly drawn sample of the fund’s own investors deliberates and sets guidelines.	Unproven at scale; advisory in its first form; specific legal hurdles (IV.A). None of these is structural in the way the rows above are.

To this paper’s knowledge, no other proposal answers all of these constraints at once. The Investor Assembly does not ask the managers to become virtuous, the courts to reinterpret duty, the legislature to overcome capture, or the public to become informed investors. It requires only that a fund change a default.

II. The Remedy: An Assembly That Directs the Allocator

II.A — The case for sortition

Aristotle classed selection by lot as democratic and election as oligarchic. Jeremy Lent restates the case: elections select for charisma, wealth, ambition and connection, which is why campaign-finance rules and “better candidates” leave the structure intact, whereas sortition resists corruption, blocks the concentration of power and spreads the work of governing.¹⁰ Luke Kemp observes that almost no money goes into innovating how democracy represents people at scale, and that rewarding such work like AI startups would change that. Roman Krznaric points to the new wave of citizens’ assemblies, which tend to produce bolder, longer-horizon proposals than politicians working short cycles, and which build social trust.¹¹

This is not speculative. The OECD had documented 733 deliberative minipublics in 34 countries by 2023, including national assemblies in Ireland, France and British Columbia; several Irish assemblies’ conclusions were later endorsed by referendum. Juries, the oldest form, agree with judges about 75% of the time in the classic study, and countries with jury trials show less perceived corruption.¹²

II.B — Lateralised governance: the structural argument

Hutchinson and Yarrow argue that modern governance is “half-brained”: it attends overwhelmingly to power, control and coercion, the grasping activities McGilchrist assigns to the left hemisphere, and without right-hemisphere input it becomes easy prey for lobbyists pressing narrow interests.¹³ McGilchrist makes the organizational point directly: bureaucrats and those who hold the big picture must be different people, and the bureaucrats must serve the whole. He has watched bureaucracy in universities and hospitals expand like a neoplasm, take resources from the host organization it exists to serve, and begin instructing the doctors and professors.¹⁴ Schmachtenberger’s demand that wisdom bind power, and Stein’s that values guide technology, express the same requirement.¹⁵

This paper proposes a name for the mechanism: sortition-binding-power. A body chosen by lot sits above the managerial and elected hierarchy it directs, not beneath it. The inversion matters. Where officials can bind a citizens’ assembly, the assembly is decoration; where the assembly binds officials, the managerial layer is returned to service. HLZ’s formula that experts be “on tap, not on top” is the same structure in procedural language.

II.C — The design as proposed

The HLZ blueprint is specific enough to audit.¹⁶ The assembly is organized at the fund level, not the company level, so that decisions are made at the portfolio level, where externalities between companies become visible. The main features:

• Membership. 150 investors drawn by lottery, stratified so that a decliner can be replaced by a similar person; paid for their time; each investor drawn at most once. Selection odds are proportional to holdings (“one share, one lottery ticket”), but inside the room every member has equal voting power and status. At 150 members, a 2% minority has a 95.2% chance of at least one seat.
• Opt-out. Investors who want to decide for themselves, such as large pension funds, can opt out and use pass-through voting. HLZ propose that the right persist even after the assembly has reported, because investors may be morally outraged by a guideline. Because the default changes rather than the investment contract, funds need not renegotiate with anyone.
• Process. Annual, in the fall before proxy season. A first weekend in fifteen groups of ten, reshuffled midway so that members meet widely; the assembly sets its own agenda; a ten-person governance committee chooses what information and which experts the full body hears; a month of materials; a second weekend of deliberation. The output is a set of voting guidelines, ranging from specific (curb emissions while cost stays under a stated figure per ton) to general (prevent undue animal suffering).
• Decision rule. A 60% quorum. HLZ’s sampling calculation shows that, at 150 members, a position held by 49% of investors wins a 60% vote about 0.1% of the time.
• Authority. At first, consultative: the fund keeps the right to overrule but must justify any overruling in detail. HLZ expect escalation if the guidelines prove ineffective.

A Dutch retail-sector pension fund ran a pilot in early 2024: 55 randomly drawn participants, three days in Utrecht, 49 proposals approved for the board, advisory in role.¹⁷ The proposal is brand new, and the paper’s own authors say so. But an advisory pilot is the correct first step, not a defect: it is how a design earns evidence.

II.D — Why this is capitalism’s own lever

The reflexive objection is that an assembly is socialism by lottery. The reverse is closer to the truth. Three arguments:

(1) It restores property rights. Today an intermediary casts your retirement capital’s votes on matters you may find abhorrent. The assembly hands the decision to a sample of the actual owners. Weighting the lottery by shareholding reinforces the point: the more of the fund a person owns, the better the odds of a seat.

(2) It corrects the data feed of the invisible hand. Prices coordinate well only if they reflect reality. Externalities falsify them: a firm that exports costs reports profit that the economy actually paid. The assembly does not replace the allocation mechanism; it fixes the objective function the mechanism serves. This accepts the strongest claim of the market’s defenders, that allocation by prices outperforms the alternatives, and assigns it the place McGilchrist assigns the left hemisphere: indispensable in service, ruinous in command. Who sets the dial is a political question, and the assembly relocates that question from the legislature, where capture has made it unanswerable, to the fund, where owners can answer it.

(3) It pays for itself. HLZ’s budget for a 150-person, two-weekend, fully in-person assembly is about $1.1 million. Against the roughly $480 million in annual fees on Vanguard’s $1.2 trillion of S&P 500-indexed assets, that is about 0.23% of fee revenue.¹⁸ BlackRock discloses a proxy-voting team of roughly seventy people, and the assembly’s cost is small next to that existing overhead. A fund that announces an assembly may also gain a selling point, and HLZ expect it to shield managers from the political attacks of both camps. The incentive to adopt is the manager’s own.

III. The Levinasian Analysis

III.A — What the assembly instantiates

For Levinas, ethics begins when the Other’s face addresses me, and the claim precedes any calculation of benefit. His paradigmatic Others are the stranger, the widow and the orphan: those with no power to compel. But the Other is never alone. A third party is always present, and with the third come comparison, judgment, institutions, the state. Levinas calls this the birth of justice, and insists that justice remain answerable to the ethical relation it grows from, or it hardens into totality.¹⁹ An Investor Assembly is exactly such an institution of justice: a procedure for weighing claims among many.

It has five features that embody this ethic better than any existing mechanism in corporate governance:

1. Status is removed. Equal votes, anonymity during the process, and decision by what HLZ, borrowing Habermas, call “the forceless force of the better argument.” Charisma, wealth and connection, the attributes elections select, are not rewarded.
2. Encounter is built in. Fifteen tables of ten, deliberately reshuffled, are a literal arrangement of face-to-face address. Deliberation here means people listening to people, not an algorithm tallying preferences.
3. The ordinary person is received. Paid time, prepaid expenses and briefing materials are hospitality extended to a stranger drawn by chance.
4. Focal regard serves open regard. Experts are summoned and questioned by the assembly, which chooses whom to hear. The measuring mode of attention is retained, in service.
5. Judgment rests with those who bear the consequences. The value-values trade-off is made by owners who will also bear its financial cost.

III.B — What Levinas requires of the design

Levinas also exposes the design’s limits. These are stated here as specifications, not objections. Each is a place where the design, as published, stops short of what the ethic demands, and each is a modification the design can accept.

Specification 1: the absent Other. The assembly seats owners. HLZ are candid about scope: the proposal improves shareholder democracy and does not claim to resolve a company’s accountability to society at large.²⁰ The people most affected by externalities (workers whose wages are depressed, downstream communities, future generations, the nonhuman world) are by definition not in the room. They enter only as testimony. Levinas’s Other is precisely the one who cannot compel a hearing. The design should therefore add standing provisions for the absent: testimony from affected persons in their own voices, not only from experts; a designated advocate for non-owners and future generations; and a rule that every guideline be accompanied by a statement of who bears its costs and was not present. The Seventh Generation question belongs in the agenda by default.²¹

Specification 2: the said and the saying. Expert testimony delivers the said: data, estimates, cases. The Other’s claim lies in the saying: the exposure of a person to a person. A spreadsheet of system-wide costs, presented alone, risks reproducing “standing reserve” inside the assembly itself. The mitigation is HLZ’s own governance rule, under which the assembly selects its experts and information, together with the testimony provision above.

Specification 3: two senses of asymmetry. The design weights lottery odds by shareholding because capital risk is asymmetric: the owner bears the loss. Levinas’s asymmetry is a different thing: the Other’s priority over me, regardless of power. The two should not be confused. The strongest defense of the weighting is that it is the price of admission to the shareholder-democracy frame and is offset inside the room, where members hold equal status and equal votes; and HLZ argue that wealthy members’ opportunity costs lead many to decline, so the bias partly cancels itself. The specification is to publish the stratification and the gap between room and investor population, as HLZ propose, and to treat weighting as a pragmatic concession, never as an ethical principle.

Specification 4: permission, not emission. A procedure creates conditions under which responsibility can appear; it cannot manufacture it. Assemblies can err, and a narrow, focal self-interest can prevail in a given room. The blindness belongs to the same reality as the orientation that sees, and the two remain in dynamic tension. The claim for the Investor Assembly is therefore comparative and structural: it supplies anonymity, equality, information and time, which are the conditions under which open regard has room to appear, and it asks of its members no virtue they do not already have.

III.C — Open regard and focal regard in the fund

The mapping is functional. The stewardship team, the proxy advisor and the voting engine work in focal regard: skilled at execution across five hundred companies, and appropriate to that work. The assembly works in open regard: it attends to the whole body of owners and, once the specifications above are met, to those outside the room as well. HLZ’s own sequence, in which the assembly writes guidelines and the fund executes them with a duty to explain any departure, follows that order: open regard decides what the work is for, and focal regard does the work.

It also respects a distinction Levinasian design requires. At the level of orientation, nothing is worth doing except in the light of the Other, and that orientation cannot itself be weighed. At the level of institutions, weighing is unavoidable and just. The assembly weighs value against values, and does so under an orientation it cannot generate for itself. The structure repeats at every scale: fund, polity, biosphere. At each, focal regard serves, or usurps, open regard.

IV. Implementation, Legal Pathway, and Objections

IV.A — Legal hurdles

1. Fund law and fiduciary duty. Registered funds operate under the Investment Company Act of 1940. The SEC’s 2019 guidance requires voting in the client’s best interest and does not limit best interest to financial interest; clients may already direct votes on explicitly value-laden guidelines, as faith-based guidelines show. ERISA-governed retirement plans are harder: the Supreme Court’s Dudenhoeffer decision excluded nonpecuniary benefits from an ERISA fiduciary’s “exclusive purpose.” HLZ and others argue the case concerned prudence in an employee-stock-ownership context and should not be stretched to loyalty, and Fisch and Schwartz argue that managers have a duty to consult investors’ views. The ERISA question is the most serious fund-law risk and needs counsel and, possibly, a safe-harbor ruling.²²

2. Mandate and prospectus language. Index funds promise to track a named index. HLZ scope the assembly to voting guidelines and engagement, not portfolio composition, and so leave that promise intact. The index-tracking constraint is a contractual fact, not an ethical argument, and it marks the boundary of the HLZ design. Anything that changes what is held requires a different product, such as a fund whose published eligibility rules are set by an assembly (IV.D). Contractual limits of this kind are legitimate objects of design and do not show that the untracked values are unreal (I.B).

3. Delaware, and controlled companies. McRitchie concerns the duties of directors to a single firm. The assembly acts on the owner side, through the fund’s voting rights, and directors’ duties are not what it alters. To this paper’s reading, nothing in McRitchie restricts how owners may vote or what they may instruct a fund to do. McRitchie blocks court-imposed altruism by managers; it leaves intact the owners’ own tools, which are to withhold votes from or vote out directors who ignore published guidelines, and to amend bylaws. Those tools have limits that the design should plan for: in Delaware, charter amendments must be initiated by the board, and bylaws may not intrude on the board’s managerial authority, so owner action works mainly through director elections and bylaw proposals.²⁸ The more serious problem is the controller. Where one holder or family controls the vote, no fund can out-vote it. Four levers remain: (a) votes that only unaffiliated shareholders decide, such as approving an extension of a dual-class structure, which Bebchuk and Kastiel propose to make sunset-bound; (b) votes on directors and published, advance guidelines, which let would-be IPO issuers price the consequences in advance; (c) the fund’s responses to index-provider consultations, which is where eligibility rules for controlled companies are set and have been changed in both directions;²⁹ and (d) the economic record, which shows that the valuation premium of dual-class firms fades after listing and, in the later versions of the leading life-cycle study, gives way on average to a discount relative to comparable single-class firms.²³ This paper does not claim that assemblies can divest index funds from controlled companies; the HLZ design does not support that claim. Nor does it rest on investor flight to other funds as a substitute: HLZ argue that exit is unlikely to change company behavior, because the price pressure it creates attracts less scrupulous buyers, and that voice is the effective lever. What the assembly can add is public, reasoned explanation of its findings to the fund’s investors, which HLZ count as a gain in throughput legitimacy, not as an exit threat. It claims that the assembly can set the voting stance and public guidelines that make controlled structures costlier to maintain.

The legitimacy lever. At a controlled company, then, the assembly acts through voice and exposure, not through outcome or exit.³⁰ A public, reasoned position reached by a representative sample of the owners is harder to dismiss as an activist stunt than a conventional proposal, and votes of unaffiliated holders now carry legal weight: under the 2025 amendments to DGCL §144, a controlling-stockholder transaction (other than a going-private transaction) receives a statutory safe harbor if it is approved by an independent committee or by a majority of the votes cast by disinterested stockholders, so a fund voting under assembly guidelines can help decide whether that protection is available.³¹ Beyond the single company, repeated and documented override of a deliberated owner majority could build political pressure for sunset requirements. That chain is plausible and unproven, and it runs against the recent direction of Delaware legislation, which in 2025 widened controller protections. A claim of this kind can become unfalsifiable if the success criterion shifts from outcomes to “legitimacy,” so the paper states its indicators in advance: (a) the number of funds adopting assemblies, and how often they follow rather than overrule them; (b) issuer responses to published guidelines, such as changes in governance terms; (c) the terms of controlled IPOs, including sunset provisions, after guidelines are public; (d) legislative or exchange-rule changes that cite assembly findings. If none of these appears after several annual cycles, the legitimacy hypothesis fails.

4. Antitrust. The common-ownership critique holds that shared ownership across competitors dampens competition. It is now live litigation: eleven states led by Texas sued the Big Three over coal output; the court largely refused to dismiss in 2025; Vanguard has settled; BlackRock and State Street remain defendants, and reporting indicates the Justice Department has considered joining.²⁴ The risk is real and the assembly does not dissolve it. It does change the structure the theory depends on. The alleged conspiracy is among managers pursuing an agenda; under an assembly, the instructions come from the fund’s own investors through a published process, and the answer to “who gave Wall Street a mandate to decide” is: the owners, by lot. HLZ also expect the Big Three to have their own reason to adopt: shedding discretionary voting power is a way to answer breakup proposals. Whether a court accepts that distinction is an open question.

5. Political risk. HLZ argue the assembly is politically neutral a priori: liberals expect humans to be prosocial and conservatives expect them to be selfish, so each side expects to win and neither has reason to fight the design as such. The claim is testable; the pilot will be the test.

Summary: what each lever can and cannot do. The table collects the levers discussed above and in I.C. It states what each can do at a widely held company, where no one holder decides the vote, and at a controlled company, where one holder or family does. The assessments are this paper’s own, and several rest on hypotheses flagged in the notes.

Lever	Widely held company	Controlled company
Withholding or voting against directors under published guidelines	The strongest lever. Index-fund stakes are large enough to decide outcomes with other holders; the board-diversity campaign is the precedent.	Cannot change the result. Value is signal and public record.
Precatory shareholder proposals backed by the guidelines	Non-binding but hard to ignore at high support; HLZ’s leverage argument rests on director votes.	The controller can ignore them. The unaffiliated-share tally is the measurable signal.
Bylaw amendments (DGCL §109)	Available to a holder majority, within limits on intruding on board authority.	The controller can block; not a primary lever.28
Charter amendments (DGCL §242)	Board must propose first, so owner action is indirect.	Same, plus the controller’s veto.
Disinterested-stockholder approvals (2025 safe harbor; dual-class extensions)	Funds’ votes can decide conflicted director and officer transactions.	The strongest handle: fund votes can decide whether a controller transaction gets the safe harbor, and whether an extension passes the unaffiliated holders.31
Published advance guidelines	Shape the proposals activists file and the boards anticipate (HLZ).	Let IPO issuers price the consequences in advance. A hypothesis.
Fund responses to index-provider consultations	Little relevance.	Eligibility rules for multi-class firms; they moved both ways between 2017 and 2023. A hypothesis.29
Public, reasoned explanation of findings	Adds legitimacy of process (HLZ).	Exposure of overridden owner majorities. Unproven; indicators in IV.A.3.
Exit by the index fund	Unavailable by construction; HLZ doubt it would change behavior.	Same.
Assembly-governed index (new product)	Can set screens or weights.	Can exclude or underweight controlled structures. Requires a new fund and a different tracking promise; outside the HLZ design.

IV.B — Adoption path

1. Pilot, advisory. One large fund, one assembly, public report, and a reasoned-response rule: if the fund overrules the assembly, it must explain, in detail and in public. This rule is the ratchet. Transparency about divergence converts an advisory body into a binding one by reputation, before any change of law.
2. Private contract, competitive adoption. No legislation is required. A fund can adopt the assembly through its own governance documents. In a competitive industry, a first mover that attracts assets from investors who want a voice pressures the rest.
3. Existing plumbing. Pass-through voting infrastructure already exists: BlackRock had about $2.6 trillion eligible for its Voting Choice program at the end of 2023, and Vanguard’s Investor Choice spans about $4 trillion of eligible assets. A fund also holds contact details for every investor, so surveying and stratifying the population is cheaper than for a government.
4. Complementarity with TSC. The Shareholder Commons supplies the diagnostic map of externalities; the assembly supplies the legitimacy to act on it. TSC’s analysis is natural testimony for an assembly to hear, and an assembly is a natural body to adopt TSC-style guidelines. They are partners, and the assembly addresses the case TSC itself does not resolve: where an investor’s ethics and wealth point in different directions.
5. Regulatory route. HLZ note that the design can also be read as a mandate to answer excessive concentration of voting power. That route is available if private adoption stalls.

IV.C — The strongest objections, steelmanned

Each objection below was pressed in earnest, and each looks decisive until examined. They are answered in their strongest form.

“It is only a proposal; no assembly has directed a Big Three vote.” True, and the authors say so. But an absence of track record is a reason to run the pilot, not a reason to prefer the alternatives, which do have track records, and the records are of failure (I.C). A Dutch pension-fund assembly has already run. The cost of the next step is a fraction of a percent of one fund’s fee revenue.

“It can’t bypass Delaware.” The assembly does not need to. It does not alter directors’ duties; it alters who instructs the owner’s vote and with what guidelines (IV.A.3). McRitchie blocked court-mandated altruism by managers; it does not remove owners’ right to vote.

“Vanguard’s own data show investors abandoning ESG policies.” The figures are accurate: the share choosing the Glass Lewis ESG policy fell from 45% in 2023 to 12% in 2026, while the company-board-aligned policy rose from 17% to 38%.²⁵ Four points reduce what they prove. First, Vanguard attributes the fall mainly to a change in the mix of participants: participation grew more than sixfold to over half a million investors, and within the ESG U.S. Stock ETF, whose investors have been in the program since 2023, ESG-policy selection rose from 73% to 78%. Second, the menu changed: a new wealth-focused option took 22% in 2026. Third, no policy exceeded 38%, and younger investors chose the ESG policy at more than twice the rate of those aged 62 to 80 (38% against 16%); this heterogeneity is exactly why a legitimacy mechanism is needed rather than a click-count. Fourth, and decisively, the data measure self-selected, individual, uninformed choices from a menu the manager wrote. That is the regime HLZ say fails, for reasons of rational apathy. It cannot test the hypothesis that a representative group, informed and deliberating, would decide differently. The hypothesis is unproven. But the data are no evidence against it.

“150 people cannot legitimately bind millions.” Representation by lot is a statistical claim, and the statistics are favorable: a 2% minority is present with 95.2% probability, and a 49% position almost never prevails under a 60% rule. Juries bind defendants with far less statistical cover. The alternative is not “everyone”; the alternative is a few dozen professionals nobody chose.

“Assemblies will be ignorant or captured.” Capture requires identifiable expertise to buy, a career to bargain, and a known individual to reach; members are drawn by chance, can remain anonymous during the process, have nothing marketable to sell, and serve for a year. Ignorance is addressed by experts “on tap.” The fair comparison is to the status quo, in which a small set of proxy advisors and stewardship teams are themselves exposed to lobbying and conflicts of interest. Facilitation, human or AI-assisted, ensures equal speaking time.

“Who sets the agenda? What about Arrow?” The assembly sets its own agenda, which removes the manager’s agenda power that makes pass-through voting hollow. Empirical work on deliberative polls finds that deliberation tends to make preferences more single-peaked, which dissolves much of the cycling problem Arrow-type results predict.²⁶

“The hemisphere language is rhetoric.” The mapping is functional, not neuroanatomical. The claim is that institutions can be arranged so that a mode of attention that grasps and measures either serves or commands a mode that attends to wholes and relations. Hutchinson and Yarrow make this claim in economics on independent grounds, and McGilchrist applies it to organizations. The strongest test is not the label but the prediction: a system in which the measuring layer commands will systematically fail to register what it cannot measure, and the evidence in I.B is of that failure.

“It is socialism, or it is a cartel.” It is ownership exercised, with votes allocated to owners and odds weighted by stake (II.D). The cartel objection is the common-ownership critique and is addressed in IV.A.4; it is a litigation risk to manage, not a conceptual defeat.

IV.D — Extensions

The empty chair. Specification 1 in III.B, applied as a standing rule of procedure.

An assembly-governed index. Where the HLZ design stops at the index-tracking promise, a new fund can define the promise differently: a published benchmark whose eligibility rules, such as governance terms for controlled companies, are set by an assembly. This is not in the HLZ paper; it is the route by which untracked values could enter the tracked object itself, and it would test the paper’s thesis in the market.

Other intermediaries. The Dutch pilot was at a pension fund. The same mechanism applies to any institution that votes other people’s money, and, at the political scale, to bodies that must answer to citizens. HLZ cite a Norwegian citizens’ panel as a further precedent.³²

V. Conclusion: The Hinge

The Investor Assembly is the clearest existing design for placing the machinery of capital allocation beneath a body that can deliberate, face to face, on what the machinery cannot see. Against it stand real legal hurdles, which this paper has named, and an absence of track record, which only a pilot can cure. The alternatives have track records, and the records are of jamming.

But an institution can only make room. It can arrange anonymity, equality, information and time; it cannot supply the disposition that makes use of them. Roger Sperry’s argument, which McGilchrist cites, is that value systems sit upstream of every other lever for the world’s crises.²⁷ No assembly changes that. What an assembly can do is place real people, drawn by chance, around a table where the stranger’s claim can be heard, and that is what hospitality has always meant. Whether it will be heard remains open; the point is that the table exists.

Notes

1. Oliver Hart, Hélène Landemore & Luigi Zingales, “How to Implement Shareholder Democracy,” Harvard Business Law Review 16(2) (2026), pp. 355–86 (PDF); earlier working-paper version (Chicago Booth Stigler Center No. 350, 2024) and summary on the Harvard Law School Forum on Corporate Governance. All HLZ figures and design details in this paper are taken from the published version unless noted. ↑
2. “Focal regard” and “open regard” are this paper’s adaptation of McGilchrist’s contrast between narrow-beam, focal attention, which he assigns to the left hemisphere, and open, sustained, vigilant attention, which he assigns to the right (The Matter with Things, Introduction, “Some Core Ideas of the Hemisphere Hypothesis”; page references to be added). McGilchrist does not use “regard” as a technical term. He grounds the contrast in Goethe’s Verstand and Vernunft (Goethe: “Die Vernunft ist auf das Werdende, der Verstand auf das Gewordene angewiesen”, in Maximen und Reflektionen [1833], Sämtliche Werke, Aufbau-Verlag, vol. 18 (1972), 545–73, at 563, as quoted in McGilchrist’s Chapter 22, “Time” (endnote 66); and the Eckermann conversation of 13 February 1829, Gespräche mit Goethe in den letzten Jahren seines Lebens, part III, in Biedermann (ed.), Goethes Gespräche, vol. 6, Leipzig, 1889–96, cited in his Chapter 28, “The Sense of the Sacred” (endnote 120), where he equates Verstand with the left hemisphere and Vernunft with the right; endnote numbers are those of the epub and should be checked against the print edition). The English rendering is McGilchrist’s; he also glosses the contrast as ratio and intellectus in Latin, and dianoia and nous in Greek and in his parable of the master and his emissary (The Master and His Emissary). In the Cosmoxenia framework the same asymmetry appears as host and guest dispositions, understood as directional dispositions rather than fixed roles, with host-priority ontological and not temporal; the Metanarrative essay and the earlier case studies develop that vocabulary, which this paper does not require. Case Study II (AI Alignment) holds that alterity-centred design cannot weigh at the global level (nothing is worth doing except in the light of the Other) while Levinas requires weighing in mundane interactions; III.C applies that distinction. ↑
3. HLZ, Section I: index funds held 0.5% of S&P 500 capitalization in 1975, 2% in 2005, 11% in 2015, 19% at end-2024; the Big Three held 36% of mutual fund assets in 2025 (12% in 2000); their median combined stake in S&P 500 companies was about 22% in 2025 (7.1% in 2000), and about 29% of votes cast in 2024. The State Street board-diversity campaign is described there; its own reporting counted 681 companies adding female directors by 2020. ↑
4. HLZ, I.A, citing Florida CFO Jimmy Patronis calling it undemocratic for asset managers to use their power to influence societal outcomes, and Fisch & Schwartz, Corporate Democracy and the Intermediary Voting Dilemma, 102 Tex. L. Rev. 1 (2023). ↑
5. HLZ, I.B: voting divergence from Morningstar analysis (Stewart, 2023); credit-analyst partisanship from Kempf & Tsoutsoura (2021); Friedman separation and its failure from Hart & Zingales, “Companies Should Maximize Shareholder Welfare Not Market Value” (2017). ↑
6. The Shareholder Commons, theshareholdercommons.com. Its framing in terms of beta versus alpha, system-level investing and universal owners is standard in its materials; its theory of change is financial self-interest at the portfolio level, which is the sense in which it is described here as utilitarian. ↑
7. Max Scheler, Formalism in Ethics and Non-Formal Ethics of Values (value ranking); Martin Heidegger, “The Question Concerning Technology” (Bestand, standing-reserve); Iain McGilchrist, The Matter with Things and The Master and His Emissary (hemispheric modes of attention; Scheler’s hierarchy). ↑
8. The 2024 Court of Chancery decision in McRitchie v. Zuckerberg dismissed a suit that sought to hold Meta’s directors to duties owed to diversified investors; see the Columbia Blue Sky Blog commentary, “McRitchie v. Zuckerberg: Fiduciary Duties Are Firm-Specific” (May 31, 2024). Exact docket details and date to be confirmed before publication. ↑
9. HLZ, I.D, citing Brav et al., “The Proxy Voting Choice Revolution” (ECGI Working Paper No. 875, 2025). ↑
10. Jeremy Lent, “Democracy Was Never Designed to Work”, drawing on Aristotle, Politics. Also collected in the Hospitality Addenda. ↑
11. Luke Kemp, video; Roman Krznaric, video (asabiyyah is Ibn Khaldun’s term for social cohesion). ↑
12. HLZ, VI: OECD, Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave (2020), counting 733 minipublics in 34 countries as of 2023 in the version HLZ cite; Kalven & Zeisel, The American Jury (1966); cross-national jury and corruption studies (Wong 2007; Voigt 2009). ↑
13. Harold Hutchinson & George Yarrow, “To See or Not to See, That Is the Question: Moving On from a Half-Brained System of Economic Governance” (RPI Europe). ↑
14. Iain McGilchrist, video, from 13:34, on the separation of bureaucrats and those who see the big picture. ↑
15. Daniel Schmachtenberger on “wisdom-binding-power” and Zak Stein on axiological design, collected in the Hospitality Addenda linked in note 10. ↑
16. HLZ, III–V. Details: fund-level organization (III.A); opt-out (III.B); size and sampling-error figures (III.C); lottery weighting, one draw per investor, equal status (III.D); agenda and process (III.E); consultative status with detailed justification for overruling, and the withheld-director-votes leverage argument against the “precatory” objection (III.E, V.E). “On tap, not on top” is from Landemore & Pénigaud, as cited by HLZ in V.A. ↑
17. Cooper, van der Stoep & Bauer, Deelnemersdialoog report (Pensioenfonds Detailhandel, 2024); Cooper & Bauer, Netspar (2024); Bauer et al., “Investor Democracy,” ECGI Working Paper No. 1125 (2026). HLZ, VI.C. ↑
18. HLZ, IV.D, give the budget (about $1.1 million), $1.2 trillion indexed to the S&P 500 at Vanguard, a four-basis-point average expense ratio ($480 million in annual revenue), and a BlackRock proxy team of about seventy people at an assumed $150,000 each. The published text describes the assembly’s cost as “one seven-thousandth of a basis point”; the stated figures give about 0.23% of fee revenue, or roughly one hundredth of a basis point, so that sentence appears to contain an arithmetic slip. This paper uses the recomputed figure. Confirm with the authors before publication. ↑
19. Emmanuel Levinas, Totality and Infinity (1961) on the face and the stranger, widow and orphan; Otherwise than Being (1974) on saying and said, and on the third party and justice. Page references to be added. ↑
20. The authors’ London Business School panel acknowledged this limit, stating their aim as improving shareholder democracy rather than settling broader corporate accountability to society (LSE/LBS Wheeler Institute write-up). That write-up also reports selection probability proportional to shareholding, consistent with the published paper. ↑
21. Cross-references: Case Study VIII-b (the Seventh Generation Principle) and Case Study XI (the planetary scale, where the “absent” includes the nonhuman). ↑
22. HLZ, IV.C: 1940 Act; SEC, Commission Guidance Regarding Proxy Voting Responsibilities of Investment Advisers (2019); ISS Catholic faith-based guidelines; Fifth Third Bancorp v. Dudenhoeffer, 573 U.S. 409 (2014); Schanzenbach & Sitkoff (2020) for the strict reading and Naftali Ben Zion (2024) against it; Fisch & Schwartz (2023). This is not legal advice; counsel review is required. ↑
23. Bebchuk & Kastiel, “The Untenable Case for Perpetual Dual-Class Stock,” 103 Va. L. Rev. 585 (2017); Cremers, Lauterbach & Pajuste, “The Life Cycle of Dual-Class Firm Valuation,” Review of Corporate Finance Studies 13(2): 459–93 (2024). The 2018 version (discussed on the Harvard Forum) described the premium as becoming insignificantly negative six to nine years after listing; the later working-paper versions (2020, 2022; sample 1980–2019) state that dual-class firms on average come to trade at a discount to comparable single-class firms, and the published abstract says the premium tends to dissipate as the wedge between voting and cash-flow rights widens. The introduction of the published version is reported to state that the discount begins about seven to nine years after the IPO, with the data tables showing a significant average discount in roughly years seven and eight; the full text was not accessible when this was drafted, so confirm the wording, and the table figures and what they measure, before quoting. “Steep” is not the authors’ word and is not used here. Whether public advance guidelines are priced at IPO is a hypothesis, not an established finding. ↑
24. Texas v. BlackRock, Inc., No. 24-cv-00437 (E.D. Tex.); motions to dismiss largely denied August 1, 2025; Vanguard settlement and continuing claims against BlackRock and State Street per the Climate Case Chart; reported DOJ consideration of joining per Bloomberg Law. Current status should be re-checked at publication. The common-ownership literature: Azar, Schmalz & Tecu (2018); the breakup proposal is Posner, Scott Morton & Weyl (2017). ↑
25. Vanguard, press release on 2026 Investor Choice data: 2023 shares 17% board-aligned, 45% Glass Lewis ESG, 36% Vanguard fund policy; 2026 shares 38% board-aligned, 22% Egan-Jones wealth-focused, 12% ESG, 28% fund policy, 1% mirror voting. Age split and the ESG-ETF cohort (73% to 78%) and Vanguard’s mix-shift explanation per ESG Today. Participating assets of about $151 billion in 2026 (from $9 billion) per OneStop ESG, against roughly $4 trillion of eligible assets per 401(k) Specialist, which is under 4% by simple division. The 2% opt-in figure in HLZ predates the 2026 expansion. ↑
26. List, Luskin, Fishkin & McLean, “Deliberation, Single-Peakedness, and the Possibility of Meaningful Democracy: Evidence from Deliberative Polls,” Journal of Politics (2013). Citation details to be confirmed. ↑
27. Roger Sperry, “Science and the Problem of Values,” Zygon 9(1) (1974): 7–21, quoted by McGilchrist in the Epilogue of The Matter with Things (his endnote cites pp. 7 and 9). Sperry argues that corrections to symptoms such as pollution and poverty will follow only after the underlying human values change first. McGilchrist presents his own book as an attempt to address that concern. Page references to the print edition to be added. ↑
28. Delaware General Corporation Law §109 (stockholders’ power to adopt and amend bylaws) and §242 (charter amendments require a board resolution before the stockholder vote); CA, Inc. v. AFSCME Employees Pension Plan, 953 A.2d 227 (Del. 2008) (limits on bylaws that mandate board decisions). Stated from general knowledge of Delaware law, not from the HLZ paper; confirm with Delaware counsel before publication. Many large companies are incorporated elsewhere, so the analysis varies by state. ↑
29. S&P Dow Jones Indices barred new multiple-share-class companies from the S&P 500, MidCap 400 and SmallCap 600 effective July 31, 2017 (existing constituents grandfathered), then reversed that rule on April 17, 2023 after a consultation held October to December 2022 (Davis Polk and Wilson Sonsini client updates, April 2023). FTSE Russell adopted a different rule in 2017: a minimum of 5% of voting rights in the hands of unrestricted public shareholders (Gibson Dunn, August 2017); both it and MSCI include multi-class firms, per the Wilson Sonsini update. The history shows that index eligibility is a real lever, that providers consult their clients, and that the outcome has run in both directions. That a fund’s consultation responses, directed by an assembly, could move such rules is a hypothesis, not an established mechanism. ↑
30. Broccardo, Hart & Zingales, “Exit versus Voice,” 130 J. Pol. Econ. 3101 (2022), cited by HLZ in I.C: voice is more effective than exit in pushing companies toward objectives other than profit maximization. Berk & van Binsbergen, “The Impact of Impact Investing,” on the small effect of divestment on a firm’s cost of capital; cited from memory and to be confirmed. “Mathematically toothless” overstates the literature, since the results depend on model assumptions. ↑
31. Delaware Senate Bill 21, signed March 25, 2025, amending DGCL §§144 and 220: safe harbors for controlling-stockholder transactions via an independent committee or a majority of disinterested stockholders’ votes cast (both for going-private transactions); a narrowed definition of controlling stockholder (at least 33.3% of voting power plus managerial authority for a non-majority holder); and exculpation of controllers for care breaches. See Harvard Forum summary (April 1, 2025) and Gibson Dunn, Morgan Lewis and Willkie client updates. Commentators describe the bill as favoring corporations, controllers and management, and it was passed after reincorporation threats. How a fund’s vote is counted as a “disinterested” vote in a given transaction needs counsel review. ↑
32. HLZ, VI.C (the Framtidspanelet, 56 citizens). The panel’s subject and mandate are to be confirmed against the HLZ text and the Norwegian source before publication. ↑