Administrative Agencies and the Power to Execute

Congress legislates; the executive executes. Agencies are executive staff, so on a unitary-executive reading they should answer to the elected President — which is where the law has now moved (Loper Bright stripped their power to “say what the law is,” 2024; Trump v. Slaughter overruled Humphrey’s Executor and restored at-will removal, 2026). The live puzzle Chris names: the President has robust power to decline to enforce a law (DACA) yet, until 2026, could not fire the staff executing it — backwards, if agencies are just the executive’s hands. The deeper fork is who fills the gaps Congress leaves: an insulated expert agency (illegitimate — unaccountable) or the elected executor (legitimate — removable). The whole fight is who holds the unavoidable discretion, and whether that holder is accountable.

Links: Corruption (the test that decides whether directing an agency is legitimate or corrupt: statutory purpose vs. private end; and insulation → the unaccountable actor corruption warns of), The Represented Unit (the 16th+17th built the administrative state this page is about — the downstream chapter), The Industrial Revolution and the Remaking of Political Order (the global origin — Wilson/Goodnow imported the continental “science of administration”; this page is where that lands in US law), Government Formation (the Constitution as a distributed protocol; the executive as the fast-execution node), Registration and Scope Creep (agencies as the scope-creep engine), Wickard v. Filburn, Originalism and the Limiting Document (read the vesting clauses as written), Chiafalo v. Washington, The Gödel Governance Problem

The constitutional problem — the “fourth branch”

The Constitution vests all legislative power in Congress (Art. I §1), the executive power in the President (Art. II §1), and charges him to “take Care that the Laws be faithfully executed” (Art. II §3). There are three branches, and the enumeration is a limiting one (see Originalism and the Limiting Document). An administrative agency sits athwart that scheme: it makes rules (looks legislative), adjudicates violations (looks judicial), and enforces them (executive) — often all three, inside one body, staffed by unelected experts partly insulated from the President. This is the “fourth branch” the text never authorizes, and it is the structure the 16th + 17th Amendments made politically possible: unlimited revenue plus senators unmoored from state control. This page is the downstream chapter of that story.

Chris’s position — a unitary-executive reading

Stated and developed with Chris; the through-line is that agencies are executive staff, not a branch:

1. Agencies shouldn’t make law. “It makes no sense these agencies can make law; the Constitution has no provision for such a thing.” Loper Bright (below) is “a big win.”
2. Humphrey’s Executor makes little sense; Seila was right to revisit — and Chris’s view is now the law (Slaughter, 2026, finished the job).
3. Congress legislates, the executive administers, so every agency “falls under the privy of the executive to execute.” The President is the accountable head of the whole executive branch.
4. The asymmetry is backwards (see below): a President may issue an EO to not enforce part of a law (DACA) but — pre-2026 — could not hire and fire the agency at will. More control over whether to execute than over who executes.
5. Congress’s vagueness passes discretion to the executor — “if Congress is too lazy to make things explicit enough, then the executor gets to fill in the gaps” (the accountable President, not an insulated agency). But on reflection this is Chris’s fallback, not his preference: the fix should be to stop the vagueness (nondelegation), because normalizing executive gap-filling just lets each administration legislate via the executive — see his resolution below.
6. Preferred end-state: “the executive is the one who determines how to execute clear law, and not hand everything over to an agency.” Some administrative staff is unavoidable — administration must be done — but it is staff under the executive, not an autonomous rule-making state.

The four grounded cases

The Fed wrinkle (Trump v. Cook, 2026, companion). The same Court carved out the Federal Reserve (5-4, Roberts): central-bank independence is a “distinct historical tradition,” and any change “must come from Congress, not the courts.” A pure unitary-executive theory has no room for this exception — it’s either a principled sui generis carve-out or an unprincipled one that proves the rule. (Open question below.)

The asymmetry Chris flagged — is “backwards” right?

The puzzle: a President wields robust power to decline enforcement (DACA) but had restricted power over personnel (removal). Enforcement discretion is deep-rooted — Heckler v. Chaney (1985) makes agency non-enforcement presumptively unreviewable, on the logic that the executive can’t pursue every violation and must set priorities; choosing them is executing. So the President could effectively nullify a statute’s application while being told he couldn’t fire the officials applying it. From a unitary-executive standpoint that is backwards: the President’s most core power (directing his own subordinates) was weaker than a more peripheral one (enforcement triage). The 2026 correction (Slaughter) resolves the asymmetry in exactly Chris’s direction — both powers now robust.

But the DACA half deserves a sharper cut than “he ignored the law.” There are two different acts inside DACA, and only one is clean:

• Non-enforcement (declining to deport a defined class) — squarely executive, Chaney-protected. Not controversial as a matter of power.
• Affirmative conferral (granting work authorization, a lawful-presence posture, downstream benefits) — this looks like making policy with the force of law, and it’s why DACA was litigated. Tellingly, DHS v. Regents (2020) struck the rescission as arbitrary-and-capricious under the APA without ever ruling on whether DACA itself was lawful — the Court dodged the merits. So DACA is not a pure “President ignores a law” specimen; the contested part was the affirmative program, i.e. the executive doing something that resembled legislating. The asymmetry is real, but the DACA example is doing double duty — it mixes clean non-enforcement with contested benefit-creation.

The tension inside Chris’s own position — and its resolution

Two of Chris’s commitments pull against each other, and naming the pull is where the position gets sharp (this is arguments-over-rhetoric applied to his view, not a gotcha):

• “Agencies shouldn’t make law” (Loper Bright), and
• “If Congress is vague, the executor fills the gaps.”

But gap-filling is a species of lawmaking. If a statute says “set safe emission levels” and the executive picks the number, that number carries the force of law — the very thing Loper Bright and the nondelegation worry are about. So the two commitments can’t both be about whether quasi-legislative discretion exists. The reconciliation — which Chris’s own words point to — is accountability, not absence:

The discretion Congress leaves is unavoidable. The question is never whether someone fills the gap, but who — and whether that someone is removable by the voters. Gap-filling by the elected, at-will- removable President is legitimate execution (one throat to choke); gap-filling by an insulated expert agency (the old Chevron regime) is illegitimate because unaccountable.

That is the exact same “who holds the discretion” structure as the corruption fork (whose purpose fixes the office?) and the represented-unit argument (who does the Senator answer to?). The vault keeps landing on it: the design question is almost never “should this power exist” but “who holds it, and can they be fired.”

The nondelegation fork — two cures for a lazy Congress

Chris’s “if Congress is too lazy, the executor fills the gaps” is one of two constitutional responses to vague statutes, and they conflict:

• (a) Force Congress to be specific — nondelegation revival. Under the lenient “intelligible principle” test (J.W. Hampton 1928; Whitman v. American Trucking 2001) Congress may delegate broadly. A revival (signaled in the Gundy 2019 Gorsuch dissent) would invalidate the vague delegation itself — there’d be no gap to fill, because Congress would have to make the call. This is the reading most faithful to “all legislative Powers… vested in Congress.”
• (b) Let the executor fill the gap (Chris’s lean). Given the gap, the accountable President fills it.

They diverge on the root: (a) says the gap shouldn’t exist; (b) says given it exists, keep the filler accountable. (b) is more workable (Congress will always be somewhat vague) but concedes that quasi-legislative discretion is permanent and merely relocates it to the executive — which is fine on accountability grounds but does not fully answer the Article I objection that someone other than Congress is making binding rules.

Chris’s resolution — (a), make Congress do the work. The executive-gap-filling route is itself an innocuous-thread pull: normalize it and “each administration would basically legislate via agency” — the exact thing Loper Bright was meant to stop, re-entering through the executive’s door instead of the agency’s. He doesn’t like executive gap-filling either; he only acknowledges the President has some discretion over the manner of execution (the irreducible residue). The primary cure should be forcing Congress to be explicit — put the accountability where the Constitution already put the power. So the stack is: nondelegation to shrink the gaps first; unitary-executive accountability only for the irreducible remainder.

Connection to corruption — insulation is the capture invitation

The removal fight is the corruption question one level down. An agency insulated from the elected executive answers to… whom? On the unitary view, a nominally-neutral but unremovable body drifts to serving its own institutional purpose — Olson-style capture, the exact “unaccountable actor serving a private/institutional end behind a public charter” the corruption page identifies as the masked failure mode. Humphrey’s “independence,” from this angle, wasn’t neutrality; it was an invitation to capture with no removal remedy. Restoring removal (Slaughter) is a re-accountability move: it doesn’t guarantee good agencies, it guarantees a fireable one — feedback where there was insulation (cf. the friction-vs- feedback distinction in Gödel governance).

Chris’s positions on the hard questions

• The Fed carve-out is illegitimate (“BOO”). Chris rejects Cook: “I don’t like how ‘traditionalism’ makes law — we have the text to answer that.” A textualist/originalist has no room for a sui generis central-bank exception invented from “distinct historical tradition”; the vesting clauses don’t carve out monetary policy, so neither should the Court. He reads Cook as SCOTUS dodging the bigger question it had just answered everywhere else — deciding the FTC cleanly then flinching at the Fed. (Note the tension the Court created for itself: it rejected the independence rationale in Slaughter and then honored it in Cook twelve hours later. Chris’s objection is that “tradition” is doing the work the text should do.)
• Accountability’s relocation risk is real — which is why the cure should be Congress, not the President. Chris grants the worry: presidential control “could turn into unraveling as each new president reshapes the landscape every time.” An executive-centered fix trades stable-but-unaccountable agencies for accountable-but-whipsawing ones. His conclusion follows directly: make Congress the accountable one. Durable rules set by the legislature (hard to pass, hard to undo — friction) beat rules that flip every four years with the White House. This is the same reason he picks nondelegation (a) over executive gap-filling (b) above: put both the power and the accountability on Congress, where the text already assigns them.

Will Congress do the work? — Chris’s diagnosis (pessimistic, and structural)

Chris’s honest answer to the question below: “I wish I had the answer… my hopes are not high.” But the pessimism is structural, not a mood — he names the exact failure paths, and they’re the deepest point of the session because they explain why the accountability cure is so hard to reach.

The proposed forcing function: let the President “rule via margins until Congress realizes it won’t stop without them” — i.e. make executive overreach salient enough that the legislature reclaims its own power in self-defense. This is the Madisonian bet: Federalist 51 assumed “the interest of the man must be connected with the constitutional rights of the place” — that an officeholder identifies with his branch and will defend its prerogatives, so ambition counteracts ambition.

Why the bet fails — the loyalty the design assumed was swapped out. Two failure paths, both fatal:

1. Co-partisan capture of the check. “Often a majority in Congress approves of the president, so they don’t mind it happening or even encourage it.” Madison’s mechanism assumed institutional loyalty; real officeholders have partisan loyalty. A congressman of the President’s party won’t defend congressional power against his own team — so “ambition counteracting ambition” becomes ambition aligning with ambition. (This is the “separation of parties, not powers” problem: the operative cleavage is party, not branch, so the branch-based check idles whenever one party holds both.)
2. The out-party routes around instead of repairing. “Those that do not like it will look for alternate legal paths such as SCOTUS rulings to get their way without fixing the problem.” The losers don’t reclaim legislative power (the front-door fix); they seek a judicial win — the enforcement-bypass again — which resolves the instant grievance while entrenching the bypass pattern for next time. Winning the case beats fixing the structure, every time, for whoever is currently losing.

The result is incentive-incompatible self-repair — the Gödel-governance core in one sentence: the structural fix requires every actor to forgo the advantage the broken structure hands them whenever they’re winning, so no one with the power to repair it has the incentive to. Restoring the original accountability requires getting the structure back to where it was, but each player prefers to exploit the current break while they’re ahead. Chris’s read on the recent wins (Loper Bright, Slaughter): “glad to see some people coming to their senses… but there is still more to go” — the courts have removed one bypass (agency lawmaking) and restored one control (removal), but the deeper repair (Congress reclaiming the legislative function it delegated away) runs straight into the incentive wall above, and courts can’t force that from outside.

Open Questions (still live)

• What minimal administrative staff is unavoidable, and how do you keep “staff that executes” from ratcheting back into “agency that legislates” (the scope-creep problem applied to the executive’s own machinery)? Chris’s end-state — the executive determines how to execute clear law, minimal handoff — needs a boundary that itself resists creep.
• Will Congress actually do the work? The nondelegation cure assumes a legislature willing to be explicit. The whole reason broad delegation arose is that being specific is politically costly (you own the hard call). If nondelegation forces specificity and Congress simply won’t, the result may be paralysis rather than accountability — is that an acceptable price, or does it just re-route the pressure back to the executive anyway?

Tags

politics, libertarian-law, history, philosophy