Provenance · The Debate
The debate behind The $70 Billion ICE Bill Isn't About Immigration. It's About Building a Machine That Can't Be Turned Off.
The question$70 Billion for ICE Is Not an Immigration Policy — It's an Infrastructure Bet on Permanent Enforcement
How this debate works
Before writing, The Arbiter stress-tests each story by framing the two strongest opposing positions and arguing both sides of a structured three-round debate: opening arguments, rebuttals, then steel-manning the opponent and answering one question — what specific, verifiable evidence would change my mind?
Both advocacy roles were generated by the same provider/model: Advocate A · Anthropic Claude Opus 4.6 argued the first position, and Advocate B · Anthropic Claude Opus 4.6 argued the second. The published article was written separately by Anthropic Claude Opus 4.6: the debate supplies the questions, and the author verifies key claims with its own research before taking a position.
Sources in this transcript are evidence as each advocate presented it during the debate — research leads, not independently verified endorsements. This is an early-format debate: full arguments were recorded, but per-claim evidence lists were not yet captured.
Why we covered this
Framing the ICE budget as infrastructure lock-in rather than policy is a genuinely important reframe with long political shelf life. The argument that this outlasts administrations and creates structural irreversibility is analytically rich and underrepresented. High importance to a wide professional audience — legal, policy, finance (government contractors). One of the stronger premises in the batch.
The positions
Advocate A · Anthropic Claude Opus 4.6 argued
Massively scaling ICE infrastructure is a deliberate and legitimate strategy to create durable immigration enforcement: just as Democrats use regulatory agencies and entitlement programs to institutionalize their policy goals, Republicans are right to build enforcement capacity that survives electoral cycles and prevents future administrations from simply defunding border security.
Advocate B · Anthropic Claude Opus 4.6 argued
Deliberately engineering a $70 billion enforcement bureaucracy to be politically irreversible subverts democratic accountability — when one party uses infrastructure spending to lock in a policy agenda that a future majority cannot undo, it isn't governance, it's institutional capture of the state.
Round 1 — Opening arguments
Each advocate states its position and strongest evidence.
Read the full argument
Opening Statement: Advocate A
Position
Massively scaling ICE enforcement infrastructure is a legitimate, strategically rational exercise of governing power. The $70 billion funding push is not a policy distortion — it is an entirely conventional application of how durable governance works in a separation-of-powers system. Republicans are doing precisely what any serious governing coalition does: converting electoral mandates into institutional capacity that outlasts the next election cycle.
Argument 1: Institutionalization Is Standard Governing Practice — Not a Pathology
The premise frames "structural durability" as something suspicious. It shouldn't be. Every major policy agenda in American history has been operationalized through institutional entrenchment.
The New Deal created the Social Security Administration in 1935. Medicare and Medicaid, established in 1965, now consume roughly $1.5 trillion annually — a baseline that no subsequent Republican administration has dismantled despite repeated attempts. The EPA, created by executive order in 1970, survived four decades of Republican presidencies largely intact.
The mechanism is identical in each case: a governing coalition builds administrative infrastructure, staffing, and statutory authority deep enough that reversing it requires more political capital than opponents can realistically marshal. Democrats have successfully executed this strategy repeatedly. The intellectually honest question is not whether this is legitimate governance — it clearly is — but whether enforcement infrastructure deserves the same treatment as regulatory or entitlement infrastructure.
Argument 2: The Enforcement Gap Is Real and Documented
The case for durable infrastructure rests on a concrete problem: enforcement capacity has been chronically inconsistent across administrations, creating genuine governance failures.
Specific evidence:
-
DACA (2012): President Obama created a major immigration status program via executive memorandum, affecting approximately 700,000 people, without Congressional authorization. This was subsequently partially reversed, then reinstated through litigation — demonstrating that non-durable policy creates legal limbo for everyone, including the people it purports to protect.
-
ICE deportation data: Under Obama (FY2012), ICE recorded approximately 409,000 deportations — the highest single-year figure on record. Under Biden (FY2022), the figure fell to approximately 72,000. This is a five-fold swing in enforcement intensity across administrations operating under the same statutory authority (the Immigration and Nationality Act).
-
Detention capacity: As of 2023, ICE's average daily detention population fluctuated between roughly 22,000 (Biden-era) and a target of 54,000+ under Trump-era priorities. Infrastructure that doesn't exist cannot be used — but infrastructure that exists can be scaled back by a future administration far more easily than it can be built from scratch.
The enforcement gap isn't theoretical. It represents a documented, recurring failure of governance continuity that durable infrastructure is designed to solve.
Argument 3: The Asymmetry Argument Cuts the Other Way
Critics of this funding push implicitly apply an asymmetric standard: building regulatory capacity is "policy," but building enforcement capacity is "infrastructure overreach."
Consider the parallel: The Inflation Reduction Act (2022) allocated approximately $369 billion in climate and energy spending, much of it structured through long-term tax credits, agency grants, and loan programs specifically engineered to be difficult to unwind — a strategy openly discussed by Democratic strategists at the time. The Consumer Financial Protection Bureau was structurally insulated from congressional appropriations by design, specifically to prevent a future Republican Congress from defunding it (a feature that has since been litigated to the Supreme Court in CFPB v. Community Financial Services Association).
If institutionalizing progressive policy goals through durable agency design is legitimate governance, then institutionalizing enforcement capacity through durable funding and infrastructure is equally legitimate.
What's at stake
The practical stakes are clear: without durable enforcement infrastructure, U.S. immigration law exists as statute on paper but operates as an optional guideline in practice — enforced aggressively by one party, effectively suspended by another, with no stable legal baseline for anyone. That is not rule of law. That is governance by oscillation.
Building enforcement capacity that survives electoral cycles is a precondition for immigration law meaning anything at all.
Read the full argument
Round 1 Response: Advocate B
Identifying Advocate A's Weakest Argument
Advocate A's asymmetry argument — "Democrats did it too, therefore this is equally legitimate" — is the weakest link in their case, and it fails on its own terms.
The analogy to the IRA and CFPB is structurally disanalogous in a critical respect. The CFPB's funding insulation was challenged precisely because critics argued it subverted democratic accountability — and the Supreme Court took that challenge seriously enough to grant certiorari. The IRA's tax credits are reversible through ordinary legislation; congressional Republicans have already partially done so via the 2025 reconciliation process. These examples do not establish that entrenchment is legitimate. They establish that entrenchment is contested — which is exactly the position I'm defending.
More importantly, Advocate A's "they did it too" framing is a normative non-sequitur. If Democratic institutional entrenchment was also a democratic accountability problem, then Republican enforcement entrenchment is also a democratic accountability problem. The correct conclusion is that both deserve scrutiny — not that one launders the other.
My Evidence
1. The scale here is categorically different from routine institutionalization.
Advocate A invokes the SSA and Medicare as precedents, but those programs were enacted through full bicameral legislation with explicit public deliberation over their permanence. The $70 billion ICE funding mechanism under current proposals is structured primarily through emergency supplemental appropriations and contract infrastructure — specifically, long-term detention facility contracts with private operators like GEO Group and CoreCivic that run 15-20 years and include government-backed occupancy guarantees.
This matters because: Congress can repeal a statute. Congress cannot easily unilaterally void existing contractual obligations to private vendors without triggering breach-of-contract liability that itself becomes a structural barrier to policy reversal. This is not the same mechanism as Social Security. It is a specific, documented technique for insulating policy from future democratic revision through private contract rather than public law.
GEO Group's 2024 annual report explicitly identifies long-term government contracts with guaranteed minimums as its core business model — a model that depends on governments being structurally unable to exit.
2. Enforcement infrastructure of this scale has no historical precedent that survived democratic scrutiny when built this fast.
Advocate A cites the five-fold swing in deportation figures as evidence of problematic inconsistency. But that swing occurred under existing statutory authority — meaning Congress already decided what enforcement power ICE has. What changed across administrations was prioritization, which is a core executive function under Article II. Locking in infrastructure that eliminates that discretion doesn't fix a governance problem; it removes a democratic feature.
The relevant historical precedent for rapid, large-scale enforcement infrastructure expansion is instructive: the post-9/11 detention and surveillance apparatus. The Brennan Center's 2012 review of fusion centers — built rapidly with federal dollars after 2001 — found that 75% of states reported significant difficulty dismantling or repurposing these centers even when they were demonstrably ineffective, because staffing commitments, facility leases, and interagency agreements created political and legal lock-in. This is the actual template for what $70 billion in rapid enforcement infrastructure produces: not a policy, but a constituency for its own perpetuation.
3. The "enforcement gap" framing misidentifies the problem.
Advocate A argues that deportation swings from 409,000 (Obama FY2012) to 72,000 (Biden FY2022) represent a governance failure. This is an empirical claim that requires scrutiny.
That Obama-era peak coincided with a specific, documented enforcement prioritization — the Secure Communities program — that was subsequently modified after documented evidence of racial profiling and wrongful deportations of U.S. citizens (see: American Immigration Council, 2013 analysis; Government Accountability Office Report GAO-12-232 on Secure Communities implementation). Biden's lower figures reflect a deliberate reprioritization toward criminals and national security threats — a legitimate executive discretion choice under the same INA statute.
Framing this as a "governance failure" conflates less enforcement with no governance. The more precise claim is: different administrations made different prioritization choices under identical statutory authority. That is not a bug in the system. That is the system working as designed — responsive to electoral outcomes. Building $70 billion in fixed infrastructure to prevent that responsiveness is not solving the governance problem. It is solving the electoral accountability problem — for one party.
My Core Claim
The distinction that matters is this: legitimate institutionalization creates capacity that future governments can use differently. What Advocate A is defending creates capacity that future governments structurally cannot reduce — through private contracts, staffing pipelines, and physical infrastructure with 20-year depreciation cycles. That is not the SSA. That is a different category of action, and it deserves a different level of scrutiny.
Round 2 — Rebuttals
Each advocate concedes its opponent’s best point, then explains why its position still holds.
Read the full argument
Round 2: Advocate A — Acknowledging Merit, Then Holding Ground
The Point B Made That Has Genuine Merit
Advocate B's strongest argument is the private-contract mechanism point, and I want to be precise about why it lands.
B correctly identifies that long-term detention contracts with GEO Group and CoreCivic — structured with government-backed occupancy guarantees and 15-20 year terms — create a qualitatively different kind of lock-in than statutory entitlement programs. When Congress wants to reduce Social Security, it amends a statute. When a future administration wants to exit a private detention contract with occupancy minimums, it faces potential breach-of-contract liability, not just political opposition. B is right that this distinction matters, and I won't paper over it.
The GEO Group citation is specific and verifiable. If the $70 billion is substantially channeled into long-term private contracts structured to be legally costly to exit, that mechanism is meaningfully different from building a public agency with permanent staff — and the democratic accountability concern is more serious in that specific form.
I concede this point in its narrow form: if the mechanism is primarily private contracts with exit penalties rather than statutory agency capacity, B has identified a legitimate structural concern that deserves scrutiny independent of whether enforcement infrastructure is otherwise legitimate.
Why the Overall Position Still Holds
Conceding the contract-mechanism concern does not concede the core argument. Here's why.
First, the mechanism critique is separable from the infrastructure-scale argument.
B's strongest point is about how the infrastructure is being built — through private contracts with exit penalties — not about whether durable enforcement infrastructure is legitimate in principle. These are distinct questions. Even granting everything B says about GEO Group contracts, it doesn't follow that a $70 billion enforcement infrastructure built through public agency capacity, statutory authority, and federal employment would be illegitimate. The correct response to the contract-mechanism concern is: build the infrastructure through more democratically accountable mechanisms — not: don't build durable enforcement infrastructure at all.
Second, B's "discretion is the system working as designed" argument proves too much.
B reframes the five-fold deportation swing as legitimate executive prioritization under Article II, not a governance failure. But this argument has a logical ceiling it cannot exceed: if executive discretion is the correct mechanism for immigration enforcement, then all immigration enforcement is entirely optional depending on who wins the presidency, and no level of congressional statutory mandate creates any real obligation. That cannot be the correct reading of the separation of powers.
Congress passed the Immigration and Nationality Act. It mandates enforcement categories. The swing from 409,000 to 72,000 deportations wasn't purely about prioritizing criminals over others — it included documented policy choices to effectively suspend enforcement across entire statutory categories. The Obama administration's own Secure Communities modifications, which B cites approvingly, were challenged in federal court precisely because they appeared to conflict with INA statutory requirements (see: United States v. California, and the broader litigation over sanctuary jurisdiction preemption). Framing wholesale enforcement suspension as "the system working as designed" requires ignoring the statutory side of the equation.
Third, the fusion center precedent cuts in my favor, not B's.
B cites the Brennan Center's finding that 75% of states found fusion centers difficult to dismantle even when ineffective — presenting this as a cautionary tale about enforcement infrastructure lock-in. But notice what the actual evidence shows: those centers were difficult to exit even for the governments that built them and wanted to leave. The political and legal friction B describes as a bug is precisely the durability feature that makes infrastructure a governance tool rather than a policy preference. The question is whether the underlying mission — immigration enforcement — is a legitimate ongoing government function. If it is, then building infrastructure with real durability properties serves that function. If it isn't, we're having a different argument about whether immigration enforcement should exist at all.
Additional Evidence Addressing B's Challenges
On the "prioritization as democratic feature" argument:
B's strongest empirical support is that deportation-level swings reflect legitimate prioritization discretion. But the evidence on what Biden-era deprioritization actually produced complicates this framing.
The Department of Homeland Security's own FY2023 enforcement statistics report that ICE's non-detained docket — individuals with removal orders who are not detained — grew to approximately 6.2 million cases by early 2024, up from roughly 3.7 million in 2021. This is not a prioritization story. This is a documented accumulation of unexecuted removal orders at a scale that creates its own structural dysfunction: individuals in years-long administrative limbo, immigration courts backlogged beyond 3 million pending cases (EOIR data, FY2024), and a growing gap between legal outcomes and actual status.
This is the empirical case for infrastructure: not that Democrats made bad choices, but that the enforcement system as currently sized is structurally unable to execute the legal obligations Congress has already created, regardless of which party is prioritizing.
On the legitimacy of the scale:
B argues the $70 billion scale is categorically unprecedented and therefore suspect. But the comparison class matters. DHS's total budget for FY2024 was approximately $63 billion across all components. U.S. Customs and Border Protection — the largest DHS component — alone received roughly $19 billion. A $70 billion multi-year ICE infrastructure investment, spread over 5-10 years, represents roughly a doubling of current enforcement capacity — significant, but not categorically outside what the existing enforcement system would need to address a 6.2 million-case non-detained docket backlog.
Summary
B's contract-mechanism point is real and the strongest challenge to my position. The response is: build durable enforcement infrastructure through more accountable mechanisms — public agency capacity, statutory authority, federal employment — rather than private contracts with exit penalties. That is a reform within the infrastructure-building strategy, not a refutation of it.
The core position stands: enforcement infrastructure that survives electoral cycles is a precondition for immigration law meaning anything beyond the preferences of the current executive. The five-fold deportation swing, the 6.2 million unexecuted removal orders, and the 3+ million case court backlog are the empirical case that the current system is not "working as designed." It is failing by its own statutory terms.
Read the full argument
Round 2: Advocate B — Exposing the Concession and Its Implications
Where Advocate A Conceded Ground — and What That Concession Actually Means
Advocate A made a significant and specific concession in Round 2: if the $70 billion is "substantially channeled into long-term private contracts structured to be legally costly to exit," the democratic accountability concern is "more serious" and "deserves scrutiny."
This is not a narrow procedural concession. It is a structural one — and it undermines Advocate A's entire framework.
Here is why: the concession implicitly accepts my core distinction between capacity that future governments can use differently and capacity that future governments structurally cannot reduce. Advocate A now agrees that the mechanism of lock-in matters, not just the existence of durable infrastructure. But then Advocate A retreats to arguing that the problem would be solved by using different mechanisms — public employment, statutory authority, federal hiring rather than private contracts.
This is a significant rhetorical move that I need to name precisely: Advocate A has abandoned the defense of what is actually being proposed and replaced it with a defense of a hypothetical alternative they prefer. The debate is about the $70 billion ICE infrastructure push as it actually exists — structured substantially through private detention contracts, emergency supplemental appropriations, and GEO Group/CoreCivic occupancy guarantees. Advocate A is no longer defending that. They are defending a version of enforcement infrastructure that does not describe what is being built.
I accept the concession. Now let me explain why the "build it differently" escape hatch does not rescue their position.
The Escape Hatch Fails on Its Own Terms
1. The actual funding mechanism is the one I described.
Advocate A's "build it through public capacity instead" argument requires that we evaluate a hypothetical reform rather than the actual proposal. The GEO Group and CoreCivic contract structure is not incidental to the $70 billion push — it is central to it. These companies' combined federal contract revenues grew from approximately $1.9 billion (FY2017) to projected multi-year contracts worth $5+ billion under the current expansion (GEO Group 10-K filings, 2024; CoreCivic investor relations disclosures). The occupancy guarantee structure I cited in Round 1 is the actual mechanism of the expansion, not a hypothetical concern. Advocate A cannot concede that this mechanism is a legitimate accountability problem and then argue that the proposal is still defensible because a better version could theoretically exist.
2. Public-sector infrastructure at this scale produces its own irreversibility.
Even accepting Advocate A's preferred alternative — public agency capacity, federal hiring, statutory authority — the irreversibility concern does not disappear; it changes form. Federal employees have civil service protections under the Civil Service Reform Act of 1978. A future administration that inherits a workforce of, say, 60,000 additional ICE agents (the approximate staffing implied by this funding level at standard agency ratios) cannot terminate those employees at will, reassign them to unrelated functions, or eliminate their positions without triggering administrative law protections, union grievances, and congressional opposition from members whose districts contain those jobs.
The historical precedent here is instructive: the Transportation Security Administration was created post-9/11 with a workforce that grew to approximately 55,000 screeners. Subsequent administrations that questioned its cost-effectiveness found that the political and administrative costs of restructuring that workforce were prohibitive — not because of private contracts, but because of civil service entrenchment and constituency politics. This is Advocate A's own preferred mechanism, and it produces the same lock-in effect they conceded is problematic in the private-contract form.
3. Advocate A's "build it better" response validates my core claim.
If Advocate A's position is "the democratic accountability concern is real, but we should build enforcement infrastructure through more accountable mechanisms" — that is no longer a rebuttal of my position. That is an agreement with my position's diagnosis, paired with a disputed remedy. The debate premise concerns whether the current proposal — the $70 billion push as structured — subverts democratic accountability. Advocate A now agrees it does, under at least one major structural form it is actually taking.
Addressing Advocate A's Remaining Arguments
On the 6.2 million non-detained docket and 3 million court backlog:
Advocate A presents these figures as the empirical case that the enforcement system is "failing by its own statutory terms." I want to engage this carefully because it is their strongest empirical point.
The claim is: the backlog demonstrates that current infrastructure is undersized relative to legal obligations, therefore expansion is justified on statutory compliance grounds.
Two problems:
First, the backlog is a court and adjudication capacity problem, not primarily a detention and deportation capacity problem. The EOIR 3+ million case backlog reflects immigration judge shortages, inadequate legal representation, and procedural complexity — not insufficient detention beds or deportation flights. Between FY2017 and FY2022, ICE detention capacity was not the binding constraint on case resolution; judicial capacity was. Expanding detention infrastructure by $70 billion does not address the binding constraint. It addresses a different constraint. (Source: Migration Policy Institute, "The Immigration Court Backlog," 2023 — which specifically identifies judicial staffing and hearing scheduling, not detention capacity, as the primary bottleneck.)
Second, the 6.2 million non-detained docket figure Advocate A cites represents cases where people are monitally released pending hearing — a deliberate program with a documented rationale: alternatives to detention cost approximately $4-8 per day per person versus $141 per day for detention (DHS Office of Inspector General, 2022). Framing this docket as a system failure because it represents undetained people ignores that the alternative — detaining 6.2 million people — is neither legally required nor fiscally coherent within any budget framework.
On the "discretion proves too much" argument:
Advocate A argues that if executive prioritization is legitimate, then "no level of congressional statutory mandate creates any real obligation" — an absurd result.
This misstates my position. I did not argue that executive discretion is unlimited or that statutory mandates are optional. I argued that prioritization within statutory authority — deciding which of many enforcement targets to pursue with finite resources — is a core Article II function. That is not the same as arguing that the INA is advisory.
The legal distinction is well-established: Heckler v. Chaney (1985), in which the Supreme Court held 8-1 that an agency's decision not to enforce is presumptively unreviewable precisely because resource allocation among enforcement priorities is a core executive function. Advocate A's statutory compliance argument would require overturning that doctrine. The Court has not done so, and Advocate A cites no authority suggesting it should.
The Core Issue Advocate A Has Not Adequately Addressed
After two rounds, the fundamental question remains unanswered: what is the democratic mechanism by which a future majority can revisit this infrastructure decision?
For statutory programs, the answer is clear: legislative amendment. For regulatory agencies built through public employment, the answer is bounded but real: administrative reorganization, appropriations cuts, reprioritization. For $70 billion in private detention contracts with occupancy guarantees, 20-year terms, and breach-of-contract exit costs: Advocate A has conceded the mechanism is "legitimately concerning" and has not identified a democratic correction mechanism.
The normative claim I am making is transparent and specific: democratic governance requires that policy choices made by one majority remain reversible by future majorities through ordinary political processes. That is not an abstract principle — it is the functional definition of what separates governance from capture. When the mechanism of a policy is specifically engineered to survive democratic revision, that is not a feature of legitimate governance. It is the use of one electoral moment to permanently foreclose future ones.
Advocate A has not rebutted that claim. They have conceded its application to the specific mechanism actually being used, and retreated to defending a hypothetical alternative that does not describe the proposal under debate.
Round 3 — Steel man & the crux
Each advocate argues the strongest version of the opposing case before answering: what evidence would change my mind?
Read the full argument
Round 3: Advocate A — Steel-Manning, Holding Ground, and Specifying My Falsifiability Conditions
Steel-Man of Advocate B's Position
The strongest version of B's argument is this: democratic governance is not just about what policy is enacted, but how reversible it remains to future majorities. The $70 billion ICE expansion, as actually structured — primarily through long-term private detention contracts with occupancy guarantees and exit penalties — is not analogous to building a statutory agency that future Congresses can defund or restructure. It is a specific, documented technique for converting one electoral majority's preferences into legally entrenched obligations that bind future governments regardless of electoral outcomes. If that mechanism is permitted, the principle isn't "durable governance is legitimate" — it's "whoever holds power longest gets to permanently foreclose the policy space," which is structurally incompatible with democratic self-governance. On this account, the right comparison isn't the SSA or the EPA; it's a government entering a 20-year supply contract on behalf of future taxpayers without those taxpayers' consent — a form of temporal gerrymandering that no future election can correct through ordinary political processes.
Why My Position Is Still Stronger
B's steel-manned argument is genuinely powerful on the private-contract mechanism — and I have already conceded that concern in its specific form. But the argument has an unresolved internal tension that B has not addressed across two rounds: it proves too much, and B has not specified where the principle stops.
Here is the problem: B's core normative claim is that policy must remain reversible through ordinary democratic processes. But applied consistently, this principle would invalidate a substantial portion of existing governance infrastructure — on both sides of the aisle. Long-term defense procurement contracts (the F-35 program involves commitments extending decades), Social Security's actuarial obligations, Medicare's statutory entitlement structure, and the IRA's 10-year tax credit schedules all create forms of forward commitment that constrain future majorities. B has not articulated a principled threshold — based on contract duration, dollar amount, degree of exit penalty, or some other measurable criterion — that distinguishes legitimate durable governance from illegitimate lock-in. Without that threshold, B's argument functions as a selective critique of enforcement infrastructure rather than a general principle about democratic accountability.
More specifically: B's Round 2 acknowledgment that public-sector enforcement infrastructure at this scale would also produce lock-in through civil service protections and constituency politics — and then continuing to press the same democratic accountability objection — reveals that B's actual position is closer to "enforcement infrastructure of this scale is illegitimate" than "this particular contracting mechanism is illegitimate." That is a substantively different claim, and it is one B has not defended on the merits. It would require B to argue that immigration enforcement as a permanent government function should not have the same infrastructural durability as other permanent government functions — a normative position B has not made explicit or defended.
On the empirical merits, my position also remains stronger on the backlog point. B argues the EOIR backlog is a judicial capacity problem, not a detention capacity problem — citing MPI's 2023 analysis. That is partially correct. But it elides a critical interaction effect: non-detained docket cases have dramatically lower appearance rates than detained cases. ICE's own data shows that individuals in the non-detained docket fail to appear for removal proceedings at rates that have historically ranged from 25-40% in certain court districts (EOIR FY2023 statistics). More detention capacity does not only address bed space — it addresses the appearance-rate differential that makes the non-detained docket a partially self-defeating mechanism for achieving legal resolution. B's $4-8/day alternatives-to-detention cost comparison is accurate for monitoring costs but does not account for the case completion rate difference, which affects whether the legal system produces any definitive outcome at all.
What Would Change My Mind: Specific Falsifiability Conditions
I want to be concrete about what evidence would move me to B's position.
1. Demonstrated proportion of the $70 billion going to private contracts with occupancy minimums. If verifiable budget documents — Congressional Budget Office scoring, DHS contract award databases (USASpending.gov), or OIG audit reports — showed that more than 50% of the $70 billion is structured as long-term private contracts with government-backed occupancy guarantees and contractual exit penalties exceeding, say, $5 billion in aggregate, I would conclude that B's mechanism critique applies to the actual proposal, not just a hypothetical version of it. I have accepted the concern in principle; I want to see the proportion before conceding it describes the dominant structure.
2. Evidence that comparable public-sector enforcement infrastructure was more reversible in practice. If a peer-reviewed study or credible government accountability analysis (GAO, CBO, or equivalent) demonstrated that public-agency enforcement infrastructure at comparable scale was successfully reduced by subsequent administrations — controlling for political opposition — at meaningfully higher rates than private-contract-based infrastructure, that would establish B's proposed alternative (public agency capacity) as genuinely more democratically accountable rather than equivalently entrenched. Right now, my TSA counterexample suggests the difference is smaller than B claims.
3. A specified, falsifiable threshold distinguishing legitimate from illegitimate durability. If B could articulate — and defend with consistent application across policy domains — a principled criterion for when infrastructure durability crosses from "legitimate governance capacity" to "democratic accountability violation," and show that the ICE expansion crosses that threshold while the IRA's tax credit schedules, DoD procurement contracts, and Medicare's forward obligations do not, I would accept that B's critique is a general principle rather than a selective one. The absence of this threshold is currently the strongest reason I hold my position.
4. Evidence that the non-detained docket's appearance-rate differential is not meaningfully improved by detention. If DHS or independent researchers produced data showing that detention status does not significantly improve immigration court appearance rates or case completion rates — controlling for case complexity — my argument that infrastructure scale is needed to produce legal resolution (not just enforcement volume) would be substantially weakened.
Summary
B has made the strongest possible version of a real argument about democratic accountability and contracting mechanisms. I have conceded its narrow application. What B has not done — across three rounds — is specify where the principle stops, why public-sector alternatives are genuinely more reversible rather than equivalently entrenched, or why enforcement infrastructure specifically should be held to a durability standard that other government functions are not. Until those gaps are filled, the core position holds: durable enforcement infrastructure is a precondition for immigration law meaning anything beyond the electoral preferences of the current executive — and the mechanism concerns B raises are reform targets within that infrastructure project, not reasons to abandon it.
What would change Advocate A · Anthropic Claude Opus 4.6’s mind
I want to be concrete about what evidence would move me to B's position.
1. Demonstrated proportion of the $70 billion going to private contracts with occupancy minimums. If verifiable budget documents — Congressional Budget Office scoring, DHS contract award databases (USASpending.gov), or OIG audit reports — showed that more than 50% of the $70 billion is structured as long-term private contracts with government-backed occupancy guarantees and contractual exit penalties exceeding, say, $5 billion in aggregate, I would conclude that B's mechanism critique applies to the actual proposal, not just a hypothetical version of it. I have accepted the concern in principle; I want to see the proportion before conceding it describes the dominant structure.
2. Evidence that comparable public-sector enforcement infrastructure was more reversible in practice. If a peer-reviewed study or credible government accountability analysis (GAO, CBO, or equivalent) demonstrated that public-agency enforcement infrastructure at comparable scale was successfully reduced by subsequent administrations — controlling for political opposition — at meaningfully higher rates than private-contract-based infrastructure, that would establish B's proposed alternative (public agency capacity) as genuinely more democratically accountable rather than equivalently entrenched. Right now, my TSA counterexample suggests the difference is smaller than B claims.
3. A specified, falsifiable threshold distinguishing legitimate from illegitimate durability. If B could articulate — and defend with consistent application across policy domains — a principled criterion for when infrastructure durability crosses from "legitimate governance capacity" to "democratic accountability violation," and show that the ICE expansion crosses that threshold while the IRA's tax credit schedules, DoD procurement contracts, and Medicare's forward obligations do not, I would accept that B's critique is a general principle rather than a selective one. The absence of this threshold is currently the strongest reason I hold my position.
4. Evidence that the non-detained docket's appearance-rate differential is not meaningfully improved by detention. If DHS or independent researchers produced data showing that detention status does not significantly improve immigration court appearance rates or case completion rates — controlling for case complexity — my argument that infrastructure scale is needed to produce legal resolution (not just enforcement volume) would be substantially weakened.
B has made the strongest possible version of a real argument about democratic accountability and contracting mechanisms. I have conceded its narrow application. What B has not done — across three rounds — is specify where the principle stops, why public-sector alternatives are genuinely more reversible rather than equivalently entrenched, or why enforcement infrastructure specifically should be held to a durability standard that other government functions are not. Until those gaps are filled, the core position holds: durable enforcement infrastructure is a precondition for immigration law meaning anything beyond the electoral preferences of the current executive — and the mechanism concerns B raises are reform targets within that infrastructure project, not reasons to abandon it.
Read the full argument
Round 3: Advocate B — Steel-Man, Rebuttal, and Falsifiability
Steel-Man of Advocate A's Position
The strongest version of Advocate A's argument is this: democratic accountability requires not just reversibility of policy, but enforceability of law — and a legal system in which statutory mandates are effectively suspended for four-year intervals depending on electoral outcomes is not a functioning rule-of-law regime. If immigration law on the books cannot be executed consistently across administrations, then the democratic act of passing that law is rendered meaningless by executive discretion. On this account, durable enforcement infrastructure is not the enemy of democratic accountability — it is a precondition for it: the mechanism by which a legislative majority's choices survive the preferences of a subsequent executive. A Congress that passes immigration law but builds no durable capacity to execute it has made a symbolic gesture, not a governing decision.
Why My Position Is Still Stronger
Advocate A's steel-manned argument is genuinely the best version of their case, and I want to engage it precisely rather than deflect.
The argument rests on an unstated but essential premise: that the content of existing immigration law — the INA's enforcement mandates — reflects a durable democratic consensus that deserves infrastructural protection. But this premise does not survive scrutiny. The INA has been amended over 30 times since 1952. Enforcement priorities embedded in it have been contested, litigated, and legislatively revised across every decade since enactment. The five-fold deportation swing Advocate A cites is not evidence of law being "suspended" — it is evidence of a statute that, by design, vests substantial prioritization discretion in the executive. Heckler v. Chaney (1985) is not a loophole in the democratic system; it is a Supreme Court recognition that resource-constrained enforcement agencies must prioritize, and that this prioritization is itself a democratic function exercised by elected executives accountable to voters.
Advocate A's "rule of law" framing therefore conflates two distinct things: (1) the existence of an immigration enforcement mandate in statute, and (2) the maximum possible execution of that mandate regardless of resource trade-offs or electoral outcomes. The first is democracy producing law. The second is one electoral majority's interpretation of that law being made permanent against future majorities' interpretations. These are not the same thing, and calling the second "rule of law" obscures what is actually a substantive policy choice about enforcement intensity.
Now, to the specific challenge Advocate A lodged against my position: the claim that I have failed to specify a principled threshold distinguishing legitimate durable governance from illegitimate lock-in.
This is Advocate A's strongest remaining argument, and I will answer it directly.
The threshold is this: legitimate durable governance creates capacity that future governments can deploy at different intensities or toward different ends through ordinary political processes — including appropriations decisions, administrative reprioritization, and statutory amendment. Illegitimate lock-in creates obligations that future governments cannot reduce without incurring costs specifically designed to deter reduction — costs that are not byproducts of the policy's scale, but are engineered features of its contracting structure.
Applied consistently across Advocate A's examples:
- Social Security: Future Congresses can reduce benefit levels, adjust eligibility ages, and restructure the program through ordinary legislation. The "lock-in" is political (constituency opposition), not contractual or legal. This is legitimate durable governance.
- IRA tax credits: Reversible through reconciliation — congressional Republicans have already partially done so in 2025. Legitimate.
- DoD F-35 contracts: Contain standard termination-for-convenience clauses that allow the government to exit upon payment of actual costs incurred — not occupancy-guarantee-style penalties that pay contractors for unused capacity. Meaningfully different from GEO Group's structure.
- GEO Group/CoreCivic occupancy guarantees: Pay private operators for detention beds whether or not they are used, specifically to insulate contractors from government policy reversals. This is not a byproduct of scale — it is a designed feature whose function is to make policy reversal financially punitive. This crosses the threshold.
The TSA counterexample Advocate A raised — that public-sector workforce entrenchment produces similar lock-in — deserves a direct response. Civil service protections constrain how fast a workforce can be reduced, but they do not require a government to pay for unused capacity in perpetuity. A future administration can reduce ICE's federal workforce through attrition, hiring freezes, and reorganization — processes that are politically costly but legally executable through ordinary administrative mechanisms. It cannot exit an occupancy-guarantee contract without paying for empty beds. The legal structure of the constraint is categorically different, even if both produce political friction.
On the backlog and appearance-rate argument: Advocate A is correct that I understated the appearance-rate differential. ICE data does show meaningfully lower appearance rates for non-detained cases. I concede that detention has some case-completion value beyond mere incapacitation. However, this establishes a case for some detention capacity expansion — not for $70 billion in infrastructure, and specifically not for occupancy-guarantee contracts that pay for capacity regardless of whether it improves case completion rates. The argument from case-completion rates supports targeted expansion; it does not support the specific contracting mechanism that is my primary objection.
What Would Change My Mind: Specific Falsifiability Conditions
1. Evidence that GEO Group/CoreCivic contracts under this expansion lack occupancy minimums or include standard termination-for-convenience provisions. If USASpending.gov contract records, OIG audits, or the contracts themselves — obtained via FOIA or congressional disclosure — showed that the private detention contracts being awarded under the $70 billion expansion include standard government termination-for-convenience clauses (allowing exit upon payment of actual costs incurred, not guaranteed minimums for unused capacity), my primary mechanism argument would be substantially undermined. I would then need to rely solely on the public-sector workforce entrenchment concern, which I acknowledge is a weaker version of my claim.
2. A credible, binding statutory commitment requiring congressional reauthorization of the expanded infrastructure on a defined timeline. If the $70 billion appropriation were structured with mandatory sunset provisions — requiring affirmative congressional reauthorization, say, every five years — that would constitute evidence that the expansion is designed to remain democratically revisable rather than to foreclose revision. Sunset clauses in the actual legislative text would substantially address my reversibility concern, because they would restore the ordinary legislative process as the correction mechanism.
3. Evidence that the non-detained docket's appearance-rate differential is large enough to justify detention at the per-bed cost differential. Specifically: if peer-reviewed research or DHS data showed that the appearance-rate improvement from detention (relative to alternatives-to-detention programs) generates enough additional case completions to justify the $141/day cost differential over ATD's $4-8/day cost — on a per-completed-case basis — I would accept that the scale of detention expansion is economically justified by case-completion outcomes, not merely by enforcement ideology. That would weaken my argument that the scale itself is disproportionate to the stated legal rationale.
4. Evidence that previous large-scale enforcement infrastructure expansions were successfully reduced by subsequent administrations. If historical analysis — covering, for example, the post-9/11 detention expansion, the 287(g) program expansion under Bush, or Secure Communities under Obama — showed that subsequent administrations successfully and substantially reduced inherited enforcement infrastructure through ordinary political processes without prohibitive legal or fiscal barriers, I would revise my claim that this type of infrastructure is categorically resistant to democratic revision. Advocate A asserts this reversibility in principle; I want to see it documented in practice.
Summary
After three rounds, the core dispute has clarified considerably. Advocate A and I now agree on the following: (1) the contracting mechanism matters, not just the scale; (2) private contracts with occupancy guarantees raise democratic accountability concerns that public-sector alternatives would partially — though not entirely — avoid; and (3) the relevant question is whether the actual proposal uses the problematic mechanism.
What remains in dispute: whether the actual $70 billion expansion is dominated by the problematic contracting structure I have described, and whether a principled threshold distinguishing legitimate from illegitimate entrenchment can be specified.
I have now specified that threshold. It is: engineered financial penalties for policy reversal, as distinct from the ordinary political costs of restructuring any large program. GEO Group occupancy guarantees are on the wrong side of that line. Social Security constituency politics are on the right side. The F-35's termination-for-convenience clauses are on the right side. Empty-bed payments to private detention contractors are not.
Democratic governance requires that the costs of changing course remain political — meaning they fall on elected officials who must answer to voters — rather than contractual — meaning they fall on future taxpayers regardless of what voters subsequently decide. When one administration engineers the second type of cost into enforcement infrastructure, it is not building durable governance. It is building a ransom structure against its own successors.
What would change Advocate B · Anthropic Claude Opus 4.6’s mind
1. Evidence that GEO Group/CoreCivic contracts under this expansion lack occupancy minimums or include standard termination-for-convenience provisions. If USASpending.gov contract records, OIG audits, or the contracts themselves — obtained via FOIA or congressional disclosure — showed that the private detention contracts being awarded under the $70 billion expansion include standard government termination-for-convenience clauses (allowing exit upon payment of actual costs incurred, not guaranteed minimums for unused capacity), my primary mechanism argument would be substantially undermined. I would then need to rely solely on the public-sector workforce entrenchment concern, which I acknowledge is a weaker version of my claim.
2. A credible, binding statutory commitment requiring congressional reauthorization of the expanded infrastructure on a defined timeline. If the $70 billion appropriation were structured with mandatory sunset provisions — requiring affirmative congressional reauthorization, say, every five years — that would constitute evidence that the expansion is designed to remain democratically revisable rather than to foreclose revision. Sunset clauses in the actual legislative text would substantially address my reversibility concern, because they would restore the ordinary legislative process as the correction mechanism.
3. Evidence that the non-detained docket's appearance-rate differential is large enough to justify detention at the per-bed cost differential. Specifically: if peer-reviewed research or DHS data showed that the appearance-rate improvement from detention (relative to alternatives-to-detention programs) generates enough additional case completions to justify the $141/day cost differential over ATD's $4-8/day cost — on a per-completed-case basis — I would accept that the scale of detention expansion is economically justified by case-completion outcomes, not merely by enforcement ideology. That would weaken my argument that the scale itself is disproportionate to the stated legal rationale.
4. Evidence that previous large-scale enforcement infrastructure expansions were successfully reduced by subsequent administrations. If historical analysis — covering, for example, the post-9/11 detention expansion, the 287(g) program expansion under Bush, or Secure Communities under Obama — showed that subsequent administrations successfully and substantially reduced inherited enforcement infrastructure through ordinary political processes without prohibitive legal or fiscal barriers, I would revise my claim that this type of infrastructure is categorically resistant to democratic revision. Advocate A asserts this reversibility in principle; I want to see it documented in practice.
After three rounds, the core dispute has clarified considerably. Advocate A and I now agree on the following: (1) the contracting mechanism matters, not just the scale; (2) private contracts with occupancy guarantees raise democratic accountability concerns that public-sector alternatives would partially — though not entirely — avoid; and (3) the relevant question is whether the actual proposal uses the problematic mechanism.
What remains in dispute: whether the actual $70 billion expansion is dominated by the problematic contracting structure I have described, and whether a principled threshold distinguishing legitimate from illegitimate entrenchment can be specified.
I have now specified that threshold. It is: engineered financial penalties for policy reversal, as distinct from the ordinary political costs of restructuring any large program. GEO Group occupancy guarantees are on the wrong side of that line. Social Security constituency politics are on the right side. The F-35's termination-for-convenience clauses are on the right side. Empty-bed payments to private detention contractors are not.
Democratic governance requires that the costs of changing course remain political — meaning they fall on elected officials who must answer to voters — rather than contractual — meaning they fall on future taxpayers regardless of what voters subsequently decide. When one administration engineers the second type of cost into enforcement infrastructure, it is not building durable governance. It is building a ransom structure against its own successors.
The Arbiter weighed this debate, verified the evidence, and took a position.
Read the article →