Justice Kagan opens her Callais dissent with a hypothetical. She knows it’s stylized, and she says so. Imagine a state shaped like a rectangle, with one of its six congressional districts a near-perfect circle in the middle. The circle is ninety percent Black. The other five districts, surrounding it, are ninety percent white. Voting is racially polarized. Black voters, concentrated in the circle, elect a representative of their choice. Then the legislature redraws the map — slices the circle into six pie pieces, one for each new district. The Black population is now dispersed. They can still vote. Their votes still count. But across every election, every cycle, those votes will lose.
This is, Kagan tells us, racial vote dilution in its most classic form. And as of last Wednesday, the Court has held that — under the right circumstances — Section 2 of the Voting Rights Act has nothing to say about it.
Almost every account of Callais you’ll read this week will frame the holding as a weakening of the VRA, or a narrowing of Gingles, or an extension of Rucho into territory that Allen v. Milligan had seemed to foreclose. All of those framings are accurate. None of them gets at what the Court actually did.
The Court did not weaken Section 2. The Court did not narrow Gingles. The Court broke a link.
I want to explain what I mean by that, because the distinction is not rhetorical. It is structural, and it has consequences that the standard “weakening” frame will miss.
In 2008 I published a paper called “Arguments-Based Collective Choice.”1 The setup is simple. A group has to decide which actions to take, but actions need justification. A justification is a path: it starts from a first principle, runs through a sequence of intermediate reasons, and arrives at the action. Call such a path an argument. Each link in the argument — each connection between principle and reason, reason and reason, reason and action — has a cost of breaking. Some links are cheap to challenge: nobody much believes them, nobody will defend them. Some links are expensive to challenge: many people accept them, breaking them would require a major rhetorical or institutional effort. An action is sustainable if the argument that supports it is, in the relevant sense, immune to challenge — if the cheapest link in the chain still costs more to break than any individual would gain from overturning the action.
The interesting thing — the thing I spent thirty pages working through — is what happens when arguments share links. If a single high-cost link supports many actions, the calculus changes. Breaking the link doesn’t just overturn one action; it overturns every action whose argument routes through that link. The arithmetic of cost-versus-benefit runs differently when a link is load-bearing for an entire body of doctrine. My standing example in the paper is the Privileges and Immunities Clause: a link that supports many actions simultaneously, which is precisely why no single losing party can afford to break it. The cost of breaking such a link is not really the rhetorical or institutional effort it takes to write the opinion. It is the cumulative consequence of overturning all the actions the link was holding in place.
The framework predicts something specific. A determined actor — a court, in particular — can rationally undertake to break a high-cost shared link if and only if the cumulative gain from overturning all the actions the link supports exceeds the cost of breaking it. This is not a critique. It is a description of how a body of doctrine can be unmade.
Now consider the active argument that supported Louisiana’s District 6 in 2024. Schematically: Equal Protection of the Laws, as a first principle, supports the further proposition that Section 2 of the VRA can require race-conscious districting in cases of vote dilution; that proposition, applied to the facts the Robinson court found, supports the conclusion that Louisiana was required to draw a second majority-Black district; therefore the State may — indeed, must — draw such a district. That is the argument. It runs from principle through reasons to action.
This argument was stable under existing law. The crucial link — the one connecting Section 2 to “race-conscious districting can be required” — had been carrying serious weight for decades. Justice Alito himself, writing for the Callais majority, notes that “for over 30 years, we have assumed for the sake of argument” that compliance with Section 2 can constitute a compelling interest justifying race-based districting. That is the language of a high-cost link. The Court, by its own account, has been treating it as expensive to break, partly because so many other actions depend on it.
What did the Callais Court actually do? Read the holding carefully. The Court did not say that Louisiana cannot have a second majority-Black district. The Court did not say that race-conscious districting is forbidden. The Court did not even say that Section 2 fails to provide a compelling interest. What the Court said, almost in so many words, is that Section 2, properly construed, did not actually require what the Robinson court thought it required. The link between “Section 2” and “race-conscious districting required here” — the link that justified Louisiana’s race-based response — turns out, on the new reading, not to have been there.
This is what link-breaking looks like in practice. The Court did not strike down an action; it removed the connector that the action depended on. Whether the action survives is now a separate question, contingent on whether some other argument can be constructed to support it.
The reason the distinction is not merely semantic — the reason “the Court broke a link” says something more general than “the Court raised the bar for §2 plaintiffs” — is that the broken link was not unique to Louisiana. It was the same link supporting Allen v. Milligan (Alabama, 2023), Cooper v. Harris (North Carolina, 2017), Alabama Legislative Black Caucus v. Alabama (2015), Wisconsin Legislature v. Wisconsin Elections Commission (2022), and behind those, the entire Gingles edifice constructed since 1986. Every one of those decisions ran an argument that routed through the link the Callais Court has now broken. The Court did not have to overrule those cases individually. It did not even have to mention most of them. By breaking the shared link, it overturned the structural support for all of them simultaneously, and for every prospective claim that would have routed through the same link.
This is what Justice Kagan means when she writes, in the final paragraph of her dissent, that today’s decision “renders Section 2 all but a dead letter.” It is not rhetorical excess. It is a structural observation. Once the load-bearing link is gone, the actions it was supporting do not stand on their own. They were never standing on their own. They were standing on the link.
A point worth flagging, because it changes how to read the case. Justice Thomas, joined by Justice Gorsuch, would go further. His concurrence argues that Section 2 does not regulate districting at all — that the “voting qualification, prerequisite to voting, or standard, practice, or procedure” language reaches only access-to-the-ballot questions, never how district lines are drawn. That is an argument for breaking a link further upstream — a link so foundational that breaking it would overturn essentially every Section 2 districting case in American constitutional history. Thomas’s position is a more aggressive use of the same operator the majority is using. He would have broken a link the majority left intact. He is two votes shy.
The presence of his concurrence on the docket is itself a piece of analytical information. It tells us that the structure of the Callais majority is a compromise position within a broader project, not its outer limit.
One more piece of the puzzle, briefly, before I close. Once the link from “Section 2” to “race-conscious districting required” has been broken, the same physical map — a Louisiana congressional plan with two majority-Black districts — can in principle still be supported by a different argument, routed through a different principle. The Court is explicit, in fact, that partisan motive is a constitutionally permissible justification under Rucho v. Common Cause (2019). A legislature that announces a partisan rationale for a particular districting choice is, under current doctrine, almost always shielded from racial-gerrymandering attack. Whether a given map is constitutional, in other words, no longer depends primarily on what the map does. It depends on what the map is called, and on which principle the labeler is permitted to invoke. This is a constitutive-classification problem of the kind Maggie Penn and I have been working on in a different context: the equilibrium response of a regulated population to a classification rule.2 It is also, I think, where most of the real action is going to be in post-Callais litigation. But the labeling regime, and its asymmetries, deserve their own treatment, and that is not the point of this post.
The point of this post is simpler, and I want to state it plainly. The standard account of Callais is that the Court weakened Section 2. That account is true, but it understates the operation. The Court did not weaken a statute. The Court broke a structural link in a body of argument that supported many decisions, and the link’s removal cascades through all of those decisions at once. This is why Justice Kagan reaches for “all but a dead letter” rather than for some more measured formulation. The structure she is describing — the simultaneous unraveling of an entire line of doctrine through the breaking of a single shared link — is the structure my 2008 paper was about, and it has now arrived in a constitutional opinion of major consequence.
There is a second analytical question, just as important, that this post does not address: why did it take thirteen years? From Shelby County (2013) to Brnovich (2021) to Callais (2026), the Roberts Court has been moving through the Voting Rights Act in a sequence that is neither random nor arbitrary. The question is why this case, why this term, why now. That question requires a different formal apparatus — one I developed with Deborah Beim and Tom Clark in a paper called “Why Do Courts Delay?” — and I will take it up in the next post.
For now, what I want to leave you with is this: when you read commentary on Callais over the next several days, watch for the framing. If the analyst tells you the Court weakened Section 2, ask what is meant by “weakened.” If the answer is that the Court raised the bar for vote-dilution plaintiffs, the analyst is describing a partial truth. The Court did something more general than that, and more consequential, and the apparatus for seeing it clearly has been around for almost two decades. We do not need new theory to understand what just happened. We need to apply theory we already had.
With that, I leave you with this.
Notes
1 John W. Patty, “Arguments-Based Collective Choice,” Journal of Theoretical Politics 20(4): 379–414 (2008). The framework distinguishes principles, reasons, and actions, with arguments connecting them through links of varying cost; the central results concern when sets of arguments are stable against challenge by individuals with stakes in the actions supported.
2 See Elizabeth Maggie Penn and John W. Patty, “Algorithmic Fairness with Feedback,” arXiv:2312.03155, and the broader project on classification and social outcomes that will appear as a Russell Sage Foundation book. The connecting idea is that classification rules induce equilibrium responses in the populations they classify; the rule and the population are jointly determined.