THE EPHOR'S MIND: SPARTA HAD TWO KINGS, AND BOTH COULD BE ARRESTED

Sparta had two kings at all times, from two separate royal houses, and both of them could be arrested. Not by each other — by five annually elected men called ephors, whose entire function was to watch the kings and stop them when they overreached.

The Institution Most Retellings Skip

Two ephors accompanied every Spartan king on campaign specifically to observe him in the field. If a king made a rash decision, took a bribe, or exceeded his authority, the ephors traveling with him reported back to the three who had stayed behind. When the army returned, the ephors could summon a king to explain his conduct — and if the explanation did not hold up, they could fine him, put him on trial, or in the most serious cases depose him entirely.

Aristotle described the resulting arrangement as a genuine balance between oligarchy and democracy. Not because Sparta trusted its kings less than other Greek cities trusted theirs, but because Sparta built structural accountability into the position itself, regardless of who held it or how capable he was.

This is the piece of Spartan government that most modern retellings leave out, because it complicates the version of Sparta that reads best on a poster: the unbreakable warrior-king, answerable to no one, ruling through pure strength of will. The actual system was closer to the opposite. Even the two most powerful men in Sparta, from houses tracing their lineage back to legendary founders, operated under continuous, structural, non-negotiable oversight.

And it was not held in reserve as a formality. Herodotus and later sources record the ephors actually exercising it. King Cleomenes I faced accusations serious enough to threaten his position. The regent Pausanias — who commanded the decisive victory at Plataea the year after Thermopylae — was later investigated by the ephors over conduct they judged had crossed a line, and by the ancient accounts ended up sealed inside a temple where he had taken sanctuary, left to starve rather than seized directly. The man who had delivered Sparta's greatest battlefield success in a generation was not exempt from the check the moment his conduct gave the ephors reason to apply it.

Why the Chapter Isn't Really About Kings

The reason The Spartan Protocol opens its tenth chapter with the ephors is that the lesson underneath the institution lands on a problem the previous nine chapters do not address: what happens to a man's judgment once he has built real capability and nobody is checking it anymore.

Every chapter up to that point is about building something — strength, discipline, tolerance for hardship, protected sleep. A man who has actually done that work for a year, or five, is operating at a level most of the people around him are not. The chapter's argument is that this creates a specific and underdiscussed risk: competence without accountability does not stay competence indefinitely. A man who is right often enough starts to assume he is right by default, stops seeking correction, and begins treating his own judgment the way Sparta explicitly refused to treat even a king's.

The ephors, in that reading, were not a punishment for Spartan kings being untrustworthy. They were a structural acknowledgment that building the check in from the start costs less than discovering the failure after the fact.

What the Accountability Research Actually Measured

There is a real research literature underneath this, and it is worth reporting at the strength it has rather than the strength the idea seems to deserve.

The standard reference is Lerner and Tetlock's 1999 review in Psychological Bulletin, which synthesized the accountability research across a wide range of social judgments and choices. Their framing is more careful than the popular version of it. They asked explicitly under what conditions accountability attenuates cognitive bias, has no effect on it, or amplifies it — and treated accountability as something that could go any of those three directions depending on how it was structured. Their own summary describes accountability not as a uniform improvement but as a process of internalized dialogue whose effects depend on social and institutional boundary conditions.

Accountability is not a unitary phenomenon. It can be implemented in at least two ways, and the two do not produce the same result.

Process Versus Outcome — The Distinction That Moved the Numbers

The two ways are usually labeled process accountability (you will have to explain your reasoning) and outcome accountability (you will be judged on how it turned out). A 2022 meta-analysis in Frontiers in Psychology pooled 7 studies covering 9 experiments and 1,080 participants, published between 2002 and 2017, comparing the two.

Pooled across all tasks, the authors found no significant difference between them — a standardized mean difference of −0.09 (95% CI −0.21 to 0.03). The interesting result was the moderator. In the three low-complexity experiments, process accountability was substantially better (SMD 0.96, 95% CI 0.72 to 1.20). In the six high-complexity experiments, the direction reversed and outcome accountability came out ahead (SMD −0.48, 95% CI −0.62 to −0.33).

The authors state their own limits plainly: a modest number of studies of moderate methodological quality, very high heterogeneity between experiments (I² = 96%, p < 0.001), all laboratory-based, and participants drawn primarily from U.S. undergraduate populations with no healthcare professionals in any of the included studies.

That pattern is consistent with what De Langhe, Van Osselaer and Wierenga reported in Organizational Behavior and Human Decision Processes in 2011: process accountability improved judgment quality on relatively simple tasks where outcomes were an additive combination of cues, and that advantage did not carry over to more complex tasks where outcomes depended on interactions between cues.

The Checklist Evidence, Including Where It Went the Other Way

The same split shows up in the most-cited real-world version of structured pre-decision review — the surgical safety checklist.

Haynes and colleagues reported in the New England Journal of Medicine in 2009 on the WHO Safe Surgery Saves Lives program across eight hospitals in eight cities. Comparing 3,733 consecutively enrolled patients before a 19-item checklist was introduced with 3,955 after, the death rate went from 1.5% to 0.8% (P = 0.003) and inpatient complications from 11.0% to 7.0%. That is the number that drove worldwide adoption.

Urbach and colleagues published a different result in the same journal in 2014. Ontario's policy encouraging universal checklist adoption created a natural experiment across 101 hospitals and roughly 109,000 procedures before adoption against 106,000 after. Adjusted risk of death within 30 days was 0.71% before and 0.65% after (odds ratio 0.91, 95% CI 0.80 to 1.03, P = 0.13). Adjusted surgical complications were 3.86% before and 3.82% after (odds ratio 0.97, 95% CI 0.90 to 1.03, P = 0.29). The authors' conclusion: implementation was not associated with significant reductions in operative mortality or complications.

Two studies, same journal, same intervention, different designs and different settings, pointing different directions. Neither one is the whole answer, and the honest version of this section is that the mechanism is better supported than any specific effect size for it.

What This Isn't

The chapter heads off a misreading before its protocol section, and it is worth repeating here: this is not an argument for constant self-doubt, or for deferring every decision to someone else.

Fast, trained, recognition-primed judgment — the kind Chapter 2 of the book builds under the heading of the laconic mind — remains the tool for the overwhelming majority of moments. A man standing in a phalanx line, literal or figurative, does not have time to convene a review board before deciding whether to hold position.

The ephor system operated on a different timescale entirely. It was not a check on split-second judgment. It was a check on the slower, cumulative pattern of conduct over months and years — the drift that is nearly invisible day to day and only obvious in retrospect. The chapter describes what that drift usually looks like, and it is rarely a single dramatic decision: a gradual narrowing of who a man listens to, a growing comfort with being the smartest person in every room, and an increasing tendency to treat disagreement as a problem to manage rather than information to use.

The Protocol

The Ephor's Mind card, as it appears in the book:

• Choose one person given actual standing to challenge you — not a cheerleader, someone who will say your reasoning doesn't hold up
• Build a fixed review point, scheduled in advance: a monthly check-in where you explain your reasoning out loud, not just report outcomes
• Write your reasoning down before the outcome is known — what you expect to happen, why, and what would prove you wrong
• A check nobody is empowered to actually push back on isn't a check, it's decoration

It is worth noticing that the second and third items are process accountability in the technical sense — explain the reasoning, not just the result — which is the version the low-complexity experiments favored and the version the ephors themselves practiced. They summoned a king to explain his conduct. They did not simply count his victories.

Battle Hard. — Will Power