Why it matters

When you have to argue against something — a proposal a colleague is championing, a published strategy you think is wrong, a position you are about to face in a debate — the natural move is to lead with the objection you happen to find most damning. But the objection that bothers you is rarely the one that moves the room. Red-Team Advocate is the discipline of building the strongest case against a specific thing, on purpose, calibrated to the people you actually need to persuade — and arguing it in good faith at full strength, separate from whether you privately believe it.

For example: a colleague is pushing a vendor deal, and you have a bad feeling about it. The bad feeling, voiced as “I’m not sure about this,” changes nothing. Now run it the other way. You name the audience — the finance-minded executives who will approve or kill the deal — and you ask what would land with them: not your vague unease, but the principal-agent problem buried in the contract terms, the switching costs they will care about disproportionately, the attribution disputes a CFO has lived through before. You rank those by what hits hardest with that room, you phrase each one in their idiom, and — crucially — you name the deal’s strongest defence yourself, before they can raise it. That is an argument that survives contact with a prepared opponent. The bad feeling was not.

  • What it reveals. The strongest honest case against an artifact, built to land with a specific audience — the objections a prepared opponent could actually use, ranked by persuasive force rather than by which one you noticed first.
  • How it changes the read. You stop asking “what do I dislike about this?” and start asking “what would the strongest version of the opposition argue, in front of the people who decide — and have I answered their best reply before they make it?”
  • When to foreground it. You must argue against a specific named thing for an external audience — debate prep, a dissuasion brief, getting ready for hostile review — and you want ammunition that holds up, not a list of complaints.
  • What you’d miss without it. That an argument calibrated to you misfires in front of them; that an attack on a distorted version of the artifact collapses the moment the audience can read the real one; and that a brief which hides the other side’s best defence gets ambushed by exactly that defence.
  • Where it misleads. Pushed into performance it manufactures hostility — weak objections dressed up as devastating, attacks the artifact never earned, a case so one-sided a sceptical listener stops trusting it. An overstated brief does not just fail; it spends your credibility along with the argument.

Realtime examples

See real, dated analyses where this mode built the strongest case against a position in the news, ranked by what lands hardest → Red-Team Advocate on Main Street Independent

How to invoke it in Ora

You have a specific, named artifact — a strategy, a proposal, a published commitment — and you need the strongest case against it, argued for an external audience rather than surfaced as a fix-list for yourself.

Name the artifact, name the audience, and ask:

“Make the case against [this specific thing] for [this audience] — every angle, including the weak ones. I need to argue against it.”

The phrases make the case against, argue against this, give me ammunition, prep me for debate, and devil’s advocate are what route you here. Two details do most of the work. Name the artifact precisely — a specific proposal, not a whole domain; if you hand the mode an area rather than a thing, it will stop and ask rather than attack thin material. And name the audience — this mode argues in front of someone whose persuasion is the point, so “argue against this for a sceptical board” produces a far sharper brief than “argue against this,” which leaves the mode guessing at a generic critic.

One boundary worth knowing. If you want to know what to fix in your own plan before you ship it — vulnerabilities ranked by severity, with repairs — that is the sibling mode, red-team-assessment, not this one. Red-Team Advocate builds a case for someone else to weigh; it does not hand you a punch-list.

How it works

The cleanest illustration is the role this discipline is named after, because it was once a literal job. From 1587 the Catholic Church ran a formal office — the Promoter of the Faith, known everywhere by its nickname, the advocatus diaboli, the Devil’s Advocate. Whenever someone was proposed for sainthood, the Church did not simply gather the case in favour and vote. It assigned a man whose sworn duty was to argue against the candidate: to marshal every doubt, attack the claimed miracles, dig up every flaw in the life. Canonization was granted only if the case survived that determined, sanctioned, in-house opposition. There is a revealing coda. In 1983 the office was sharply curtailed — and the rate of canonizations promptly soared, hundreds where there had been a trickle. An accidental experiment had just run: remove the appointed critic, and far more candidates suddenly clear the bar.

What the Church had built, centuries before anyone named the problem, was a structural remedy for a predictable failure of groups. Left to themselves, groups converge. The psychologist Irving Janis called the pattern groupthink: a cohesive team talking itself into a consensus its own members privately distrust, because in a room marching toward a decision the person who objects becomes “the difficult one,” the obstacle, the wet blanket. The doubts are usually there; what is missing is a way to voice them that does not cost the doubter socially. The Devil’s Advocate is the fix, and the fix is structural rather than intellectual: do not hope a brave person speaks up — assign someone the job of arguing the other side. Once opposition is a sanctioned role, the objection that would have died as awkwardness gets spoken, on the record, as a duty.

That assignment carries a discipline with it, and the discipline is the whole point: you argue the opposing case in good faith and at full strength, separate from whether you believe it. This is the move people find hardest, because it runs against instinct. The instinct is to argue the side you hold and to argue against the other side only as hard as your conviction carries you. The Devil’s Advocate does the opposite on purpose — builds the best case for a conclusion they may personally reject, because the goal is not to express a view but to test one. A position is only worth trusting once someone whose job was to break it has tried, in earnest, and failed. An advocate who pulls punches because their heart is not in it has done nothing; the half-hearted objection is exactly the one the room can comfortably wave past.

Red-Team Advocate is that discipline aimed at a single artifact and a single audience. Concretely: suppose a company has published a plan to discontinue all combustion-engine research by 2028 and go fully electric by 2030, and you have to argue against it in front of the board. The undisciplined version leads with whatever you find objectionable. The disciplined version starts by modelling the board — what they weigh, what moves them, what idiom they think in — and then builds the strongest objections calibrated to that room: that the timeline bets the whole company on charging infrastructure and supply chains the firm does not control; that abandoning a profitable engine line forfeits the cash that funds the transition itself; that committing irreversibly now forecloses the option to adjust if the market moves slower than forecast. Each objection is grounded in what the plan actually says — not a caricature of it — and ranked by which lands hardest with the board, not by which you noticed first. And then comes the move that separates a brief that survives from one that gets ambushed: you name the plan’s strongest defence yourself — that early commitment is exactly what forces the organization to move — and you handle it head-on, so the board hears your answer to their best reply before they can raise it. That is the difference between ammunition and a complaint: a complaint expresses your displeasure; ammunition survives a prepared opponent who has read the same document you have.

Framework & implementation

This section uses Ora’s own terms for the parts of an analysis, so that if you open the actual mode file they line up. Each is glossed in plain language on first use.

Pipeline execution

Red-Team Advocate is an atomic mode in the artifact-evaluation-by-stance territory (called evaluation-by-stance on Main Street Independent) — a single adversarial pass over one named artifact, not a composite of sub-analyses. It is one of two modes split from a single legacy red-team mode: the sibling red-team-assessment takes the same hostile stance but ranks findings by severity for the user’s own fix-prioritisation, while this mode ranks attacks by persuasive force for an external audience the user must argue in front of.

It runs at Gear 4, Ora’s most thorough setting: a Depth analyst and a Breadth analyst work the artifact in parallel — one drilling into the artifact’s internal logic, hidden assumptions, and the steps that quietly assume away the hard part; the other scanning wide for the optical, strategic, and reputational attack vectors the audience would find compelling — and then critique each other (cross-adversarial evaluation, where each analyst’s reading is attacked by the other) before a consolidator integrates the result.

Before any attack, the pass runs the Input Sufficiency Protocol — a gate that checks the artifact is a specific named thing (not a domain), that its scope has clear edges, that there is enough detail for attacks to be specific rather than generic, and that the external audience is named or inferable. If the input is too thin, the mode emits a three-part redirect (What I see / What’s missing / Three options) instead of attacking on weak material; if the user overrides the gate, every attack is flagged as low-specificity so they know the limitation when arguing it.

The mode’s reasoning tools ride in its ANALYTICAL PERSPECTIVES block — the lenses it loads as it works. The required, load-bearing one is the cia-tradecraft-red-team lens: the institutional adversarial-actor-modelling discipline (Team A / Team B analysis, the Red Cell, the Key Assumptions Check) that both red-team modes draw from. Alongside it sits the devils-advocacy lens — the sanctioned-opposition protocol described above, the mode’s namesake — and a set of persuasion-aware lenses the audience-modelling depends on (walton-schemes-and-critical-questions for argument structure, plus affect-heuristic, social-proof, and narrative-instinct for how an audience actually responds). Underneath them all is the kahneman-tversky-bias-catalog as the foundational lens — the inventory of cognitive biases an attack can exploit and an analyst must avoid.

Output contract

The deliverable is an adversarial advocate brief — a fixed set of sections, so the case is auditable rather than a loose rant: a Stance declaration (the literal line Stance: advocate., which fixes the posture as build-the-case-against, not surface-vulnerabilities-for-fix); an Audience model (the named external audience with their frame, priorities, and persuasion pathways); an Artifact restatement (the artifact in brief, quoted where possible, so the attacks anchor in what it actually says); Attacks ranked by persuasive force, each carrying an Attack [N] label, a persuasive-force tier (Devastating / Strong / Plausible, applied verbatim and ranked worst-for-the-artifact first), a surface tag (Internal — a logic flaw — or External — empirical, optical, or strategic), grounding in the artifact’s content, and a “why this lands with [audience]” annotation; Suggested phrasing per attack in the audience’s own idiom rather than the analyst’s voice; Residual uncertainties (the facts or framings the brief depends on that could shift — the mode’s honest account of what it would take for an attack to be wrong); Concessions, each naming a counter-move the audience will recognise as the artifact’s strongest defence, with a pre-emptive handling; and Strategic considerations naming the political, reputational, or coalitional dimensions the audience cares about. A separate confidence marker rides each major attack — confidence is whether the attack is correct; persuasive force is whether it lands — and the two are kept distinct on purpose.

Origin and evidence

The role predates the theory by centuries. The Catholic Church’s advocatus diaboli institutionalised sanctioned opposition for canonization reviews from 1587 until the office was curtailed in 1983 — the inadvertent natural experiment in which canonizations rose steeply once the appointed critic was removed. The modern diagnosis the discipline answers is Irving Janis’s. In Victims of Groupthink (1972), and in the expanded second edition retitled Groupthink: Psychological Studies of Policy Decisions and Fiascoes (1982), Janis showed how cohesive groups suppress their own doubts to preserve harmony, reconstructing foreign-policy fiascos — the Bay of Pigs invasion among them — as failures of exactly that shape: capable people in a room, none willing to be the one who breaks ranks. The empirical case that the cure works is Charles Schwenk’s meta-analysis (1990), which pooled the experimental literature and found that structured devil’s advocacy measurably improves decision quality over groups that simply seek consensus. The throughline is institutional: organizations that build sanctioned, disciplined, good-faith opposition catch their own errors earlier; organizations that let agreement go unchallenged are surprised by adversaries who were not so polite.

Applications and common uses

  • Debate and hostile-review preparation. The native use: building the strongest case an opponent could make against your position, so you walk in already knowing their best lines and your answers to them.
  • Dissuasion briefs. Talking a colleague, a board, or a committee out of a course of action by arguing against it for the specific people who will decide — calibrated to what moves them, not to your own objections.
  • Pressure-testing a published strategy. Constructing the adversarial case against a competitor’s, a regulator’s, or your own organization’s public commitment, ranked by what would land hardest with the audience that matters.
  • Argument and proposal stress-testing for external use. Where the goal is not to fix your own draft but to construct the case someone else would use against it, so you can anticipate and pre-empt it.
  • Sharpening a position by confronting its best opposition. Adopting the opposing stance in good faith to surface the objections your own side has been comfortably ignoring — institutionalised dissent aimed at a single argument.

Failure modes and when not to use it

  • Cynical overreach. Weak attacks promoted to “devastating” to inflate the brief, or the artifact’s strongest defence quietly omitted to make the case look one-sided. Both backfire: a brief built on inflated claims fails in front of a prepared audience and spends the user’s credibility with it. The mode deflates persuasive force to the honest tier and keeps concessions first-class.
  • Straw-target attack. An attack aimed at a distorted version of the artifact rather than what it actually says. This is the failure that kills an advocate brief on first contact, because the audience can read the artifact themselves and watch the attack collapse. The mode grounds every attack in quoted content and reshapes or drops the ones that do not apply to the artifact as written.
  • Fabrication. An attack resting on a claim the artifact does not make, or on intentions it does not have — distinct from straw-targeting because it can survive even when the attacked passage is quoted verbatim. It is a top-priority failure regardless of how devastating the attack would be if it were true, because it dissolves the instant anyone checks.
  • Performing hostility. Manufacturing objections to seem adversarial — attacks that fail the “would a committed opponent actually use this in front of the named audience?” check. The mode drops them rather than padding the brief to hit a quota; its purpose is high-leverage ammunition, not an exhaustive list of complaints.

When not to reach for it. When you want to find how your own artifact fails under any pressure so you can fix it — a full adversarial stress-test with vulnerabilities ranked by severity and repairs attached — that is red-team-assessment, the sibling that ranks for your own fix-prioritisation rather than for an external audience. When you want the strongest version of the same argument built and defended rather than attacked — the charitable best case for a position before any critique — that is steelman-construction, the direct opposite stance. And when you want an even-handed weighing of strengths against weaknesses with no lean to either side, that is balanced-critique; Red-Team Advocate is deliberately one-sided by design, and forcing it toward balance dissolves it into a different operation. (One more boundary: if the real problem is the framework the artifact rests on rather than the artifact within it, the mode flags that drift and routes sideways to paradigm-suspension.)

  • Red-Team Assessment — the stance-counterpart in the same territory: the same hostile posture, but ranking vulnerabilities by severity for your own fix-prioritisation rather than attacks by persuasive force for an external audience. The safer default when it is ambiguous which you want.
  • Steelman Construction — the direct opposite: instead of the strongest case against, the strongest case for a position, reconstructed at its best and only then critiqued.
  • Balanced Critique — the neutral sibling for when you want strengths and weaknesses weighed at parallel depth with no lean, rather than a one-sided brief.
  • Devil’s Advocacy, CIA Tradecraft Red Team, and the Kahneman–Tversky Bias Catalog — the lenses this mode loads: the sanctioned-opposition protocol it is named for, the institutional adversarial-actor-modelling discipline at its core, and the inventory of biases an attack exploits and an analyst must avoid.