Faster Than the Court

Read it looking for the prohibition. You will not find one. What you find instead is a list of removals. Drive-thru voting: outlawed. Harris County ran ten drive-thru sites in 2020. Twenty-four-hour voting: gone, replaced by an early-voting window of 6 a.m. to 10 p.m. Between them, those two methods carried nearly 140,000 Harris County ballots in the 2020 general election. Tents, garages, mobile units, any temporary structure: no longer permissible as a polling place. Election officials mailing unsolicited absentee applications: a state jail felony — aimed squarely at Harris County's attempt to send applications to 2.4 million registered voters, though political parties may still send whatever they like. Mail ballots now require a driver's license number or the last four digits of a Social Security number, on both the application and the return envelope, and the number must match whichever one happens to be sitting in the voter's registration record. A person assisting a voter must complete paperwork and swear an oath under penalty of perjury, and is confined to reading or marking the ballot — not answering questions about it. Poll watchers get free movement inside the polling place, and obstructing them becomes a crime.

Every one of those is facially neutral. Every one of them applies identically to a retired oil executive in Highland Park and a shift nurse in Aldine.

That much is a familiar story and a well-told one. Facially neutral procedure is a hundred-and-thirty-six-year-old American technology, and the people who write about voting law have covered it thoroughly. It is how a bill like this gets written.

It is not why it works. Why it works is a clock.

Facially Neutral Is the Old Half of the Story

Mississippi's constitutional convention of 1890 did not disenfranchise anyone either. It imposed a poll tax, a literacy test, a residency requirement, and an "understanding clause" under which a registrar could ask an applicant to interpret a passage of the state constitution. Nothing in the text mentioned race. Louisiana followed with its own version in 1898. In 1896, Louisiana had 130,344 registered Black voters. By 1904 it had 1,342.

In Williams v. Mississippi, decided in 1898, the Supreme Court reviewed the Mississippi scheme and let it stand, on the ground that the provisions did not on their face discriminate against anyone. That holding is the founding document of the method. It established that the operative question in American election law is what the statute says, not what the statute does — and once that is the question, the design problem is solved. You do not write a prohibition. You write a procedure, and you tune the procedure until the people you want gone are the people who cannot complete it.

Congress figured this out and, in 1965, built the only mechanism that has ever actually worked against it. Section 5 of the Voting Rights Act did not outlaw specific practices, because the drafters had watched a century of specific practices get outlawed one at a time and replaced faster than they could be litigated. Instead it froze the process: covered jurisdictions could not change any voting procedure at all without federal preclearance.

The congressional finding underneath that design is the thing worth taking out of the statute and keeping, because it is not about race, or Mississippi, or 1965. It is a claim about speed. A legislature can pass a new procedure in ninety days. A lawsuit takes four years. Everything else follows from that ratio. It means case-by-case litigation cannot win against legislative iteration no matter how good the case law gets, because the loser of any given round simply drafts again while the winner is still waiting on appeal. Whoever controls the clock wins regardless of who wins the argument.

Preclearance is the only remedy anyone has ever built that answers the ratio directly. It does not out-argue the legislature. It stops the legislature from moving until someone catches up.

Texas came under coverage in 1975 and stayed there for thirty-eight years.

Sixty-Eight Days, and What They Measure

Now lay out the 2021 calendar.

May 30, 2021: Texas House Democrats walk out before the midnight deadline, breaking quorum, and Senate Bill 7 — SB 1's predecessor — dies. Abbott calls a special session.

Note the date, because it disciplines everything that follows. SB 7 already contained the drive-thru ban, the twenty-four-hour voting ban, the poll-watcher provisions, the assistance oath. The bill's shape was in hand at the end of May.

July 1, 2021: the Supreme Court decides Brnovich v. Democratic National Committee. It is not a preclearance case; Section 5 has been dead since 2013. It is a Section 2 case, and it supplies a set of "guideposts" for judging whether a voting rule results in denial of the vote on account of race. Among them: the size of the burden imposed, weighed against "the usual burdens of voting"; the size of any racial disparity in a system considered as a whole; the state's asserted interests, including fraud prevention, which need not be supported by evidence of actual fraud. And one more, the sharpest of the set — the degree to which the challenged rule departs from standard practice as it existed in 1982, when Section 2 was amended.

July 8: the first special session convenes. Seven days after Brnovich.

August 31: SB 1 passes. September 7: signed.

The tempting reading is that Brnovich designed the bill — sixty-eight days from decision to signature, and look how neatly the provisions fit the new test. That reading is wrong, and the thing that kills it is sitting in the first line of the calendar. Texas legislators had the design in May. They did not need a July opinion to teach them to write neutral subtractions; they had been drafting into the post-Shelby environment since 2013, and the underlying technique is older than the Voting Rights Act.

What Brnovich did was not design. It was ratification, and it arrived seven days before a special session convened to pass the bill it ratified.

Look at the 1982 guidepost against the list of removals. Drive-thru voting did not exist in 1982. Twenty-four-hour voting did not exist in 1982. Counties mailing unsolicited absentee applications to millions of voters did not exist in 1982. Every method SB 1 eliminated was an innovation, most of them very recent — and under a test that measures departure from a 1982 baseline, abolishing an innovation is not a departure. It is a return. The novelty of the thing is precisely what makes it safe to kill.

Nobody in Austin had to be told this. That is the point, and it is a stranger point than instruction would be: the legislature and the doctrine converged on the same shape from opposite ends, independently, and the Court's contribution was to confirm mid-draft that the shape was safe. Inside that constraint, the drafting was careful work — SB 1 required counties of 55,000 or more to offer twelve hours of weekday early voting, which expanded access in smaller, more rural, more Republican counties, and it added an hour to Sunday voting. A bill that simply hated voting would not do that. Doctrine set the outer boundary; partisans optimized inside it. The two are not in tension. They are the division of labor.

So the sixty-eight days are not a design window. They are an interval — the gap between a court confirming a method is safe and a state shipping it. That is a tempo measurement, not a causation claim, and tempo is the whole subject.

The Clock Runs Both Ways

There is one more date on the calendar and it belongs to the other side.

July 12: more than fifty House Democrats fly to Washington and break quorum again, this time to lobby Congress for federal voting legislation. They stay away for over a month. They return in August. The federal bills they went to Washington for — the Freedom to Vote Act, the John Lewis Voting Rights Advancement Act — fail to clear a Senate filibuster in January 2022.

The quorum break bought thirty-eight days and spent them on a remedy that never arrived.

That sentence is usually filed as a footnote about Democratic tactics. It is better read as the pattern confirming itself from the opposite end. The opposition's only real asset in the summer of 2021 was the same asset everyone else was playing — delay. A quorum break does not defeat a bill. It moves a bill in time. Thirty-eight days is a legitimate acquisition; the question is only what you buy with it, and they bought a federal statute that required sixty Senate votes or a carve-out that nobody had. They spent a tempo asset on an argument.

Both sides were playing the clock. One side was buying an interval to ship a statute already drafted. The other was buying an interval to persuade a body that was not going to be persuaded. That asymmetry is not about who was right. It is about which side understood what the asset was for.

The Cost of Being First

The first election under the new rules was the primary of March 1, 2022.

Statewide, roughly 24,600 mail ballots were rejected — about 12 percent. The historical rate in Texas ran around 1 to 2 percent. In Harris County, of 36,878 mail ballots received, 6,888 were rejected, and the county's elections administrator attributed them directly to SB 1: the identification number a voter wrote on the envelope had to match a number already in the registration record, and voters had no way of knowing which number that was. Nineteen percent. Votebeat's analysis later found the rejections fell disproportionately on Black, Latino, and Asian voters.

By the November 2022 general election the statewide rejection rate had fallen to 2.7 percent. Harris County came in around 4.5 percent. Counties ran education campaigns. The online cure process got used. The friction decayed.

That decay is the strongest argument against reading SB 1 as durable disenfranchisement, and it deserves to be stated at full strength rather than waved at. The mechanism is not a wall. It is a tax, and the population learns to pay it, and within two cycles the measurable disparity approaches the noise floor. Anyone claiming SB 1 permanently removed a bloc of voters from the Texas electorate is claiming something the numbers do not support.

But run the decay curve against the clock and it stops being a concession.

A decay curve describes the system adapting. It does not describe the 24,600 ballots from March 2022, which were not counted then and will not be counted later. Whether those ballots would have changed a March primary result is unknowable and probably not; that was never the test. A mechanism this size is not built to flip an election. It is built to sit inside every election and collect fully from whoever arrives first — people encountering a new procedure for the first time, who skew toward first-time and infrequent voters, which is to say toward the voters whose participation was the marginal question to begin with.

And two cycles is fast only by the wrong measure. Two cycles is roughly three years. The lawsuit had not finished.

Faster Than the Court

Shelby County v. Holder came down June 25, 2013, striking the coverage formula and rendering Section 5 inoperative. Texas announced that its voter ID law — which had been blocked under preclearance — would take effect immediately. Not after remand. Within hours. That was not overreach; it was an accurate reading of what had just changed. The only thing that had been stopping the law was a mechanism that no longer existed. What Shelby removed was not a legal argument. It was the one device that had ever compensated for the ratio.

So SB 1 was litigated the old way, which is the way Congress determined in 1965 does not work.

The Justice Department sued in November 2021. On October 11, 2024 — three years and one month later — Judge Xavier Rodriguez struck down SB 1's voter-assistance provisions as violations of Section 208 of the Voting Rights Act. In March 2025 he found that parts of the law obstructing assistance to voters with disabilities violated the Americans with Disabilities Act and the Rehabilitation Act. In August 2025 the Fifth Circuit reversed the assistance ruling, holding that the plaintiffs lacked standing to challenge the disclosure requirement and the expanded oath, and that the ban on compensated assistance was not preempted by Section 208.

Four years. Two elections held under the law before the first ruling. Another held before the reversal. The plaintiffs won a round in year three and lost it in year four, and every ballot cast in the interval was cast under the statute as written.

That is not a failure of the litigation. It is the litigation performing exactly as the 1965 Congress predicted it would perform, which is why they built something else, and why the something else was taken apart in 2013.

Now set the two curves side by side, because they are the finding. The friction imposed by SB 1 decayed to background inside three years. The lawsuit against SB 1 took four, and lost. The mechanism did not need to be permanent. It needed only to outlive its own review, and it did — comfortably, by a full cycle, without anyone having to defend it on the merits for the elections that actually happened under it.

The pattern states cleanly as a sequence. A facially neutral procedure is enacted. Courts, asked what the text says rather than what it does, uphold it. Congress eventually builds a pre-review mechanism because it recognizes that post-review is too slow. The mechanism is dismantled. Post-review is narrowed. A legislature that already had the draft ships it in the interval after a court confirms the draft is safe. The friction collects its full toll in the first cycle, decays into the background over the next two, and outlasts the lawsuit either way.

Nobody was prohibited. Nobody has to be. The prohibition was always the least efficient version of the technology — loud, litigable, easy to name, and above all slow, because a prohibition invites a lawsuit that might finish before it does. What replaced it does not announce itself, cannot be photographed, and has never once required anyone in the building to say the quiet part. It only requires that the ballot be slightly harder to reach for some people than for others, and that the difference persist longer than the count.