AMERICANDIVIDENDFUND EST·MMXXVI American Dividend Fund Est. July 4, 2026 · A nonpartisan policy laboratory

Proposal № 060 of 250  ·  Released September 1, 2026

The Detention Clock

Immigration detention is civil. Nobody in it is serving a sentence, and 3 in 4 of the 65,000 people held in November 2025 had never been convicted of anything. A bed costs $165 a day; supervised release costs about $5 and produces court appearance rates above 90 percent. The Supreme Court said in 2001 that six months is the outer edge of reasonable and in 2018 that the statute sets no clock at all. Write the clock: a hearing in 7 days, a judge's finding every 30, and release at 180.

The PerimeterShare on X

The problem

There are two kinds of custody in America. One follows a conviction, and it has a sentence, a judge, and an end date. The other is called civil, and immigration detention is its largest form. In mid-November 2025 ICE held 65,135 people, and 73.6 percent of them, about 48,000, had never been convicted of a crime. They were held not as punishment, because the law says it is not punishment, but to make sure they turned up for a hearing that, in the immigration courts, is on average about 900 days away.

The courts themselves are the reason for the wait. The Executive Office for Immigration Review counted about 570 judges in January 2026, carrying more than 2 million pending cases, about 3,300 per judge; at the 2024 peak, the Syracuse University clearinghouse that tracks the docket counted 3.7 million. A person detained while waiting for that queue is being held to guarantee attendance at a proceeding the government itself cannot schedule.

The Supreme Court has been asked twice what the limit is. In Zadvydas v. Davis in 2001 it held that a person with a final removal order who cannot actually be removed may be held for six months, presumptively, and after that must be released unless removal is likely soon. In Jennings v. Rodriguez in 2018 it held that the detention statute itself contains no requirement of a bond hearing, at six months or ever, for people whose cases are still pending. The result is that the clock exists for the people whose cases are over and does not exist for the people whose cases have not begun.

The cost is not hidden. The fiscal 2025 bed rate is $164.65 per person per day. The average stay was 44 days in September 2025 and rising, from 42 days in fiscal 2022 toward an estimated 52 in fiscal 2026. The government's own alternative, supervised release with check-ins and monitoring, costs between $4.20 and $8 a day, and the programmes that use it report court appearance rates of 93 to 97 percent. The country is paying thirty times the price of the thing that works, to hold people who have not been convicted, for a hearing it has not scheduled.

№ 007 argued for a perimeter that is firm about the line and humane in the method. № 055 put a judge between the state and a front door. This proposal puts a clock on the cell.

The proposal

*A statutory Detention Clock for civil immigration custody. A hearing before an immigration judge within 7 days of detention. Continued detention only on a judge's written finding of flight risk or danger, renewed every 30 days. Release to supervision at 180 days if the case is not decided, with the clock running against the government, not the detainee. And the six-month limit of Zadvydas written into the statute for anyone with a final order who cannot be removed.*

How it would work

  1. Seven days to a judge. Every person taken into civil immigration custody sees an immigration judge within 7 days for a custody hearing. The government carries the burden of showing, on the record, that the person is a flight risk or a danger. This is the hearing Jennings said the statute does not require. The statute will now require it.
  1. Thirty days between findings. Detention continues only while a judge's finding stands, and the finding expires after 30 days unless renewed on the record. A person cannot be forgotten in a cell because nobody had to look again.
  1. The 180-day release. If the case has not been decided in 180 days, the person is released to supervised alternatives: check-ins, monitoring, and a reporting address, the programmes that already exist and already produce appearance rates above 90 percent. The case is not dismissed. The government does not lose its claim. It loses only the bed, and it may have the person back in court on the same docket. The clock runs against the state because the state controls the docket; a detainee cannot be made to pay, in days, for the government's failure to hire judges.
  1. The exception, narrowly. A person with a conviction for a violent felony, or a finding of danger by a judge, may be held past 180 days, but the 30-day renewal still applies and the finding must be made each time. This is the same population № 101, The Removal Rule, will address with certainty of removal; until then, they are the exception that keeps the rule honest.
  1. **Zadvydas in statute.** For anyone with a final order of removal who cannot be removed, because no country will take them, the six-month limit becomes law rather than a presumption courts may read differently, and release is to supervision with a duty to report.
  1. The published clock. Every facility publishes its median and its longest current detention, monthly. A facility with someone at day 400 has to say so, and a judge has to have signed for it thirteen times.

The numbers

What detention costs now. 65,000 people at $164.65 a day is about $10.7 million a day, or roughly $3.9 billion a year, before the cost of transport, medical care and litigation.

What the alternative costs. The same 65,000 on supervised release at $8 a day is about $190 million a year. If the clock moved half the detained population to supervision, the saving is on the order of $1.8 billion a year. This proposal does not claim all of it, because some of the people released will need to be found again, and finding people costs money. It claims most of it.

What the courts need. A 7-day custody hearing for everyone detained is roughly 300,000 hearings a year at recent arrest rates, about 500 per judge, on top of a docket already at 3,300. The clock will not work without judges; № 099 proposes the court that can hire them. Until then the clock forces the question in the only way that has ever worked in this country: by making delay expensive to the party that can end it.

The honest objections

"Release them and they disappear." Some will. The programmes report 93 to 97 percent appearance, which means 3 to 7 percent do not, and with a population of tens of thousands that is real people who will have to be found. Against that: the 48,000 people held today with no conviction are, by definition, not the dangerous ones, and № 022 gives the country the exit record it has never had. The choice is not between certainty and risk. It is between a system that spends $4 billion a year to hold people who mostly would have appeared anyway, and one that spends a tenth of that and loses a few percent.

"A clock rewards the immigrant who drags the case out." The opposite. The clock runs against the government, and the detainee gains nothing from delay because release at 180 days is to supervision with a duty to appear, and the case continues. The party that can shorten the wait is the one that schedules hearings and hires judges, and this is the first proposal that gives it a reason to.

"Mandatory detention was written by Congress on purpose, for criminals." It was, in 1996, and item 4 keeps it for the violent. What the statute did not foresee was a docket of two million and a detained population three-quarters of which has no conviction at all. A rule written for the dangerous is now applied mostly to the merely present. The clock restores the distinction Congress meant.

"This is soft on the border." It is not about the border. A person detained at the line and removed in a week never reaches the clock. The clock bites on the interior, on the person held for months waiting for a court date, and it costs the taxpayer nearly $4 billion a year to keep that person in a bed rather than on a monitor. Firmness is the removal order; the cell is just the waiting room, and the country is paying hotel prices for it.

"Judges will rubber-stamp the 30-day renewals." Some will, and the published clock in item 6 is the check: a facility where every renewal is granted and the median stay never moves will be visible, and so will the judge. It is the same answer this catalog gives everywhere. A rule with a public count is not a guarantee; it is a floor, and today there is none.

Sources

  • TRAC Immigration, Syracuse University, Immigration Detention Quick Facts, detained population and criminal history as of November 16, 2025; immigration court backlog counts, tracreports.org
  • U.S. Department of Homeland Security, ICE Detention Statistics, fiscal 2025 bed rate and average length of stay; Congressional Budget Justification FY2025 and FY2026
  • Executive Office for Immigration Review, Adjudication Statistics, immigration judge counts and pending cases, January 2026, justice.gov/eoir
  • ICE, Alternatives to Detention programme cost per participant; National Immigrant Justice Center, The Real Alternatives to Detention (2025); Congressional Research Service on ATD appearance rates
  • Zadvydas v. Davis, 533 U.S. 678 (2001); Jennings v. Rodriguez, 583 U.S. 281 (2018)
  • 8 U.S.C. § 1226 (detention pending decision) and § 1231 (detention after final order)
  • Proposals № 007 (The Humane Perimeter); № 022 (The Turnstile); № 055 (The Warrant Standard); № 099 (The Immigration Court, forthcoming); № 101 (The Removal Rule, forthcoming)