Amazon's bar raiser program asks one question of every senior hire: is this person better than the median of the team they're joining? The answer is meant to ensure that each hire raises the team's average, that talent compounds over time instead of diluting.
It is the right instinct and an incomplete theory. Bar raisers raise the hiring floor. They do not raise the floor on which decisions are made and ratified. A company can hire above its own ceiling and watch the new hires produce at the company's existing level. The hire moves the average. The environment bounds the output.
Why this happens, what it costs, and what to do about it is the subject here.
What “sophistication” means
The argument turns on a word, so the word needs a definition.
By sophistication I do not mean credentials or pedigree. I mean three things together: structural rigor, intellectual honesty, and domain mastery. A sophisticated decision is one whose reasoning is traceable, whose assumptions are named, whose tradeoffs are acknowledged, and whose author understands the actual domain well enough to know which simplifications break the conclusion and which do not.
A sophisticated environment is one that rewards this kind of decision and rejects its absence. The reward and the rejection both have to operate. An environment that rewards rigor but cannot detect its absence is theater. An environment that detects sloppiness but does not reward rigor produces conformity.
The floor of an institution is the level of sophistication required for a decision to be ratified there. Below the floor, decisions are rejected. Above the floor, decisions can pass. The floor is the institution's ratification bandwidth—its metabolic capacity for sophisticated reasoning. I will use “floor” as the rhetorical anchor throughout because it carries the right intuitive freight. Where the argument requires precision, the technical term is ratification bandwidth: the maximum complexity of a decision the institution can metabolize without distortion.
Part of this capacity is measurable. Cognitive ability—operationalized as IQ, GT scores, AFQT, or any of the validated psychometric instruments—correlates strongly with the ability to construct and evaluate sophisticated arguments. The correlation is not the whole story; sophistication includes domain training, intellectual discipline, and cultural permissions that pure cognitive measurement does not capture. But cognitive capacity is a real input, the research is robust across half a century of meta-analyses, and pretending otherwise concedes ground to people whose interest is in not having the conversation. The conversation is worth having.
Sophistication is also domain-specific. An institution can have high sophistication on one dimension and low sophistication on another. A board with decades of geopolitical and military judgment may have zero capacity to evaluate biomedical engineering claims. A technology company expert at consumer product design may be unable to evaluate enterprise sales motion. A bank's commodity trade-finance unit may know exactly which counterparties to blacklist while the same bank's asset-management division aggressively markets paper from those counterparties. The floor that matters for any particular decision is the floor on the domain that decision depends on. An institution's overall floor is not a scalar. It is a profile across the domains it operates in, and the decisions that fail are the decisions that depend on the dimension where the profile collapses.
The game
Economists know a game called the Keynesian beauty contest. Each player guesses a number from zero to one hundred, trying to guess two-thirds of the average. The math says the equilibrium is zero. If everyone reasons fully, everyone guesses zero. The math is right and almost no one plays that way.
In practice the winning guess in a population of college students is around twenty-two. In a population of game theorists, around eight. The winning number tracks the sophistication of the population, not the equilibrium of the math.
The dynamic is recursive. A player capable of reasoning to the equilibrium has to decide whether the rest of the population can too. If they cannot, the player must bound their reasoning to one step above where the population stops. A player capable of level-eight thinking, playing against a population that averages level-two, must play at level-three to win. Their additional capacity does not help. It hurts, if they use it.
This is the dynamic that runs in every institution. The proposal that requires level-six reasoning to evaluate fails the review by people who reason at level-three. The high-capacity proposer either bounds their reasoning to level-four to clear the room or persists at level-six and watches the proposal die.
How sophistication gets compressed
The game describes the proposer's adaptation. It does not describe the ratifier's mechanism—what actually happens, cognitively and procedurally, when a sophisticated analysis lands in front of someone who cannot fully evaluate it.
The ratifier knows in some sense that the proposal exceeds their evaluative capacity. They cannot say so. The institution requires that they make a decision, take responsibility for ratifying or rejecting, appear competent in doing so. So they resolve their uncertainty by anchoring on features they can evaluate: the proposer's confidence, the stylistic markers of the presentation, the surface coherence of the conclusions, the degree to which the proposal aligns with prior decisions, the political coalition supporting it, the analogies the proposer chose to use. They cannot evaluate the structural reasoning. They evaluate the surface presentation. The decision they make is a decision about those features, not about the substance.
This is Crawford-Sobel in operational form. The original game-theoretic result, from Crawford and Sobel's 1982 paper on strategic information transmission, shows that as the preferences of the proposer and the receiver diverge, communication between them becomes coarser. The same compression operates here, but the divergence is not in preferences—it is in evaluative capacity. The proposer can transmit structural reasoning. The ratifier can receive surface features. The information loss happens at the decision boundary because the channel cannot carry what the proposer is sending.
The proposer learns. Over enough cycles of having sophisticated proposals evaluated on surface features, the proposer notices that confident language and stylistic conformity predict ratification more reliably than rigor. The proposer adapts. They start to present at the surface level the ratifier can evaluate, because doing otherwise loses every time. The sophistication that did not survive transmission stops being transmitted. The institution converges on the bandwidth its ratifiers can carry, and high-bandwidth signal stops being attempted at all.
This is the mechanism. It is not malicious. It is not even fully conscious. It is the equilibrium that emerges when the proposer's signal exceeds the ratifier's channel capacity, generalizing across the institution until the institution loses the capacity to know what its high-capacity people would have said. The information is lost at the source because the proposers know it cannot reach the destination.
The compression is silent. The institution sees the decisions it ratified, not the decisions it could not metabolize. The high-capacity individuals look indistinguishable from medium-capacity individuals because both produce output at the bandwidth the institution can carry. The floor is invisible to the people inside it precisely because everything they see has already passed through the filter the floor imposes.
The selection regime artifact
A finding from organizational psychology gets repeated everywhere: cognitive ability predicts career success up to about IQ 120, and then stops mattering. The finding is empirically real and well-replicated across decades of meta-analytic work—Schmidt and Hunter, Gottfredson, the broader g-factor literature. The interpretation is wrong.
The interpretation says that above 120, other factors—social skill, drive, judgment—do the work. The implication is that companies should hire smart-enough people and select the rest on character.
The interpretation confuses a property of the measurement environment with a property of the underlying ability. The ceiling at 120 is a ceiling on the environment, not on what the brain can do. Move the environment and the ceiling moves with it.
You can see the artifact in other domains. Wealth stops predicting happiness around a particular threshold—in the consumption environment where the data was collected. Move to a higher-cost city and the threshold moves up. Move to a higher reference group and it moves up again. The number is not a fact about human satisfaction. It is a fact about the regime.
Height stops predicting basketball success above six-foot-eight—in the NBA, a population already filtered to six-foot-five. Within that pool the variance in success comes from skill, conditioning, injury history. Take the same measurement on the general population and height is back to being the dominant variable.
The same pattern. The ceiling is a property of the selection regime, not of the measured trait.
Apply this to organizations and you find that the IQ-120 ceiling reflects an institutional ceiling on what gets ratified. The 120 threshold is roughly the upper end of what college-educated middle managers can follow without effort—which is to say, roughly the level of the people doing the ratifying in most companies. Above that level, sophistication starts to be a liability. The proposal that requires careful structural reasoning to evaluate fails the political review because the reviewers reason in shortcuts and conclude the proposal is incoherent.
This is not a fact about cognition. It is a fact about the room.
Necessary, not sufficient
The floor is the necessary condition for an institution to produce sophisticated output. It is not the sufficient condition.
A company can have a high floor and still fail through incentive misalignment, capital structure problems, regulatory capture, market timing, executive corruption, or any number of other variables. A million-dollar house can have a leaky roof and a rat infestation. The high purchase price does not guarantee structural soundness on every dimension. It guarantees only that the dimensions the buyer chose to prioritize are addressed.
The argument is not that the floor is the only variable. It is that the floor is the variable that determines whether the other variables can be addressed intelligently. Incentive distortion in a low-floor environment cannot be diagnosed because the diagnosis would require sophisticated analysis the environment cannot ratify. Regulatory capture in a low-floor environment persists because the regulators cannot evaluate the analysis that would expose it. Capital structure problems in a low-floor environment compound because the structural problems are invisible to the people who would have to fix them.
The floor is the meta-variable. It does not solve the other problems. It determines whether the other problems can be solved.
This is why low-floor companies can have brilliant individuals, generous capital, strong markets, and still produce mediocre output. The brilliance does not propagate. The capital is misallocated. The market position erodes. The floor is the gate through which sophistication has to pass, and at a low floor, very little passes.
What the institution loses
When the institution hires above its ceiling and bounds the new hire's output to the existing level, the loss is invisible. It is the difference between the decisions that were made and the decisions that could have been made. The institution sees the products it shipped. It does not see the products it could have shipped if its floor were higher.
This is the asymmetric accounting that protects low floors from examination. Successes are attributed to culture and process. Failures are attributed to individuals. The floor that produced both never enters the analysis.
Consider Boeing. The 737 MAX shipped without telling pilots about the MCAS system that could push the plane's nose down based on a single faulty sensor reading. The engineers who designed the system knew about the single-point-of-failure. They flagged it. The decision to omit redundancy and to hide the system from pilots was made above their level—by managers operating in an environment shaped by FAA negotiation politics, schedule pressure, and decades of financialization that had reoriented the company from engineering toward shareholder return.
Two airplanes crashed. 346 people died. Boeing paid roughly $20 billion in costs and lost decades of brand value.
Boeing's failure was multi-causal. The engineering compromise, the regulatory capture, the schedule pressure, the executive incentives that prioritized financial metrics over safety—all of it contributed. The framework here does not claim the floor was the only cause. It claims the floor was the mediating variable. Each of those pressures could have been resisted by a sufficiently sophisticated decision environment. The floor at Boeing was low enough that the pressures determined the outcome instead of being constrained by the analysis.
The investigations afterward attributed the failure to specific individuals and to cultural problems. The cultural problem had a name. The name was floor.
The same pattern operates in domain-specific ways. Theranos is often described as fraud rather than as a floor failure, and the description is half right. The board was sophisticated in some dimensions—Henry Kissinger, George Shultz, James Mattis brought decades of geopolitical and military judgment to the room—and absent in the dimension that mattered. None of them could evaluate a blood-diagnostics technology claim. The floor on biomedical engineering was approximately zero. The board ratified decisions because the dimension where ratification mattered had no functional floor at all.
This is the refinement the framework requires. The floor is domain-specific. An institution can have a high floor in one area and a low floor in another, and the decisions that depend on the low-floor domain will be ratified at the low-floor level regardless of how sophisticated the institution looks on its high-floor dimensions. Theranos's sophistication on personality, brand, and political relationships could not compensate for its absence of sophistication on the technical dimension where the product had to work.
Wells Fargo's cross-selling targets required fraud to meet, and the employees who knew could not get the analysis ratified. Volkswagen's emissions defeat device was an engineering decision made in a regulatory environment whose floor permitted the obfuscation. In each case the high-capacity individuals had the analysis. The institutional floor on the relevant domain prevented the analysis from determining the outcome.
Cases that do not get used enough
The conventional case set—Boeing, Theranos, Wells Fargo, Volkswagen—has been worked so thoroughly across organizational literature that the examples can begin to do the analysis instead of being analyzed by it. The framework predicts that the dynamic operates everywhere, which means the worked examples should be available anywhere. Two more recent cases extend the pattern.
Credit Suisse and Greensill Capital. Beginning in 2017, Credit Suisse's asset-management division began marketing supply-chain finance funds backed by Greensill-originated paper. Internal risk teams flagged concerns from the start. Whistleblower emails in 2018 questioned the decision to entrust client money to Greensill. The bank's own commodity trade-finance unit had blacklisted one of Greensill's principal counterparties, Liberty Commodities, in 2016. The warnings continued for years. Senior executives in Zurich, London, and Singapore dismissed them and expanded exposure. By March 2021, when Greensill collapsed, Credit Suisse had $10 billion in funds tied to the paper. The bank recovered roughly $5.9 billion. The Swiss regulator FINMA later found that Credit Suisse had “seriously breached” risk-management obligations.
The case is a clean test of the framework. The high-capacity individuals—the commodity trade-finance unit that knew Liberty Commodities was uninvestable, the internal risk teams that flagged Greensill repeatedly, the whistleblower who put it in writing—produced accurate analysis. The leadership operated at a floor that could not ratify the analysis. The product of the floor was approximately $4 billion in losses and the eventual collapse of a 167-year-old bank.
Facebook and the Civic Integrity team. Frances Haugen joined Facebook in 2019 as a product manager on the Civic Integrity team—the unit responsible for studying and mitigating the platform's role in misinformation, polarization, and political violence. Her team's research, and the research of related integrity teams, documented the algorithmic mechanisms by which engagement-maximization amplified divisive content, harmed teenage mental health, and degraded the information environment. The research was rigorous. It was internal, expensive, and conducted by people the company had hired specifically to do this work.
The research did not change the products it studied. After the 2020 election, leadership dissolved Civic Integrity. Two months later, the January 6 insurrection happened. Haugen left in 2021 and brought the internal documents with her. The documents showed that Facebook's own researchers had repeatedly identified harms the leadership chose not to address.
The case is structurally identical to Credit Suisse. The high-capacity individuals had the analysis. The institutional floor could not metabolize the conclusions. The product of the floor was a platform whose harms the company's own research documented and whose research the company's own leadership ignored. The institution did not lack the capacity to know. It lacked the capacity to ratify.
What the bar raisers miss
Amazon's bar raiser solves the hiring problem and leaves the operating problem untouched. You can hire a software engineer capable of designing a distributed system that scales to a billion users and assign them to ship a feature whose architecture was decided in a Slack thread by people who could not read the engineer's proposal.
The hire is above the bar. The decisions are not. The engineer ships what the decisions allow, which is what the floor permits, which is roughly what the previous engineer at the lower bar would have shipped. The hire's capacity is paid for in salary and absorbed by the institution as cost. Nothing of the capacity reaches the product.
Over time the high-capacity engineer notices. The trajectory has two endings.
The first is exit. The engineer leaves for an environment whose floor allows their capacity to operate—a startup they found, a research lab, a competing company with a higher bar. The exit interview cites culture fit. The institution concludes the engineer was not a team player. The institution does not raise its floor.
The second is atrophy. The engineer stays. Over a decade of bounding their reasoning to what the floor permits, the deployment infrastructure for their capacity decays. The capacity is still there in some sense. The muscle memory of using it is gone. By the time the engineer would have authority to operate at a higher level, they cannot. They have become an artifact of the floor that bound them.
Either ending is bad for the engineer. Both are worse for the institution. The first removes the capacity from the system entirely. The second converts the capacity into political competence at the floor level—which is to say, the engineer becomes one more defender of the floor that wasted them.
There is a third path the framework should acknowledge. Many high-capacity individuals do neither exit nor atrophy. They stay, partially atrophy on the institutional work, and displace their unused capacity into side projects, writing, research, open-source contributions, communities of practice outside the institution. Linus Torvalds did this for years at Transmeta while building Linux. The engineers who built the early internet protocols did most of the work outside the institutions that employed them. The pattern is old.
The institution still wastes the capacity in its own domain unless it explicitly supports the external deployment. Tolerant employers—Transmeta with Torvalds, IBM with Knuth, most modern technology companies with their engineers' side projects—capture 60–80% of a person who would otherwise have exited entirely. Demanding employers, who insist on 100% mental energy from people who cannot deliver 100% to any single job, capture 0% because the person leaves. The arithmetic favors tolerance. Most employers do not perform the arithmetic.
The framework's central claim is that the leader sets the floor—that what gets ratified, and therefore what the institution can produce, follows from who is doing the ratifying. The claim is falsifiable. Two cases test it in opposite directions.
Satya Nadella inherited Microsoft from Steve Ballmer in 2014. The Microsoft Nadella inherited had a floor compressed by a Windows-everywhere worldview that punished cross-platform thinking, a stack-rank performance system that punished collaboration, and a strategic position that had missed mobile. The talent was still there. Microsoft had brilliant engineers, brilliant researchers, brilliant product people. The floor would not let them.
Nadella changed what Microsoft would ratify. He killed stack ranking. He pivoted to cloud-first with Azure, which required ratifying complex multi-year infrastructure investments at the level of engineering sophistication those investments demanded. He embraced open source, reversing decades of institutional posture that had treated Linux as the enemy. He acquired GitHub and let it operate as it had been operating. The market capitalization roughly tripled over the next several years. The high-capacity people who had been quietly displacing their capacity into side projects started bringing it back inside.
Post-merger Boeing is the inverse case. The 1997 acquisition of McDonnell Douglas brought in leadership whose floor was set by financial metrics rather than aerospace engineering. Phil Condit moved the headquarters from Seattle to Chicago in 2001, partly to put physical distance between executives and the engineers in Everett. Successive CEOs prioritized shareholder return, share buybacks, and outsourcing. The engineering floor that had built the 747—the design culture that had made Boeing the company it was—eroded under leadership that did not ratify decisions at engineering levels.
The 737 MAX is downstream of this floor reset. The single-point-of-failure in MCAS, the omission from pilot training, the regulatory negotiations to avoid the simulator requirement—none of these would have survived the floor that existed before the merger. They survived the floor that existed after.
Both cases support the framework in the right direction. The leader who can operate at a higher floor than the institution inherited raises the floor and the institution compounds. The leader whose floor is lower than the institution's existing engineering or operational standard lowers the floor and the institution decays. The mechanism is the same in both directions: the leader is the person whose ratifications define what the institution will produce.
This means the answer to “what makes a great CEO” is not a list of personal virtues. It is a question about ratification capacity. Can this person evaluate the substance of the decisions that come to them, or will they evaluate the surface features? Can they tell the difference between a rigorous proposal and a confident one? Can they reward sophistication when sophistication is what the moment requires, or will they reward conformity because conformity is what they can recognize? The answer determines the floor. The floor determines everything else.
What to do
Raising the floor is not a matter of hiring better people. The bar raiser already does that. Raising the floor is a matter of designing the environment in which decisions are made and ratified so that sophisticated decisions survive contact with the institution.
The mechanisms are not mysterious. They have been built before.
Bell Labs ran a culture where mathematicians and physicists could publish, present, and challenge each other at a sophistication level the rest of AT&T could not match. The floor was set by the people in the room. The room was filtered to maintain the floor. The Labs produced the transistor, the laser, information theory, the C programming language, Unix. AT&T proper produced telephone service. Same parent company. Different floor. Different outputs.
Jeff Bezos required six-page narrative memos at executive meetings, read silently for an hour before discussion. The mechanism is floor-raising. A six-page memo that survives an hour of silent reading by senior leaders cannot be sloppy. The reviewers tear apart anything inadequate. The proposer either writes at the level the room can absorb or fails. Over time proposers calibrate up. The floor of what gets ratified rises because the format of ratification demands it.
Steve Jobs ran product reviews where bad ideas were called bad ideas in front of their proposers. The mechanism was the same. The floor was visible, defended, enforced. Proposers calibrated up because the alternative was being humiliated. The humiliation was a feature of the floor, not a cultural flaw.
Berkshire Hathaway operates on Buffett's if-you-can't-explain-it-on-one-page floor. The floor filters complexity that hides imprecision. Proposals that require obfuscation to sound competent do not survive the format. The institution compounds capital at a rate that suggests the floor is paying for itself many times over.
Pixar's Brain Trust is the same mechanism in a creative domain. Brad Bird, Pete Docter, Andrew Stanton, John Lasseter—the directors of the studio's films—gather to review each other's projects in development. They are required to identify what is broken, specifically and technically, without political softening. The format demands that the criticism be substantive enough to be acted on. A director whose film is in the Brain Trust knows they will get rigorous feedback they cannot have hired any consultant to give them, because the feedback comes from people who have made the same kind of decision and made it well. The Brain Trust is what allowed Pixar to ship a string of films in the 2000s that nobody else was capable of producing. The floor was the format, the format was the room, and the room produced the output.
In each case the floor-raising mechanism has the same shape. It is a decision-ratification environment whose format requires sophistication to navigate. It cannot be gamed by political competence at a lower level because the format rewards substantive sophistication and exposes its absence.
High floors have failure modes. They can become brittle. They can become elitist in ways that filter out useful variance. They can ossify around the paradigms of their founders and miss the next thing because the next thing is not legible in the old framework. The high floors that compound advantage are the ones whose format is rigorous about reasoning rather than rigorous about adherence to a fixed worldview. The distinction matters. A floor that demands sophisticated reasoning about whatever is in front of the room is generative. A floor that demands sophistication only within a particular paradigm becomes the next institution's decay story.
The institution that wants its high-capacity hires to operate at capacity has to engineer the environment. It cannot be done by exhortation. The leaders themselves have to operate at the floor they are setting. If the leaders cannot read a six-page memo critically, the memo format produces theater. If the leaders cannot evaluate a complex architectural proposal, the architecture review produces consensus instead of selection.
Floor-raising starts at the top. The leaders who can recognize sophistication when they see it are the ones who can establish the floor. The leaders who cannot will defend the existing floor against any attempt to raise it, because raising it would expose them.
Local floors
Most readers of this essay are not in a position to set the floor for a Fortune 500 company. They are running a team, a division, a subunit. The question is whether the framework applies at smaller scale.
It does. Local floors operate within institutional floors. A VP of Engineering can build a sub-organization whose ratification environment is more sophisticated than the broader company's. The mechanism is the same—codified principles, demanding review formats, rewards for substantive sophistication. The constraint is that the local floor cannot exceed the broader institutional floor for decisions that cross the boundary. Architecture decisions can be made at the local floor if they do not require approval from outside. Resource decisions, headcount, prioritization across teams—these have to clear the institutional floor, and the local floor cannot override.
The local floor is still worth building. It preserves the capacity of the people inside it. It produces better output within its domain. It demonstrates the alternative to the broader institution, which sometimes propagates upward when the leader of the local floor becomes a leader of the institution. Most floor-raising at the institutional level starts as floor-raising at the local level, by leaders who built sophisticated sub-organizations and were eventually given broader authority.
The strategic advice for the high-capacity individual stuck in a mid-floor institution is therefore not always to exit. It is to find or build a local floor whose ratification environment matches their capacity, and to operate inside it while signaling enough compliance with the broader institution to avoid attracting political attention. This is hard and imperfect, but it is achievable. The alternative is the exit or the atrophy.
The diagnostic
The institution that wants to know where its floor is can find out by looking at three things.
Look at who leaves. The institutions that decay are losing their high-capacity people to higher-floor environments. Each exit is diagnostic. Treating each exit as an individual problem misses the pattern. If exits skew toward high-capacity individuals—the ones who could have moved the institution most—the floor is the cause, regardless of what the exit interviews say.
Look at what gets ratified, and look at the proposal flow that produces it. The diagnostic has four cases. If all proposals are good and good ones are approved, the team is high-capacity and the floor has not yet been tested—you can't tell from outputs whether the floor is high or simply unchallenged. If proposals are mixed and good ones are chosen, the floor is doing its job; the institution is functioning. If proposals are mixed and bad ones are chosen, the floor is low, the approvers can't distinguish, and your high-capacity people are about to exit or atrophy. If all proposals are bad, the floor is low and the team has already compressed to match it; the capacity may still exist in individuals but is no longer being deployed even at the proposal stage. The fourth case is the worst because it shows no asymmetry to read. By the time the team has stopped trying, the diagnostic from outputs alone returns nothing.
These three diagnostics are available to any leader willing to perform them honestly. Most leaders do not perform them honestly because the diagnosis implicates the leader. The floor is set by whoever ratifies decisions, which is the leader. To name the floor is to name the leader's own ceiling. Few leaders volunteer for that.
The prescription
Build decision-ratification environments at the sophistication level of the talent you mean to retain.
Make the floor visible. Codify it. The unwritten floor is the dangerous one because it shifts with politics and personalities. The written floor is defensible because new entrants can see what they are calibrating to.
Reward decisions made above the floor, not decisions that comply with the floor. Compliance is what the floor already produces. Sophistication is what raising the floor requires. If the institution's reward signals go to compliance, the floor stays where it is.
Audit the exits. Talk to the people who leave. Not in formal exit interviews, which produce diplomatic answers. Six months after they have left, when they have nothing to lose by being honest. Ask them where they could not operate. The answer is the floor.
Hire leaders who can operate at a higher floor than the institution currently does. They will raise the floor by example, by what they ratify, by what they refuse to ratify. The wrong leaders will lower the floor by ratifying mediocrity, by punishing sophistication, by selecting peers who will not threaten their own ceiling.
The institution that does these things will compound. The institution that does not will decay. The decay will be invisible until it is structural. By the time it is structural, the institution will have lost the capacity to recover.
What the bar raisers should have been
Amazon's bar raiser program is what an institution does when it has half the theory. It recognizes that hires shape the institution and tries to ensure good hires. It does not recognize that the institution shapes the hires. Without the second half, the first half decays into ritual. The bar raiser approves a hire who joins an environment that bounds their capacity, and the bar raiser's work was for nothing.
The complete program raises the floor on hiring and the floor on ratifying decisions. The two together produce the compounding that turns ordinary talent into extraordinary output. Either alone produces theater.
The high-capacity individuals in the institution are not a problem to be managed. They are the diagnostic, the asset, and the lever. Treat them as the problem and they become it—they exit or they decay. Treat them as the lever and they raise the floor for everyone, which raises the output for the institution, which compounds advantage every year.
The choice is between an institution that uses what it hires and an institution that wastes what it hires. There are no third options. The floor decides which one you are.