Abstract
The question this article answers: why is a casually cruel colleague at their worst when the work is quiet, and what does that imply for an organisation that cannot screen the behaviour out?
Everyday sadism is the least studied member of the dark tetrad, which is the name given to the group of four unpleasant personality traits comprising narcissism, Machiavellianism, psychopathy and sadism, and it is the only one of the four defined by an appetite rather than by an aim. This article appraises what it predicts in ordinary working behaviour, and it is deliberately organised around a situational variable rather than around the construct, because that is where the evidence is strongest.
Its central source is the nine-study programme establishing boredom as a cause rather than a correlate of sadistic behaviour, including four experiments in which boredom was induced and cruelty rose, and the condition under which the effect generalises beyond people high in the trait (Pfattheicher, Lazarević, Westgate & Schindler, 2021). A second institutional setting corroborates this programme, with 1,038 school students rated by themselves and their peers (Pfattheicher et al., 2023).
The workplace limb rests on three further sources. A ten-day experience-sampling study of 120 employees, meaning a study in which people report on themselves several times a day as the days pass, establishes that the relationship between job boredom and counterproductive work behaviour, which is the research term for the things employees do that damage the organisation or the people in it, runs in both directions within the same person from one day to the next (Kim, Kaplan, Aitken & Ponce, 2024). A two-study design showing that boredom’s route to counterproductive behaviour runs through sensation seeking provides a mechanism consistent with the experimental work (Yan, Kang & Zhao, 2025). A survey of 11,468 employees across 87 organisations also shows that job boredom spans industries rather than being confined to routine work, and that it is linked to intentions to leave and worse self-rated health (Harju, Hakanen & Schaufeli, 2014).
The article’s structural argument is that the cost of the harmful act, rather than the disposition of the person performing it, is the discriminating variable. That rests on the original behavioural demonstration of the construct, in which the free version of an aggressive act was predicted by several traits and the effortful version by sadism alone (Buckels, Jones & Paulhus, 2013), and on the pattern that the trait does its strongest predictive work in the cheapest available environment (Hidalgo-Fuentes, González-Pérez & Martínez-Álvarez, 2025).
A dedicated late chapter sets out the five kinds of person most often wrongly described this way, since the commonest practical error in this territory is not failing to spot cruelty but attributing it to enjoyment when it is produced by anger, overload, bluntness or strategy.
The practical section offers four moves and three refusals, and it is candid that the organisational evidence behind them is thinner than the laboratory evidence underneath. Three findings mark that thinness precisely: a 2026 meta-analysis, meaning a study that pools the results of the earlier studies on a question rather than running a new one, had to exclude this trait from its pooled workplace figures for want of studies (Sui, Malik, Tiliopoulos & Yu, 2026); the same Spanish employee population, surveyed once on job performance and once on organisational fairness, found sadism the strongest personality predictor of counterproductive work behaviour on both occasions (Fernández-del-Río, Ramos-Villagrasa & Barrada, 2020; Fernández-del-Río, Castro & Ramos-Villagrasa, 2022); and on the fairness question specifically, perceived fairness changed the behaviour of Machiavellian employees and left the behaviour of sadistic ones untouched. That is one dataset examined twice, not two independent replications, and the article says so.
Keywords: everyday sadism; dark tetrad; workplace boredom; counterproductive work behaviour; workplace bullying; situational triggers; evidence appraisal; organisational psychology.
PART I. WHAT ACTUALLY HAPPENS
Chapter 1: The Quiet Wednesday
1.1 Three Rooms With Nothing Going On
The scenarios below are composite. They draw on recurring patterns in research and advisory practice, and they describe no real, identifiable person or organisation.
A shared-services team is between quarters, the ticket queue has been thin for a fortnight, and one of the analysts has developed a habit of reading out other people’s email mistakes in the team channel with a row of laughing faces underneath. Nobody has complained. That is partly because each individual instance is small enough that complaining about it would make the complainer look worse than the analyst, and partly because he is perfectly pleasant in the weeks when there is work on, which makes any account of him sound unfair the moment it is spoken aloud. The person whose spelling gets quoted has started drafting everything in a private document first.
A night-shift supervisor in a distribution centre runs a genuinely tight operation between about nine in the evening and two in the morning. Then, in the flat hours before the early shift arrives, she becomes something else altogether.
She reassigns people to the least popular station at short notice, always with a reason that would survive a hearing, and she does it to the same two people every time. One of them has begun taking annual leave in single days, scattered through the month, without ever explaining why. Her manager thinks she is excellent, and on the evidence available to him she is.
A partner in a professional firm is charming with clients, unremarkable with peers, and, in the dead patch between a completed engagement and the next one starting, sends associates comments on their drafts that go a little further than the draft requires. The comments are accurate. They are also, in a way nobody can point at without sounding paranoid, arranged so that the associate reads the worst one last. Two associates who have never discussed it have independently described the experience to their mentors in almost the same words.
1.2 What the Three Have in Common
Three quite different organisations, three quite different levels of seniority, and the same two features running through all of them. The harm is small enough to be deniable in any single instance, and it is timed to arrive when there is not enough going on. I have begun with three rather than one because the pattern is not a property of a job type, a shift pattern or a hierarchy, and a single example invites the reader to explain it away with the particulars of that example. Three of them, in three different kinds of organisation, makes that harder to do.
The pattern needs a supply of empty attention, a reachable person, and an act that costs the performer approximately nothing. Every organisation I have ever worked in supplies all three, several times a month, without anybody having decided to. A third common feature makes this worth an article rather than a paragraph: in all three cases, somebody has already tried explaining to the person how the behaviour makes other people feel, and it has changed nothing at all.
1.3 Why Reaching for the Word Makes It Worse
The available word is sadist, and I would ask you to put it down before we go any further, for two reasons that have nothing to do with politeness.
The first is that the word imports a picture the research does not support. Everyday sadism, meaning the ordinary trait rather than anything clinical or sexual, is a continuous dimension on which most people sit near the bottom and almost nobody sits at the extreme, and the behaviour it predicts in ordinary life is petty rather than dramatic. The school study I will come to in a moment, which measured the behaviour directly in more than a thousand young people and asked their classmates as well as themselves, went out of its way to record that what it found occurred at a low level. That is a long way from the picture the word carries.
The second reason is more immediately practical, and it concerns your own position rather than anybody’s accuracy. To say that somebody enjoys it is to make a claim about their inner life, which is the one claim they can deny with total confidence and which you can never substantiate, and you have therefore chosen the single least defensible piece of ground available to you. It is also, in most organisations, serious enough that the conversation stops being about the behaviour and starts being about whether you were fair to say it.
What the research offers instead of a label is a trigger, telling you when the behaviour is most likely to appear, and a price, telling you what separates the people who will do it from the people who will not. Neither requires you to know anything about what is going on inside anybody. That is a smaller claim than the word makes, it is the only one that survives contact with the evidence, and it happens to be the more useful of the two. It has the further advantage of being checkable by somebody who is not you.
Chapter 2: Nothing Else to Do
2.1 The Experiment Nobody Expected to Work
The obvious explanation for cruelty is that the person is angry, and the second most obvious is that the person is under strain and passing it on; both explain it as something unpleasant arriving from somewhere and being transmitted onward. Boredom does not fit that shape at all, because boredom is an absence rather than a pressure, and until recently almost nobody had thought to look at it. Stefan Pfattheicher, Ljiljana Lazarević, Erin Westgate and Simon Schindler examined it across nine studies with 7,617 people, building the programme in three stages so each stage addressed the obvious objection to the one before (Pfattheicher, Lazarević, Westgate & Schindler, 2021).
The first stage, across six samples of 1,780 people, established that people who report being chronically prone to boredom in everyday life also report more sadistic tendencies. That is the weakest evidence in the paper, and the authors treat it that way. Two things reported by the same person on the same questionnaire will agree for a dozen reasons that have nothing to do with either one causing the other. The second stage, across four studies of 1,740 people, took the same question into settings where somebody might plausibly be bored for hours at a time: among people trolling strangers online, among military personnel, among parents describing their own conduct towards their children, and in what people reported fantasising about.
The third stage is the one that matters, and it is the reason this finding is not merely another pair of questionnaires agreeing with each other. Across four experiments with 4,097 people, the researchers made participants bored, by the straightforward method of giving them something extremely dull to sit through and then gave them an opportunity to do something unpleasant to somebody else; in one study that meant putting live worms into what participants believed was a grinder, and in others it meant quietly cutting the payment another participant was due to receive. Boredom raised the cruelty. It did so in every experiment they ran.
2.2 The Condition That Changes Everything
The single most useful thing in that paper is not the headline. It is the condition attached to it, and it is the reason this article is organised around a situation rather than around a personality. In one study, participants who had been made bored were offered several other things they could do instead, and under those conditions boredom raised sadistic behaviour only among people who already scored high on the trait. That is roughly what most readers would have predicted, and it is not the finding.
In the studies where there was nothing else available, boredom raised sadistic behaviour across the board. It did so, in the authors’ own framing, even among individuals low in dispositional sadism. I would sit with that for a moment, because it is doing considerably more work than it looks like it is doing.
The population in which boredom produces cruelty is not a fixed group of unpleasant people whose identity an organisation might in principle discover and manage, because it expands and contracts with the alternatives available, which means that the same team contains more potential offenders on a slow Wednesday than it does on a busy one, and that the number is partly set by whoever decided what the slow Wednesday would contain. The authors also examined what was driving the effect, and the route ran through excitement and novelty seeking rather than through hostility, which matters for what you do about it: a person looking for stimulation and a person discharging anger want entirely different things, and only one of those two can be satisfied by being given something interesting to do.
2.3 What This Does Not Show
I want to be careful here, because this is the strongest finding in the article and the easiest to stretch beyond what it will carry. It does not show that boredom makes people cruel in general, in the sense that a bored workforce will produce harm the way a damp wall produces mould; it shows that boredom, induced in a laboratory, raises the rate of a specific and rather narrow set of harmful acts, in settings deliberately designed so that those acts are the most interesting thing on offer. Nor does it license the reverse inference, which is the one I would most expect a reader to make and which the evidence will not support.
Finding somebody being casually cruel does not establish that they are bored. Cruelty has several other causes, and Chapter 7 sets out four of them at length.
Evidence status: evidenced. Nine studies, 7,617 people, four of them experiments with random assignment, published in a leading peer-reviewed journal. This is unusually strong for a question about why somebody behaves badly, and the strength of it is the reason the rest of the article leans on it as heavily as it does.
2.4 The Same Pattern in a Second Institution
A finding that holds only inside a psychology laboratory is a finding about psychology laboratories, so it is worth knowing that the same group went and looked for the pattern somewhere else entirely, and they went and studied it in schools, across two studies of 1,038 students aged between ten and eighteen, where they were able to do the one thing the laboratory work could not do, which is to ask other people. Sadistic behaviour was measured by self-report, meaning what each student said about themselves, and by peer report, meaning what their classmates said about them, so the finding does not rest on children volunteering unflattering things about themselves. Students who were more often bored at school were more likely to behave sadistically towards their classmates, at a strength the authors report as moderate and consistent across both studies (Pfattheicher et al., 2023).
Two things about that study are worth carrying into an organisational setting. The first is that a school is a compulsory institution containing people who cannot leave, whose attention is not always occupied, and who are in unavoidable daily contact with the same peers, which describes a very large number of workplaces more closely than anybody would like it to. The second is the sentence the authors include almost in passing, which is that the behaviour occurred at a low level.
That second point is the honest counterweight to everything else in this chapter. The phenomenon is real. It is triggered by something an organisation can influence. And it is not common.
Chapter 3: It Does Not Even Work
3.1 Ten Days, Twice a Day
The implicit theory inside the boredom-finding is that the cruelty is doing a job. Something is missing; the act supplies it, and the person feels better afterwards, which would make it a functional if repellent piece of self-medication and would tell you something important about how hard it will be to stop. JeongJin Kim, Seth Kaplan, John Aitken and Lida Ponce tested the timing of that directly, in the only design that can answer it, which is to follow the same people from one day to the next rather than comparing different people with each other (Kim, Kaplan, Aitken & Ponce, 2024).
They took 120 employees, mostly in business, finance and technology roles, and asked them twice a day for ten working days how bored they were and what counterproductive things they had done. That produced 1,161 separate daily observations, which is what makes it possible to ask whether today’s boredom precedes tomorrow’s behaviour or the other way round. Both, as it turns out.
3.2 The Loop That Feeds Itself
A day of higher boredom was followed by a rise in counterproductive behaviour the next day, which is the expected half of the result and the half that most readers will have assumed. The unexpected half is that a day of higher counterproductive behaviour was followed by a rise in boredom the next day. Whatever relief the behaviour buys does not survive to the following morning, and what arrives instead is more of exactly the state that produced it. Whatever the medicine is, it has worn off by the following morning.
The authors are careful about what this does and does not establish, and so am I. Counterproductive work behaviour is a broad category covering withdrawal, rule-bending, theft of time and unpleasantness towards colleagues, and it is not a measure of sadism, so this study tells you about the wider family of behaviours that boredom produces rather than about the particular member of that family this article is concerned with. The experimental work in Chapter 2 makes the link from one to the other, not this study.
Evidence status: evidenced as to the reciprocal pattern; inference as to sadism specifically. The daily design establishes that job boredom and counterproductive behaviour feed each other within the same person from one day to the next. The step from that family of behaviours to the particular behaviour this article is about is mine, and it rests on the experimental work in Chapter 2 rather than on this study.
3.3 Why This Matters More Than the Headline Finding
I would argue that the loop is more useful to you than the trigger, for a reason that only becomes obvious once you try to act on either: if cruelty at work were successful self-medication, then the person doing it would be responding rationally to their own situation, and the only available intervention would be to change the person, which is the intervention nobody has ever managed to carry out on anybody. Because it is not successful, because the state returns worse the following day, what you are looking at is a behaviour that repeats precisely because it fails.
That reframes the whole problem in a way I find genuinely clarifying. The supervisor in the distribution centre is not working through something and arriving somewhere. She is going round.
And the practical consequence of a loop is different in kind from the practical consequence of a trait, because a loop can be interrupted at any point on its circumference. You do not have to find the beginning of it, you do not have to establish who started it, and you do not have to change anybody’s disposition to stop it turning.
PART II. WHY IT RUNS THIS WAY
Chapter 4: The Price of the Act
4.1 When Hurting Someone Is Free
Chapter 2 established that the number of people who will behave cruelly rises when nothing else is available, which is a claim about the alternatives on offer. This chapter is about the other half of the same arithmetic, which is what the act itself costs the person performing it. The study that put everyday sadism on the map in the first place happens to be the cleanest demonstration of that, and it was not designed to make the point at all.
Erin Buckels, Daniel Jones and Delroy Paulhus set out to show that an ordinary, non-clinical form of sadism could predict actual behaviour rather than merely questionnaire answers (Buckels, Jones & Paulhus, 2013). Until 2013, the field had three dark traits and a persistent suspicion that a fourth one was missing.
Their first study offered seventy-one participants a choice of four unpleasant laboratory jobs: killing bugs in what appeared to be a coffee grinder; assisting somebody else who was doing it; cleaning dirty lavatories; and enduring cold-water pain. People higher in sadism chose the bug-killing at a higher rate than everybody else, which is the headline the study is remembered for. The number underneath the headline is the one worth keeping: just over a quarter of participants chose to kill the bugs, and just over a quarter more chose to help.
4.2 The Willingness to Work for It
The second study is where the argument of this chapter actually comes from, and it is the half almost nobody quotes. Participants were given the opportunity to blast an innocent person, somebody who had done nothing whatever to provoke them, with unpleasant noise. In one version, aggression was free, requiring nothing beyond the decision to do it; in the other, participants first had to complete a tedious letter-counting task to earn the opportunity.
When the act was free, several of the dark traits predicted who did it. When the act had to be worked for, only sadism did.
I would put the consequence as plainly as I can, because it is the structural claim of this article. Making the harmful act cost something does not reduce cruelty evenly across a population; it removes almost everybody except the small number of people who want it enough to pay. That is not a comfortable finding, since it concedes that friction will not stop the person you are most worried about, and it is nevertheless the most actionable thing in the literature, because the population it does remove is much the larger of the two and because it is a property of the environment rather than of anybody’s character.
4.3 Where the Price Is Lowest of All
If the price of the act is the operative variable, then the trait should do its strongest predictive work wherever that price is lowest, and one environment is essentially zero. Sergio Hidalgo-Fuentes, Elisa González-Pérez and Isabel Martínez-Álvarez pooled twenty-four studies from eleven countries, covering 14,044 people, on the relationship between the four dark traits and online trolling (Hidalgo-Fuentes, González-Pérez & Martínez-Álvarez, 2025). All four traits predicted it, which is unsurprising. The interesting result is the order: sadism was the strongest of the four by a clear margin, narcissism was the weakest, psychopathy sat between them and closer to sadism, and Machiavellianism sat closer to narcissism.
Consider trolling as a setting rather than a subject. The act takes seconds, and it requires no physical presence at all. The target cannot retaliate in any way that reaches the person; there is usually no record attached to a real name, and an audience is watching.
It is about the lowest-friction harm available to a human being. It is also the setting in which this trait outruns the other three.
The workplace equivalents are not hard to identify once you have the principle in hand. A comment in a channel, a name left off an invitation, feedback delivered in front of people rather than beside them, and a reassignment issued by system notification rather than in a conversation are all acts that cost the person issuing them nothing at all.
4.4 What Friction Actually Buys You
Putting the two halves together gives a picture that is more useful than either one alone. The number of people prepared to behave badly rises as the alternatives fall, which is Chapter 2, and it falls as the cost of the act rises, which is this chapter, and both of those are dials an organisation owns outright. Neither requires anybody to identify a single individual, which is why I would reach for them first.
There is an honest limit on all of this, and the second Buckels study has already told us what it is. Raising the price will not reach the person at the top of the distribution, who is the person most readers will have had in mind since the first page. Pretending otherwise would be exactly the kind of overclaim this series exists to avoid.
What raising the price does instead is stop the drift. It prevents the slow Wednesday from recruiting the six other people who would never have started it on their own and who, in an environment where doing so costs them nothing, will cheerfully join in.
Evidence status: evidenced as to the laboratory finding; inference as to the workplace application. The free-versus-effortful comparison is an experimental result on seventy-one people. The claim that the same logic operates on a team channel or a rota is mine, reasoned from that result and from the trolling pattern, and no study has tested it inside an organisation.
Chapter 5: What Boredom Is at Work
5.1 Not a Mood, and Not Rare
One version of this argument fails immediately: the one that treats boredom as a personal failing of people who cannot entertain themselves. If that were what boredom was, it would be nobody’s problem but theirs, and an organisation would have nothing whatever to work with. The largest workplace study of the question says otherwise, and it says so clearly. Harju, Hakanen and Schaufeli measured job boredom across 11,468 employees in 87 Finnish organisations, which is the largest workplace sample I am aware of on the question, and their headline conclusion is the one that matters here: job boredom is a phenomenon concerning a wide range of industries rather than a property of a few obviously tedious jobs (Harju, Hakanen & Schaufeli, 2014).
It was highest among men under thirty-six, and highest in transport, manufacturing, and arts, recreation and entertainment, which is a list that should give pause to anybody who assumed this was a problem confined to data entry, and it also travelled with outcomes an organisation already cares about, the list being longer than most people expect: intention to leave, intention to retire early, worse self-rated health, reduced ability to work, and more stress symptoms. None of that is free. That last cluster is why boredom belongs on an operational agenda even for a reader who finds the cruelty argument unconvincing, because an organisation with a boredom problem is already paying for it in turnover, and the behaviour this article is about is a second charge on the same account.
5.2 The Engine Underneath It
What connects boredom to bad behaviour is not obvious, and two independent lines of work point to the same mechanism. The experimental programme in Chapter 2 found the route running through excitement and novelty seeking rather than through hostility. A separate two-study design by Yan, Kang and Zhao, combining a scenario experiment with 97 people and a three-wave survey of 149 employees, found that workplace boredom reaches counterproductive behaviour through sensation seeking, and that the route is stronger among people with weaker self-control (Yan, Kang & Zhao, 2025).
Two different methods, two different countries, one mechanism. What the bored person is looking for is stimulation, and harm is simply the cheapest stimulation available in most offices.
I find that a genuinely useful thing to hold on to, because it explains something about these episodes that anger never explains properly. The cruelty is frequently creative, in the small, effortful, almost playful way that the ordering of the partner’s comments in Chapter 1 is creative, and creativity is what somebody does when they are looking for interest rather than discharging pressure.
5.3 There Is More of It Than There Used to Be
A trigger that is becoming more common deserves more of your attention than one that is stable, and the evidence suggests this one is rising. Elizabeth Weybright, John Schulenberg and Linda Caldwell examined boredom in a national sample of American adolescents between 2008 and 2017 and found it rising from 2010 onwards after a flat opening period (Weybright, Schulenberg & Caldwell, 2020). Katy Tam and Michael Inzlicht set that trend alongside comparable evidence from other countries and argued that digital media is part of the cause, through dividing attention, raising the level of engagement people come to expect, and functioning as a coping strategy that does not actually work (Tam & Inzlicht, 2024).
Two qualifications keep that honest, and neither of them touches the direction of travel. The trend evidence concerns adolescents and students rather than working adults, so applying it to a workforce is an extrapolation rather than a finding. The second source offers a theoretical perspective proposing a cause rather than a study establishing one.
Evidence status: contested, and briefly. That reported boredom has risen is supported by national trend data in more than one country. Why it has risen is genuinely open, and the digital-media account is a proposal rather than a result. The dispute does not change anything a manager would do, so I am not going to spend your time on it.
5.4 Why Busy Organisations Produce It Anyway
The objection I would expect from most readers of this series is that their organisation is not remotely boring, and that nobody in it has had a quiet Wednesday since 2019. That objection misreads what boredom is, and the misreading is common enough to be worth a paragraph of its own. Job boredom is not an absence of tasks. It is an absence of engagement, which is why the Finnish data finds it in industries nobody would describe as idle, and why a person can be fully occupied and thoroughly bored at the same time.
Four ordinary organisational conditions produce it reliably, and not one of them involves having too little work to do. Long stretches of waiting inside a process somebody else controls. Work whose outcome the person will never see. Tasks repeated often enough that they require no attention, and the stretch after a large piece of work finishes and before the next one is allocated.
The third scene in Chapter 1 sits squarely in the fourth of those, and it is worth noticing that a professional services firm is not an obviously boring place to work. The partner in that scene was not underemployed in any sense his timesheet would have shown.
Chapter 6: The Gap Where the Workplace Evidence Should Be
6.1 The Meta-Analysis That Had to Leave It Out
Everything so far has been about boredom, laboratories, schools and the internet, and a reasonable reader will by now be asking what happens when somebody studies this trait at work directly, to which the answer is that almost nobody has, and the cleanest demonstration of that comes from a paper which set out to do something else entirely. Sophia Sui, Sajida Malik, Niko Tiliopoulos and Lei Yu published a systematic review and meta-analysis in 2026 on dark personality and workplace bullying, screening eighty-nine records, including sixteen studies and pooling nine of them (Sui, Malik, Tiliopoulos & Yu, 2026).
They produced pooled figures for three traits, and the three came out in the order most readers would expect, with psychopathy the strongest predictor of bullying other people, then narcissism, then Machiavellianism. They did not produce a figure for sadism. They could not, because the workplace studies of it are too few and too dissimilar to pool.
I would ask you to hold that beside the subject of this article for a moment. The one trait defined by taking pleasure in somebody else’s discomfort is the one trait with no pooled workplace figure attached to it, in a review specifically about people who hurt their colleagues.
6.2 The Workplace Findings That Do Exist, and One Is Awkward
A reasonable body of workplace survey work exists, and almost all of it comes from one research group working with one employee population in Spain, across two papers published two years apart. That is worth being upfront about before using either finding, so I am putting it here rather than in a footnote.
The earlier paper is a broader study of what the Dark Tetrad predicts about ordinary job performance, and it is worth setting out because the finding is more direct than anything else in this chapter. Fernández-del-Río, Ramos-Villagrasa and Barrada surveyed 613 employees, measured all four dark traits alongside the two standard, non-dark personality frameworks that most workplace tests already use, the Big Five and HEXACO Honesty-Humility (a trait covering how sincere, fair and unassuming someone is), and tested which of them predicted task performance, contextual performance and counterproductive work behaviour once those standard measures had already been accounted for (Fernández-del-Río, Ramos-Villagrasa & Barrada, 2020). After that accounting, sadism was still the single strongest predictor of counterproductive work behaviour in the study, ahead of psychopathy, and it was the trait most consistently linked to worse ordinary task performance. A conventional personality profile would have missed both results.
The second study is the one worth setting out in full, because its result cuts against something most of us would like to believe: Fernández-del-Río, Castro and Ramos-Villagrasa surveyed 613 employees across a range of occupations, measuring all four dark traits, the counterproductive behaviours people directed at their organisation and at each other, and how fairly they felt they were treated (Fernández-del-Río, Castro & Ramos-Villagrasa, 2022). Psychopathy and sadism predicted harm aimed at the organisation, and of the four traits, sadism was the one whose direct link to harm aimed at other people the study supported.
Then they asked whether feeling fairly treated changed anything, and the answer differed by trait in a way I find uncomfortable. Employees high in Machiavellianism who felt they were treated fairly in the interpersonal sense did not go on to behave badly towards colleagues, which is an encouraging result and the one an organisation would hope for. Employees high in sadism behaved harmfully towards other people whatever their view of how fairly they were treated.
I would ask you to hold both findings loosely, and for a reason more specific than the usual caution about self-report. The two studies share the same sample size, the same country, and two of three authors, which is not proof they draw on the same 613 people but is close enough to it that I am not willing to treat them as independent corroboration. Read them as one substantial dataset, examined twice, for two different questions. That is still more than most dark-trait constructs get in a workplace setting, which is precisely Chapter 6.1’s point, and it is a great deal less than two separate studies would be worth.
6.3 Why I Am Telling You What Is Missing
I could have written this chapter as a set of confident workplace recommendations, by quietly treating laboratory findings as though they had been established in an office. Almost every treatment of this subject available to you does exactly that. So I have not. The first reason is that you would eventually find out, and a reader who catches an author overreaching once has no way of knowing which of the remaining claims to trust.
The second is that the absence is itself informative, and in this case it is informative in a direction that helps you. A field that has not established which organisational interventions work is not a field that will let you outsource the judgement, so the reasonable posture is to act on the mechanism evidence, which is strong, while holding the intervention claims loosely, which is what Chapter 8 does explicitly for every item on its list.
I looked for a review that might close the gap rather than confirm it, and the one that exists does the opposite. The most comprehensive synthesis of this literature to date draws on more than eighty-five studies on everyday sadism’s theoretical models, measurement, and prevalence, and it reaches organisations only in passing, on its way to the same online and interpersonal settings this article has already covered (Ward, 2025). If the field’s own widest-angle review adds nothing further on the workplace specifically, I am not going to invent something in its place.
Evidence status: no evidence base. No trial has been run on any organisational intervention aimed at this behaviour. The claim here rests on an absence rather than on a positive finding, and the absence was confirmed against a 2026 systematic review that went looking for the workplace literature and reported that there was not enough of it to pool.
PART III. BEFORE YOU USE ANY OF THIS
Chapter 7: The Five People You Will Mislabel
Everything up to here describes one pattern, which is harm performed for stimulation by somebody with nothing better to do. This chapter is about five people who produce behaviour that looks identical from the outside but isn't that pattern at all. In my experience, the most common error in this territory is not failing to notice cruelty, but misattributing it. If reading this article makes you more willing to conclude that a colleague enjoys causing harm, it has not worked.
7.1 The Angry One
The most obvious alternative explanation is the one this article has been arguing against, and it would be dishonest to leave it looking defeated. Anger produces harm too, and it produces it in a recognisably different shape: it has a target with a history, it arrives after something rather than during nothing, it tends to be disproportionate rather than small, and the person doing it will usually tell you why if asked directly, because from the inside it feels justified.
The distinguishing question I would put is about timing rather than about content, and it is short enough to ask yourself in a corridor. Does the behaviour cluster around particular provocations, or around particular gaps in the week? The timing is the tell, and it is the only part of this you can watch directly.
7.2 The Overloaded One
The second alternative is the person who is not under-stimulated at all but comprehensively overwhelmed, and whose behaviour towards colleagues has become curt, dismissive and occasionally unkind because the capacity to do otherwise has run out. This one matters more than the others because the remedy is close to the opposite of the remedy in Chapter 8. Giving a bored person something demanding to do is helpful; giving an overloaded person something demanding to do is the cruelty they are then accused of.
The distinguishing signal I would look for is whether the unpleasantness is efficient or elaborate. Overload produces abruptness, which is cheap in time and effort, while the pattern this article describes produces the carefully ordered comments in Chapter 1, which are neither cheap nor quick nor accidental. Watch for effort, because effort is what boredom supplies and overload removes.
7.3 The Blunt One
The third is the colleague who has always been direct, who is direct with everybody equally, who has no idea that anybody experiences it as harsh, and whose behaviour has not changed at all during the period you have been watching it, so that directness of that kind is not harm even where it is felt as harm, a distinction that is easy to state and surprisingly hard to apply, because the person on the receiving end feels something in both cases. The test I would use is one of variation rather than of intensity: the blunt colleague is blunt with the chief executive, and the pattern in this article selects for people who cannot make it cost anything.
The night-shift supervisor in Chapter 1 is not doing it to the shift manager. That is the whole of the distinction.
7.4 The Strategic Bully
The fourth is the person who deliberately harms colleagues for a reason, usually position, credit, or removing a rival. This is the Machiavellian and psychopathic territory covered by the review in Chapter 6, where the pooled workplace figures actually exist, and it is a different problem with a different shape: strategic harm is aimed at people who are in the way rather than at people who are within reach, it stops when the objective has been achieved, and it is entirely capable of being charming to the same person the following week.
The overlap between the two is real, and it is worth knowing how large. Pooling the research on how the four traits relate to each other, sadism sits closest to psychopathy, next closest to Machiavellianism and furthest from narcissism, which means the trait shares a good deal with strategic harm while still measuring something the other three do not capture between them (Bonfá-Araujo, Lima-Costa, Hauck-Filho & Jonason, 2022). I would separate the two on the question of what the behaviour is for, and the test is uncomfortably simple: if you can name what the person gains, you are probably not looking at the pattern in this article, and if the honest answer is that they gain nothing at all, you may be. Ask what it buys them.
7.5 The One Who Needs a Clinician
The fifth boundary is categorical rather than a matter of degree, and it is the one this whole series holds most firmly. Everyday sadism as studied here is a continuous trait on which the entire population sits somewhere, with most people near the bottom, and it is not sadistic personality disorder, which was removed from the main diagnostic manual decades ago and remains a matter for qualified clinicians. It has nothing whatever to do with the sexual literature that shares part of its name, and the field’s own reviewers are explicit that the two should not be conflated (Ward, 2025).
A questionnaire score is not a diagnosis. An observation is not a diagnosis either, and nothing in this article can be turned into one. I would ask you not to read that as a disclaimer bolted on for safety, because everything else depends on it.
7.6 And the Numbers Do Not Describe a Person Anyway
A sixth caution applies to every figure in this article without exception. All the evidence here comes from measuring large numbers of people and reporting how the measurements move together, and a relationship established across thousands of people does not describe the particular person standing in front of you. The nine-study programme in Chapter 2 tells you what happens to a rate when boredom rises; it tells you nothing at all about the analyst in the first scene.
That is not a technicality, and it is the reason this article gives you a trigger and a price rather than a diagnostic checklist. You are entitled to notice a pattern over time. You are not entitled to conclude anything about what somebody is.
7.7 How Common Is It, Really
This is the question readers actually arrive with, and the honest answer comes in two parts, one of which is unsatisfying. The unsatisfying part is that nobody knows the base rate in a working population. Nobody has counted. I have not found a defensible figure for the proportion of employees who would qualify as high on this trait, and I would treat any percentage you are offered as unsupported until you have seen where it came from.
What can be said is narrower and still useful. In the school study, where more than a thousand young people were rated by themselves and by their classmates, the authors record that sadistic behaviour occurred at a low level (Pfattheicher et al., 2023), while in the laboratory just over a quarter of an ordinary student sample chose to kill bugs when the alternatives on offer were cleaning dirty lavatories and enduring cold-water pain, which is a striking number and also a statement about how unappealing those alternatives were.
Between those two observations sits the position I would actually defend. Sustained enjoyment of causing harm is uncommon; participation in small, cheap, unkind acts when the environment makes them free is not uncommon at all; and the distance between those two statements is precisely the distance this article is asking you to hold.
PART IV. WHAT TO DO
Chapter 8: Four Moves and Three Refusals
Evidence status: mixed and stated per item. The mechanism evidence in this field is strong and the intervention evidence does not exist, which Chapter 6 establishes against a 2026 systematic review. Each item below carries its own status, and anybody offering you a validated method for handling a sadistic colleague is describing a product rather than a literature.
8.1 Treat Empty Time as a Hazard, Not a Windfall
This follows directly from Chapter 2, and it is the item most likely to be dismissed as soft, so I will put it in operational terms. Most organisations treat a slow period as recovered capacity, meaning either a gift to spend on catching up or an embarrassment to hide from whoever is counting. Neither treatment involves anybody deciding what people will actually be doing.
The evidence says that the interval in which nothing has been decided is the interval in which the risk sits. Plan the trough in advance, in the same way you would plan a peak.
The practical version is unglamorous. Know where the troughs are in your own operation, because most of them are predictable from the calendar; make sure somebody has decided what a trough is for before it arrives; and pay particular attention to the gap after a large piece of work completes, which is the one nobody plans because it feels like a reward.
I would add one caution: the research supplies directly. The alternatives have to be genuinely available and genuinely preferable, because the experimental finding turns on something else being worth doing, and a list of low-grade administrative tasks is not an alternative in the sense the studies mean. Evidenced as to the mechanism; practitioner judgement as to the design.
8.2 Put a Price on the Act, Not a Profile on the Person
This is Chapter 4, and it is the item I would implement first. It requires no conversation with anybody and no judgement at all about who is doing what. The acts described in this article share one property: they cost the person performing them nothing, so the question to ask of your own environment is where harm is currently free. The answer is usually a short and rather specific list: an open channel where an unkind remark reaches an audience instantly and leaves no trace anybody will act on; a rota or reassignment tool that lets one person move another person’s life around without a conversation; a review process that reaches its subject in writing without anybody standing behind it; and any route by which somebody can be excluded without the exclusion being visible to a third party.
Adding friction to those does not mean adding bureaucracy to all of them, which would cost more than the problem is worth. It means adding it at the one point where the act becomes free: a reassignment that requires a sentence of reason attached to it, feedback that is delivered by a named person rather than by a system, and a channel with an established norm that quoting a colleague’s mistake is done privately.
The honest limit on this is the one Chapter 4 already stated, and I would rather restate it than let you discover it for yourself. This will not stop the person at the top of the distribution, and it is not aimed at them. Evidenced as to the laboratory finding; inference as to the application.
8.3 Record the Pattern, Because the Instances Are Deniable by Design
Every scene in Chapter 1 has the same defensive structure: no single instance would support a complaint, but the pattern would support one easily, if only somebody had written it down. That is not an accident and it is worth understanding as a feature rather than as bad luck: harm performed for stimulation has to be noticeable to the target in order to be worth doing, and it has to be defensible in isolation in order to be survivable, and those two requirements between them produce exactly the shape you are looking at.
So the useful artefact is a dated list of instances with the target named and nothing else in it, kept by whoever is in a position to keep one. Not a characterisation, not an account of what the person seems to enjoy, and not an inference about motive. Dates, what happened, and who it happened to.
The reason to insist on that narrowness is that a list of dates is a witness statement and an assessment of what somebody is is not, which is a distinction this series has drawn before and which decides whether the record is usable when it is eventually needed. It also has a second benefit that people underestimate, which is that a list either shows clustering in the quiet weeks or it does not, and you will find out which. Practitioner judgement.
8.4 Give the Stimulation Somewhere Legitimate to Go
If the mechanism is sensation seeking rather than hostility, as two independent lines of evidence in Chapter 5 suggest, then the boring remedies are aimed at the wrong thing entirely. The intervention that follows from the mechanism is not a values workshop and not a reminder about respect, neither of which supplies any stimulation whatever; it is novelty, difficulty and consequence, meaning a problem that is genuinely hard, a piece of work whose outcome the person will actually see, a rotation that changes what the days contain, or a responsibility that carries real exposure.
I know this may read as rewarding bad behaviour, and I would meet that objection directly rather than dodge it.
The intervention is not addressed to the individual who has already done something, who should be dealt with on the merits of what they did, in the ordinary way, through whatever process the organisation has for exactly that purpose; it is addressed to the condition that expands the population of people who might do something similar next month. Boredom is the target, not the person. Practitioner judgement, reasoned from Pfattheicher et al. (2021) and Yan et al. (2025). Untested as an intervention.
8.5 Do Not Screen Candidates for It
Instruments exist, several of them are perfectly respectable pieces of research, and one or two will be sold to you inside a broader dark-side assessment without the word sadism appearing anywhere in the brochure. I would not buy one, and the reasons run deeper than cost or legal exposure. Four reasons against, briefly. The whole argument of this article is that the behaviour is produced by an interaction between a disposition and a situation, so a measure of the disposition alone is predicting one half of a two-part product; the scores were validated in anonymous, low-stakes conditions that a selection process does not resemble; the base rate is unknown, which means nobody can tell you what such an instrument would actually catch or miss in a working population; and screening people out on an inferred appetite for cruelty carries an obvious discrimination exposure for a benefit nobody has demonstrated.
There is a fifth reason I find more persuasive than the other four put together. Even if the screening worked perfectly, it would address the smaller half of the problem, because Chapter 2’s finding is that the population who will behave this way expands to include people no screen would ever have flagged. Evidenced as to the mechanism argument; practitioner judgement as to the recommendation.
8.6 Do Not Rely on Fairness to Fix It
This is a refusal I would rather not have to write, because organisational fairness is worth pursuing for a dozen independent reasons and I am not arguing against a single one of them. The finding in Chapter 6 is nonetheless specific, and it should be held rather than smoothed over. Among 613 employees, feeling fairly treated in the interpersonal sense changed how Machiavellian employees behaved towards their colleagues and did not change how sadistic employees behaved at all (Fernández-del-Río, Castro & Ramos-Villagrasa, 2022).
If that holds, a fairness intervention will improve several things but not this one, and an organisation that has done the fairness work and expects this behaviour to have gone with it will be surprised by what is left. Do both, and expect only one of them to reach this. Evidenced, from a single study. One country, self-report on both sides, and it has not been replicated.
8.7 Do Not Tell the Person They Enjoy It
The final refusal is about the conversation itself, and it mirrors 1.3. The moment you say that somebody is enjoying causing harm, you have made an assertion about their internal state that they can deny completely and that you can never evidence, and you have also, in most organisational settings, said something serious enough that the conversation stops being about the behaviour and becomes about whether the accusation was fair.
What survives that conversation is the material from 8.3, which is a dated list of things that happened, described without adjectives, with the effect on the work stated rather than the effect on anybody’s feelings. Not you take pleasure in humiliating people, which is an argument about your judgement and one you will probably lose, but on the fourteenth, the twenty-first and the sixth you quoted Rachel’s drafting errors in the team channel, she now clears everything through me before sending, and that has added a day to every piece of work she produces.
The second version has the advantage of being true whatever is going on inside the person. That is the only kind of claim worth making about somebody else’s mind. Practitioner judgement.
Chapter 9: What This Changes
9.1 The Short Version
If you read nothing else in this article, I would ask you to read this.
The best-evidenced trigger is boredom, not anger. Across nine studies and 7,617 people, including four experiments in which boredom was induced rather than merely measured, boredom raised the rate of cruel behaviour, and the route ran through wanting stimulation rather than wanting to hurt anybody.
And the population it affects is not fixed. Where participants had other things they could do, boredom produced cruelty only among people already high in the trait; where nothing else was available, it produced cruelty across the board, including among people at the bottom of the scale. That finding makes this an organisational problem rather than a personnel one.
It fails, which is why it repeats. Across 120 employees reporting twice a day for ten days, a bored day was followed by a worse-behaved day, and a worse-behaved day was followed by a more bored one. Nobody is working through anything. They are going round.
Cost is what separates the many from the few. When hurting an innocent person was free, several dark traits predicted it; when it first had to be earned through tedious work, only sadism did. Raising the price removes most people, but not the one you are thinking of.
And the workplace evidence is thinner than anybody admits. A 2026 review of dark personality and workplace bullying produced pooled figures for three traits and could not produce one for the trait actually defined by enjoying harm, because there were not enough workplace studies to pool.
9.2 So What Do You Do
Three things, and none of them involves confronting anybody about their character.
Decide what the quiet periods are for, before they arrive. Empty attention is the input, it is predictable from your own calendar, and it is the only part of this chain an organisation can remove rather than merely discourage.
Find where harm is currently free and make it cost a sentence. A reassignment that requires a reason, feedback that arrives from a person rather than a system, and a norm that a colleague’s mistake is raised privately are all cheap to introduce, and each of them converts a free act into an act with a small price attached to it.
Write down dates, not adjectives. The instances are deniable individually and the pattern is not, and the only version of this that survives being challenged is the one describing what happened rather than what somebody is.
9.3 Why Any of This Matters
Two reasons, and the second is the one I would leave you with. The first is that the standard advice on these subject points at the wrong thing and therefore fails in a way that damages the person following it: people spend months trying to establish that a colleague is cruel, which requires proving something about another person’s inner life, and they end up looking obsessive while the behaviour continues. On this evidence, the inner life was never the tractable part of the problem.
The second is more uncomfortable and I would rather not soften it. The central experimental finding in this article is not that some people are cruel; it is that when the alternatives run out, the number of people who will be cruel goes up, and the increase reaches people who score at the very bottom of the trait.
That is a finding about most of us rather than about a minority, and it makes the empty afternoon a moral fact about an organisation rather than merely an operational one. Somebody decides how much of that time exists and what it contains. Somebody chose that. In almost every organisation I have worked in, nobody has ever thought of it as a decision at all.
9.4 What I Still Do Not Know
Four things remain genuinely unknown to me, and I would rather state them as questions than resolve them by omission.
Whether the laboratory boredom effect operates at all inside an organisation, since being made bored for twenty minutes and being bored at work for eleven months are not obviously the same input. Whether any of the four moves in Chapter 8 improves anything, since not one of them has been tested against this problem and Chapter 6 establishes that the field has not tested anything else either.
Whether the fairness result in 6.2 survives replication, because it is a single study and it carries more weight in this article than a single study normally should. And what the base rate actually is in a working population, which is the question every reader wants answered and the one I have been least able to answer honestly.
A closing note on how I have handled the material, because it applies to everything in this series. Everyday sadism is the most sensational construct in this literature and the one where the temptation to write about monsters is strongest, and the evidence points somewhere a great deal less dramatic and considerably more actionable. I have tried throughout to keep the laboratory findings and the workplace inferences visibly apart, to name the absence in the middle of the field rather than write over it, and to give you a trigger and a price rather than a way of deciding what your colleague is. What I would most like to have made harder is a confident verdict about the person you had in mind while you were reading.
