189 | Brian Klaas on Power and the Temptation of Corruption
What this covers
Brian Klaas examines whether power itself corrupts people or whether corruptible people simply seek power in the first place. Hosted by Sean Carroll, the conversation draws on historical examples, behavioral research, and interviews with hundreds of powerful figures to argue that both mechanisms operate — self-selection and the corrupting effects of power itself are real — but the more actionable insight is that systems, not human nature, are the primary lever. The discussion ranges across political science, psychology, and biology to show how institutional design shapes who seeks power and how those people behave once they have it.
The episode moves through several domains. It opens with the famous parking-ticket natural experiment involving UN diplomats, where enforcement (impounding cars) changed behavior faster than cultural attitudes, illustrating Accountability and Enforcement. It explores how psychopathy and the Dark Triad traits overrepresent in leadership, examines Transparency as Accountability through bee colonies and baboon hierarchies, and traces evolutionary pressures that may have left us evolutionary mismatch between ancestral leadership-selection instincts and modern institutional needs. The conversation then pivots to tractable reforms: how police recruitment messaging (comparing militarized versus empathy-focused campaigns) shapes applicant pools, how Sortition might work as shadow governance to expose gaps between actual and ideal decision-making, and how perceived surveillance—not constant actual surveillance—deters corruption. Klaas argues against fatalism about the corrupting nature of power, arguing instead that institutions designed on the assumption that bad actors will scheme to gain power can substantially reduce abuse without requiring us to change human nature.
Klaas argues that corruption in power is best understood as the product of self-selection (corruptible people seek power) interacting with systems we design, rather than an inevitable consequence of power itself — meaning the most tractable lever for reducing abuse is reforming systems, not changing human nature.
- Both effects are real: corruptible people gravitate to power AND power biologically and psychologically changes people
- Systems mediate corruption strongly — rotten systems attract rotten people, good systems attract good people (dice-roll and parking-ticket studies)
- Simple systemic interventions (recruitment design, surveillance of the powerful, sortition oversight) can blunt corruption without solving human nature
Power systems should be architecturally designed to deter bad actors, constrain their harm, and enable their removal rather than relying on the good character of office-holders.
- Organizations should design power systems on the assumption that the worst person is scheming to get into power at every level — building systems that deter bad actors, constrain them once in power, and remove them when they overstay — not from distrust of fellow citizens but because such people exist and the system functions better when built to counteract them.
“imagine that their system has to exist to prevent the worst person who is trying to scheme to get into a position of power at every level, so they design a system that deters the bad people, that constrains them when they get power and finds a way to boot them out when they actually stay for too long”
- Some small hunter-gatherer societies maintained egalitarianism through reverse dominance hierarchies — deliberate systemic mechanisms to tear down anyone seeking too much power — exemplified by the Kung practice of communally owning and rotating arrowheads so credit for a kill goes to the arrowhead owner rather than the hunter, plus rituals of 'insulting the meat' to keep successful hunters humble.
“we're gonna create a communal ownership of the arrow heads that are used in hunting, and we're gonna rotate them regularly, and the credit for the kill doesn't go to the person who actually killed the animal, it goes to the person who owned the arrowhead”
- Rather than using sortition to govern directly (which fails because governing requires specialist knowledge), we should create randomly-selected shadow bodies — a shadow House of Representatives or shadow corporate board — with the same access to experts and information, producing non-binding decisions that expose the divergence between how actual power-holders behave (distorted by partisanship, lobbying, re-election, profit-motive) and how randomly-selected citizens would behave.
“create effectively oversight bodies that provide accountability, so my proposal, both for businesses and for politics, is to create a shadow House of Representatives or a shadow board in a company that is totally randomly selected”
- Across bee and wasp species, hive structural transparency reduces anti-hive 'corruption': in species where larval cells are visible and queen cells are conspicuously larger, 'police bees' can easily detect and behead illicit excess queens, producing lower levels of shirking and free-riding, whereas opaque hive structures make policing ineffectual — illustrating how structure governs levels of corruption-analogous behavior.
“the transparent versus the non-transparent cells have very different levels of shirking or free riding in this instance, because the police bees or police wasps in this case, are ineffectual, they can't tell very well”
- Our surveillance systems are inverted: the least powerful are the most surveilled (e.g., work-from-home chair sensors, lunch-break monitoring) while the most powerful operate in deliberately surveillance-free corner offices where the truly damaging scandals occur (Enron was not brought down by long lunch breaks) — power should be a burden carrying scrutiny, so the positions most prone to abuse should be the most surveilled, with a Goldilocks balance avoiding pervasive distrust.
“the most surveilled are the least powerful. Because you think about all the corporate scandals, the political scandals, the things that actually have created profound and disastrous consequences, they're happening at the top”
Corruptible and power-seeking individuals disproportionately put themselves forward for leadership through self-selection, making those who attain power unrepresentative of the public.
- Because power is acquired through self-selection — people who covet power put themselves forward via elections and promotions — those who attain power are by definition unrepresentative of the general public, disproportionately finding power to be an end in itself, just as a basketball tryout pool would be unusually tall rather than average height.
“if you went to a high school basketball tryout, it would be really weird if the height among the people who are trying out was average. This is exactly the same with power, you don't have a self-selection mechanism that's randomized, so people who covet power put themselves forward.”
- The dark triad — narcissism, Machiavellianism, and psychopathy — when dialed up together, makes people disproportionately likely to seek and obtain power, partly because modern power acquisition is unavoidably performative (interviews, campaigns are performances) and narcissists are attuned to others' perceptions while psychopaths display superficial charm and chameleon-like adaptability; observational research finds between four and 100 times more psychopaths in leadership than in the general public.
“this concept of the dark triad, which is narcissism, Machiavellianism, and psychopathy... if they come together in a chemical cocktail known as the dark triad in high levels, those people are disproportionately likely to seek power and disproportionately likely to get power”
- Twin-study research claiming to find a 'leadership gene' overclaims its findings because it measures power obtaining, not power seeking — and any correlation may reflect co-varying genes for affability, attractiveness, or outgoingness rather than power-seeking itself, so the evidence that power-seeking is genetic is quite weak.
“it doesn't actually cover power seeking, it covers power obtaining, and those things are not the same”
- Instead of accepting self-selection, where leaders are those who put themselves forward, political parties and organizations should proactively recruit and head-hunt people with track records of integrity and genuine service — and make power attractive to them — because even a 10% increase in this drafting mentality would meaningfully improve the quality of people in power.
“why do we accept self-selection. Why do we say that the people who are in charge of us are the people who put themselves forward... why don't we proactively recruit more?”
Systems and enforcement structures shape individual behavior more strongly than moral character, as shown by diplomatic immunity, recruitment messaging, and civil-service selection.
- Police recruitment messaging shapes the personality of applicants: militarized 'punisher'-style videos (e.g., Doraville, Georgia's SWAT-tank, heavy-metal ad) attract aggressive applicants, whereas New Zealand's empathy-focused 'Do you care enough to be a cop?' campaign radically changed applicant numbers, personality screening results, diversity, and improved relations with the Māori community — showing reform of who the police are is low-hanging fruit ignored in debates about what police do.
“almost everything in the police discussion is about what the police do, and very, very little is about who the police are”
- The natural experiment of UN diplomats' parking tickets — where pre-2002 diplomatic immunity meant no enforcement — showed diplomats from corrupt countries parked illegally far more than those from honest countries (culture matters), but once Mayor Bloomberg began impounding cars, even Egyptians 'parked like the Norwegians' almost overnight (enforcement matters), while the Norwegians' behavior had actually worsened over time the longer they could get away with it.
“in the pre-enforcement period, all of the sort of non-corrupt countries diplomats parked pretty much normally legally, and all of the corrupt countries diplomats parked illegally a lot”
- Systems mediate corruption so strongly that rotten systems attract rotten people and good systems attract good people — demonstrated by a dice-roll study where dishonest students in corrupt-civil-service India aspired to be civil servants while in low-corruption Denmark the honest students wanted civil-service jobs and dishonest ones went into business.
“rotten systems attract rotten people and good systems attract good people”
- NYPD integrity sting operations deter corruption through perceived possibility rather than frequency: only 500 real integrity tests were conducted, but 12,000 officers believed they had been tested, so 11,500 officers behaved better at genuine crime scenes they suspected might be fake — suggesting we should set up similar undercover stings for politicians (e.g., fake companies testing contract-award corruption) so power-holders constantly fear getting caught.
“they did 500 integrity tests... 12,000 police officers said, "Yes, I have been subject to an integrity test," so 11,500 had real crime scenes that they thought were fake and behave better as a result of it”
Psychopaths and dark-triad individuals are overrepresented in leadership because performative power acquisition suits their traits, yet they respond differently to reform tools than ordinary people.
- In ultimatum-game experiments, normal people reach a breaking point around 70-30 and reject unfair offers out of a sense of justice, but psychopaths accept almost any offer because their response is anger without rejection — measured via skin conductance and FMRI — suggesting their willingness to act in pure self-interest helps them rise the corporate ladder where others refuse unfair deals on principle.
“most normal people hit their breaking point about 70-30... the psychopaths would basically accept almost anything... they didn't hit a breaking point. They just say, "I'll take the $5 because it's better for me."”
- Fixing bad outcomes from people in power requires accurately diagnosing the cause — sometimes it is the system, sometimes the individual, sometimes brain chemistry — because applying the wrong tool worsens the problem; for example, shame or appeals to responsibility may reform ordinary people but are totally ineffectual on psychopaths, so we must not just say 'power corrupts' and move on.
“accurate diagnostics of understanding why a bad outcome is occurring from someone in power is crucial to fixing it. Sometimes it's the system, sometimes is the individual, sometimes their chemistry is out of whack”
- FMRI research shows psychopaths have empathy effectively switched off by default — showing muted brain responses to disturbing images — but can deliberately dial empathy up when instructed to, producing scans resembling normal individuals; this distinguishes functional psychopaths (who modulate their traits to pass as normal in boardrooms and office) from dysfunctional psychopaths (who cannot turn it off and become serial killers).
“The brains of normal people are lighting up with anguish and psychopaths have a very muted response. But then... she told the psychopaths, try to imagine what it would be like to feel the pain of these people... and all of a sudden the scans looked remarkably similar to the normal individuals.”
- Although politicians won't submit to mandatory psychopath or knowledge tests, voluntary disclosure could create competitive pressure: candidates confident they are not psychopaths would start taking psychometric tests publicly, prompting voters to wonder why others refuse — and outlets could offer optional basic-knowledge tests (publishing whether candidates agreed) — addressing the oddity that there are essentially no competence requirements to be president while a cathedral tour guide undergoes more rigorous training.
“the ones who are quite confident that they're not psychopaths would start to voluntarily do it and it would put pressure on the other ones, because the voters would then have a short kind of like, "Why aren't they doing the psychometric tests?"”
Power itself produces measurable psychological and biological changes that alter decision-making, moral judgment, and risk perception in those who hold it.
- Power produces measurable psychological changes — people in power view subordinates as more disposable abstractions and develop 'illusory control,' believing they can manipulate outcomes more than they actually can — and these are accompanied by biological changes, as shown by altered dopamine receptors in macaques whose dominance rank shifts.
“they start to think of them as more disposable, they have this mechanism called 'Illusory Control' where they believe that they can manipulate outcomes more than they actually can”
- Both mechanisms operate together: corruptible people gravitate toward power far more than the average person, and power itself changes people, so the 'chicken or egg' question resolves to 'both,' which is why the popular tagline 'power corrupts' is true but misleadingly oversimplified.
“The evidence is very strong that both happen, that corruptible people gravitate towards power much more than the average person, and that power does change you.”
- In baboon hierarchies, biological aging (measured via DNA methylation) is fastest at the bottom due to resource deprivation and harassment, slows as rank rises, but accelerates again at the very apex because the alpha has a target on his back — making the beta position optimal: low stress with high resources, analogous to 'it's not good to be the king, but it's good to be in the court.'
“as you go down to the bottom, it's the most stressful... so you age faster. As you go up, you age a little bit slower, and then when you get to the apex, when you get to the alpha male, you age a lot.”
Public discourse focuses on why current power-holders behave badly rather than questioning whether the current set of people in power is fundamentally the wrong set.
- We morally condemn the people who gain power but rarely turn the mirror on ourselves to ask why we gave unfit people power, ignoring the cognitive biases that seduce us into irrational judgments about leaders — the uncomfortable side of the coin that even in democracies makes us complicit.
“we very rarely turn the mirror back on ourselves and say, "Why did we give somebody so unfit power?"”
- The news media is uniquely bad at discussing systemic reform and very good at covering the outrage of the day, causing public attention to zoom in on why the current crop of powerful people behaves badly (the 'what they do' question) while ignoring the more fundamental 'who' question — that the current set of people in power is the wrong set.
“the news is uniquely bad at talking about systemic reform, it's very, very good at talking about the outrage of the day”
- The dynamics of power and corruption are universal across organizations — from homeowners' association tyrants obsessing over gravel rules to CEOs and politicians — but politics intensifies them because money and power are usually twinned; organizations that consciously think about power-system design tend to avoid these problems more often.
“the dynamics that exist in other organizations are basically on steroids in politics because in most systems... money and power are twinned”
- Anti-corruption reform faces a chicken-and-egg problem because the corrupt powerful people who make the rules are the ones who would have to police themselves; in a system as broken as US politics, the only realistic agents of change may be pro-social billionaires with deep enough pockets to wield power above politicians — itself a damning indictment of the system.
“you're asking the foxes to basically police themselves in the hen house and it's not gonna happen”
Systemic selection effects determine who seeks power, but evolved mental templates and cross-species evidence suggest these mechanisms operate across diverse organizational contexts.
- Much psychology and psychopathy research suffers from sampling bias, over-representing affluent white college students at Ivy League universities and Western psychopaths, so we lack good data on whether a Chinese psychopath differs from an American one — a gap that matters because cultural mediation of even neurologically-rooted traits would be a profound insight.
“they're over-representing basically affluent white college students at Ivy League universities when they claim to be explaining humans, which is a pretty big problem”
- Because the evolution of our brains has not kept pace with the rapid transformation of our societies, the mental templates we use to select leaders may have been adaptive for ancestral small-band life but are mismatched to modern hierarchical societies — though this evolutionary-mismatch hypothesis is unfalsifiable and should be held with skepticism.
“the evolution of our brains has not proceeded apace with the evolution of our societies... some of the templates that are in our brains for leadership selection might have been adaptive for a very, very different time”
- Systems that impose high costs and dangers on leadership disproportionately push good people out and select for those who find power itself rewarding or who think they can outfox the system — illustrated by Thai professionals avoiding politics because of coups and exile, and by data that 43% of Sub-Saharan African heads of state from 1960-2010 who lost power were jailed, exiled, or killed, which incentivizes election-rigging and selects for risk-takers.
“In Sub-Saharan Africa, I crunch the numbers from 1960 to 2010, the heads of state that lost power in that period, 43% of them were jailed, exiled, or killed”
- Humans uniquely evolved the ability to throw objects with great speed and accuracy, and the development of range weapons severed the link between physical size and dominance — so unlike chimpanzees where killing a male requires being big, even small humans (or toddlers with guns) can kill adults, profoundly changing how dominance and leadership are determined.
“because we're able to effectively kill other humans at a distance, we have severed the link with range weapons between size and dominance, which is not true in many non-human primate species”
- Humans uniquely have intersecting forms of hierarchy, so a person who ranks low in their corporate world can feel valued in their family or sports team — a buffer that non-human primates like macaques lack, making low rank a more singularly stressful condition for them and helping explain the self-medicating behavior of submissive monkeys.
“One of the virtues of humans is that we have intersecting forms of hierarchy, so you can be a powerless person who's low on the pecking order in your corporate world, and yet in your family or in your softball team or whatever, you can feel valued.”