Who Holds the Keys

Fascomachy · Document 6a

A promise is a claim about the future, made by a person in the present. This document is about the distance between them, and about the room in which that distance does its damage.

Evidence tier: Graded. Claims sorted by strength of evidence; limits stated, no falsifier program. Evidence Protocol

Plain-Speech Canon

This document records what happens when promises keep arriving and conduct stops following them, and who pays for the difference. It tests one popular explanation, the Dark Triad of narcissism, Machiavellianism, and psychopathy. The Dark Triad describes tendencies measured across groups. It does not identify a type of person.

The questions come in order. What happened. What room it happened in. Only then, what tends to drive the person; asked first, that question is a verdict looking for a word.

Conduct is where to start, though it is not always easy to establish. Motive can sometimes be read from admissions, threats, or demands, and rarely from broken promises alone. What a person can do next depends less on naming their partner than on whether they have time, records, someone to speak to, and a usable way out.

The word comes last.

I · A demonstration

Fictional illustration. It describes no single person and diagnoses no one.

The first promise was small, and it was kept.

They said they would call on Thursday, and they called on Thursday. They said they would meet your sister, and they met your sister and were good with her, better than you had been with her in years. You noticed that. You filed it.

The promises grew, and for a long while the conduct kept pace. They wrote the trip on the kitchen calendar in pen, not pencil, and you liked that more than you admitted. They sent you a listing for an apartment with a second bedroom and a south window, and you saved it. They started the conversation about money themselves, unprompted, which is what a person does when they have nothing to hide. You remember thinking exactly that.

Then there is a stretch you have trouble dating. The week of the trip came and went, and it was still written on the calendar, in pen. The listing expired; your phone still has the screenshot. The promises kept arriving, fluent and specific and warm, and the conduct quietly stopped following them. There was no break. There was a drift, and every inch of it arrived with an explanation. Some of the explanations were moving.

At some point the trip that didn’t happen became something that had happened to them. You had paid for the cancelled booking, and you were also doing the comforting. It did not strike you as strange until much later.

They had a key by then. It hung on the hook by the door, beside yours.

Here is the part you will return to. You never caught them lying. You were sure, each time, that they meant it; you would have said so under oath. What you had was the look of a person meaning something, across that kitchen table, and a record that did not match it. For a long time you trusted the look over the record.

Whether they meant it could wait. What it cost could not.

II · The premise

The Grammar of Manipulation, the first document in this series, argues that much of the machinery that bends judgment has no single operator. A checkout timer resets because of how a component caches state; a cancellation flow grows to twelve clicks through a series of choices that each made sense locally. The Grammar‘s claim is that manipulation often lives in environments: rooms in which sound rules of inference return wrong answers.

Distributed authorship is not the same as no authorship. Every one of those steps was a decision, and someone could have known where they led. What distribution does is thin responsibility until it is hard to pin on anyone. In a household that is harder to hide behind, because the people whose acts shaped the room are few enough to name.

Still, the opening illustration settles what happened, not why. It shows costly unreliability. Whether it also shows manipulation or control is what the rest of this essay tests, and the telling leans toward suspicion; that lean is part of what needs testing. A person whose promises outrun their conduct might be exploiting someone. They might also be depressed, overcommitted, afraid of conflict, ashamed, or simply bad at estimating the future. The pattern is consequential in every one of those cases. It points toward a manipulative disposition in only some of them. The question this essay pursues is when the Dark Triad lens tells us something those other explanations do not.

The lens comes from Delroy Paulhus and Kevin Williams, who in 2002 studied narcissism, Machiavellianism, and psychopathy in a non-clinical sample of undergraduates and found the three overlapping but distinct. They are measured as continuous scores, not categories. A high score is a position on a scale, not a diagnosis.

A tendency does nothing by itself. What matters in a household is how four things combine:

  • Disposition: what a person tends to want, and how readily they reach for others to get it.
  • Conduct: what they actually do.
  • Resources: money, housing, friends, and information, and who controls each.
  • Constraints: rent, work, children, immigration papers, and the other pressures no one in the household chose.

Exits close in three different ways, and they carry different responsibility.

  • Restrictions imposed. Someone acts to narrow another person’s options: discouraging a friendship, moving the money into one account, making each visit from family cost an argument. Evan Stark’s work on coercive control describes how acts like these can accumulate into entrapment, a pattern that shrinks a person’s freedom of action in everyday life even where no single incident looks severe. Stark developed that account mainly to describe men’s control of women, and he tied it to sex inequality. His analysis supports looking at cumulative restriction. It does not make every broken promise coercive, and it says nothing about personality traits as a cause.
  • Limits set by circumstance. These can reinforce one another. Take a hypothetical household. A lease requires proof of income. The income requires a job, and the job requires set hours. The hours work only because the partner collects the child at three-thirty. No single link is unusual. Together, they can turn an exit that exists on paper into one that is hard to use, unless someone else can do the pickup.
  • Adaptation under pressure. The person inside gives ground, one accommodation at a time, to keep the peace. Adapting to a constraint is not authoring it.

Constraints like these can give one person leverage without showing that the person built them. A partner whose availability the household schedule depends on holds power over the other’s options. Whether and how they use that power is a question about conduct.

Responsibility follows the acts that impose restrictions. It does not transfer to the person who adapted, and a high trait score neither creates that responsibility nor excuses it.

That changes the order of the questions. The popular order starts with what are they? The useful order starts elsewhere:

  1. What happened? Name the conduct, not the character.
  2. What room did it happen in? Who holds the resources, and which exits are still open?
  3. Only then: what tends to drive this person?

III · Three temperaments

The test first

The opening illustration is invented, so it can frame questions but cannot answer them. It establishes two things: repeated non-performance, and a steady supply of explanations. It leaves four unresolved:

  • Sincerity: whether the promises were meant when they were made.
  • Capacity: whether they could have been kept.
  • Motive: what the drift was for, if anything.
  • Control: whether the drift changed who held the resources and the exits.

A depressed partner, an overcommitted one, and an exploiting one can produce the same calendar page. What separates them is what surrounds the broken promise. The questions below are for reading a relationship, not for scoring a person. Each has benign answers, and each is worth asking alongside its counterexamples.

  1. Where do the failures land? Does non-performance fall mainly on commitments that cost them something and benefit their partner, or is it spread across everything, including things they wanted?
  2. Who gets kept faith with? Are promises kept to people with power over them and broken to people who depend on them? Or does the unreliability run in every direction?
  3. What happens to the record? When a lapse is pointed out, is there repair, embarrassment, or a dispute about whether it happened?
  4. Where does the cost move? Do the consequences stay with the person who failed, or do they migrate onto the partner? In the opening illustration, one person paid for the booking and then comforted the partner.
  5. What is happening to the partner’s world? Has the number of people they could tell things to gone down? If so, through what acts or circumstances?

These questions describe conduct. They do not reliably point to one trait over another; the same answers fit several temperaments and several non-trait explanations.

Three dimensions

Narcissism. Mitja Back and colleagues model grandiose narcissism as two related strategies. Admiration seeks standing through charm, confidence, and self-promotion. Rivalry protects standing by devaluing others and competing with them, a partner included. A separate literature treats vulnerable narcissism, marked by insecurity and defensiveness.

Machiavellianism is associated with a cynical view of others, attention to reputation, and strategic, planned manipulation for one’s own ends.

Psychopathy, in the triarchic model of Christopher Patrick and colleagues, combines boldness, meanness, and disinhibition. Not every questionnaire measures all three. Daniel Jones and Delroy Paulhus argued that impulsivity is one of the clearest lines between psychopathy and Machiavellianism. The line is contested: some researchers find the two overlap so heavily that they treat them as a single “dark dyad,” with narcissism apart.

These are dimensions, not characters. The same person can score moderately on all three. A partner who keeps promises to everyone except the person who depends on them presents a different pattern from one who keeps promises to no one. The selective pattern says more about what the failures may be for; the indiscriminate one can do as much damage, or more. Which warrants more concern depends on what the failures cost and who pays.

IV · What the temperaments reach for

Section II asked when the Dark Triad lens tells us something the other explanations don’t. The research offers associations and some hypotheses worth testing. It does not show that adding a trait hypothesis improves anyone’s practical decision.

One caveat covers most of what follows. Nearly all of these studies rely on self-report, and most measure people at a single point in time. They describe associations across groups. They cannot identify the motive behind one partner’s behaviour.

Admiration and rivalry

Stefanie Wurst and colleagues ran seven studies on grandiose narcissism in romantic relationships, keeping admiration and rivalry separate. Admiration was primarily associated with short-term romantic appeal. Rivalry was primarily associated with long-term romantic problems, among them less willingness to forgive, more arguments, and a more critical view of the partner.

The popular story runs from early charm to later decline. Two analyses of the same German panel study, pairfam, test parts of it, and they ask different questions. Elyakim Kislev (2023) found that increases in a person’s own rivalry over two years went with declines in their relationship satisfaction. Gwendolyn Seidman and William Chopik (2026) asked whether narcissism measured at the outset predicted how satisfaction changed over six years in 5,869 couples. Rivalry was associated with lower satisfaction in the full sample, though not among new relationships. The main analyses did not find that baseline narcissism predicted changes in satisfaction over time.

The rivalry association holds. Neither study shows narcissism steepening a decline from an early high, and neither measures cruelty.

Tactics that work through dependence

Peter Jonason, Norman Li, and David Buss linked the Dark Triad to self-reported tactics for keeping a partner, among them monitoring a partner and monopolizing their time. The survey recorded which tactics people reported using, not what those tactics did. This essay’s reading is narrower: the tactics that matter here are those that increase a partner’s dependence, such as cutting them off from others or making departure costly. Whether a given tactic does that depends on its effects in a particular relationship, which the survey did not measure.

Deliberately inducing jealousy is a related tactic. Karlijn Massar and colleagues surveyed 347 respondents: primary psychopathy was associated with reported jealousy induction and with control and revenge motives, while secondary psychopathy was associated with several motives, including testing the relationship, control, and self-esteem. Gregory Tortoriello and colleagues surveyed 237 undergraduates: both grandiose and vulnerable narcissism were associated with reported jealousy induction, grandiose narcissism with power and control motives, and vulnerable narcissism with a broader range of motives.

Infidelity and concealment

Daniel Jones and Dana Weiser found that all three traits correlated with self-reported infidelity, with different patterns for women and men. An affair is a betrayal, and a betrayal is not automatically manipulation.

Concealment is a separate question. The Grammar distinguishes deception, which falsifies the inputs to judgment, from manipulation, which turns sound judgment against itself. Concealment can do either. It works as manipulation when it controls what a partner is allowed to see, so that careful reasoning about a curated record returns the wrong answer. Not every omission does either.

Aggression and control

Aurélie Claing and colleagues (2025) surveyed 1,418 Canadian men entering treatment for intimate partner violence, measuring coercive control alongside physical, psychological, and sexual violence. They reported sequential indirect associations between cumulative childhood trauma and psychological violence and coercive control, running through the three traits and through affective dissonance, meaning emotional responses that conflict with other people’s. The data were self-reported, collected once, and drawn from men seeking treatment, so the study cannot establish that one factor caused the next. A questionnaire measure of coercive control may also miss the cumulative entrapment Stark describes.

The appendix, Mapping to the Grammar, sets each of these patterns beside its competing explanations and the conditions under which it would count as manipulation.

What the lens adds

  • The measures allow tendencies to be compared across large samples. Well supported. The admiration–rivalry distinction is one product.
  • Trait patterns can suggest where to look. Plausible, but not tested as a guide to an individual relationship.
  • A trait hypothesis improves a person’s decision. Not shown by these studies.

A trait verdict is not a prerequisite for acting. Seeking support, changing expectations, assessing danger, and leaving each need their own information, and none of them waits on a diagnosis.

V · The one who stays

The screenshot is still on the phone. It shows an apartment with a second bedroom and a south window, at a rent that no longer exists, under a message that says what about this one? It records what was said and what was listed. It does not record what anyone intended, whether the rent was ever within reach, or what the suggestion was for. It can come to stand for a story, one in which the apartment was always coming.

That leaves the person inside with a task the records cannot do. Each lapse needs an account, and each account has to fit with the last one. Some of that accounting is the pursuit of accuracy. Some is preparation to leave. Some is a way of managing uncertainty that has no end in sight.

It can also turn into something else. You want to understand, which is reasonable, and understanding needs one more fact, which is also reasonable. The fact comes with an explanation that nearly fits, so you adjust, and the adjustment raises a new doubt that also needs understanding. Meanwhile, the decision you meant to make once you understood moves back a week.

The rent is due regardless.

At that point the account can start to protect the relationship instead of describing it. Nobody has to decide this for it to happen.

The Grammar of Manipulation calls shame “the second stage of the attack,” the feeling that keeps a person silent and therefore alone. That phrasing bundles several things:

  • Deliberately induced shame. A partner answers every complaint by listing the complainer’s failures.
  • Opportunistically exploited shame. Someone notices that embarrassment keeps their partner quiet, and acts on it: saving the worst remarks for company the partner won’t contradict, or reminding them what people would think if they knew.
  • Shame with no orchestration. It arises from a person’s own standards, or from a culture that asks victims why they didn’t leave.

The first two are conduct, and both can reasonably be called an attack. The third is not, though its effect can be identical. Survivor-support services report shame as a common barrier to disclosure. That supports a narrower claim than the Grammar‘s: shame can keep a person silent whether or not anyone arranged it.

The Grammar also offers a count in place of a verdict. Who could this be told to now, and who a year ago? A smaller number is a reason to ask what happened to the missing people. It is not a finding of coercion, and an unchanged number can hide trust or access that has quietly thinned.

A record, an explanation, and a basis for action are three different things. The record is partial. The explanation may come later, from what someone says or demands, or not at all. The basis for action does not have to wait for it.

VI · The trouble with the word

“Narcissist” now does three jobs at once.

  • A trait score is a position on a research scale. It describes a tendency across people and says little about any single relationship.
  • A clinical diagnosis, such as narcissistic personality disorder, requires a qualified clinician’s assessment of enduring, pervasive patterns associated with clinically significant distress or impairment. It cannot be made from across a kitchen table.
  • An assessment of conduct says what someone did, how often, and at whose expense. It is the only one of the three a partner is well placed to make.

Popular usage takes the authority of the second, the vocabulary of the first, and the evidence of the third. The result is a verdict that sounds clinical and claims more than the testimony behind it can establish.

A word can still do real work, and the work isn’t always clinical. Said to a friend, it can make years of experience communicable in a sentence. Typed into a search bar, it can turn up other people describing the same pattern, and that company can interrupt self-blame. Said to oneself, it can end the habit of explaining the other person’s conduct on their behalf. None of these uses depends on the word being clinically accurate. The trouble starts when it is asked to settle something it cannot.

At that point it can change the question. A complaint about conduct asks whether something happened and what it cost; the person who lived through it can describe that, with or without documentation. A diagnosis concerns enduring patterns across a person’s life, assessed by a clinician, and a partner is not placed to establish it. When addressing the conduct is made to depend on settling the diagnosis, the discussion can stall over a question that never needed settling first. The other person can dispute the label without answering for what they did.

The defensible sentence is usually the plainer one: this is what happened, repeatedly, and this is what it cost. It needs no diagnosis, and no diagnosis would make it truer.

There is a temptation here that this essay shares with its reader. Having set aside the popular word, it is easy to want a more careful route to the same verdict, as if temperament, conduct, and the room were a longer road to calling someone a narcissist. They are not. The method leads toward an account of conduct and a map of the options that remain usable. Diagnosis is not where it is going.

The limit of the method

Two boundaries remain. Recognition is not freedom: naming a pattern does not produce money, housing, or a safe place to speak. Safeguards count only if they can be used: a support line nobody knows about protects no one. If this essay leaves its reader feeling able to identify a certain kind of person, it has failed at what it set out to do.

VII · What remains available

The key is still on the hook by the door.

The Grammar of Manipulation sets out counter-families that protect judgment at the level of institutions, and The Hands on the Frame maps them against the primitives row by row. What follows adapts four of those families to a household: Time, Records, Voice, and Exit. The adaptation is this essay’s own; nothing in the institutional mapping validates it for intimate life.

  1. Time. Is there room to decide, or is every choice demanded tonight? A week can allow time to check information, speak to someone privately, and weigh the options that exist.
  2. Records. Where it is safe to do so, is there an account of what happened, kept where the other person cannot reach it? Phones, shared accounts, and cloud backups are often less private than they seem; sometimes a trusted person or a support worker is the safer keeper. Records can help if conduct is later disputed. They are never a prerequisite for asking for help. A missing record does not mean nothing happened.
  3. Voice. Is there someone outside the room to speak to: a friend, a counsellor, a specialist service? Voice means independent support, not confrontation. Confronting a controlling partner can raise the danger, and none of this is advice to do it.
  4. Exit. Are there options that are usable in practice: money in one’s own name, somewhere to stay, papers in one’s own hands? Stark’s argument was that coercive control is at bottom an attack on this kind of everyday freedom. Leaving can also be a period of heightened risk, which is a reason for specialist safety planning rather than improvisation.

Recognizing a pattern supplies none of these. Conduct points toward what can be changed; categories mostly do not.

The key on the hook grants access. Whether the person inside keeps one of their own is the question that matters.

Support

  • VictimLinkBC: 24-hour, multilingual support and safety-planning referrals across British Columbia. Call or text 1-800-563-0808.

Appendix · Mapping to the Grammar

The mappings below are this essay’s interpretation, not findings from the studies in Section IV, and the Grammar itself does not map most of these situations. Each applies only when its condition holds. Appeal alone is not manipulation, warmth or specificity alone is not a primitive, and disagreeing about what happened is not destabilization.

Reported or documented conductCompeting explanationsWhat remains unresolvedProposed Grammar mapping
Specific, warm promises early onOrdinary courtship; optimism; poor forecasting; admiration-driven self-presentationWhether later conduct matched, and for whom; whether the promises were used to secure present behaviour5 (commitment capture), if promises about a shared future are used to bind present behaviour, which is the Grammar‘s own mapping for future faking; 10 (attachment exploitation), if intimacy outpaces shared history and is used to gain access
Devaluation of the person who points out a lapseDefensiveness; shame; stress; rivalry defending standingWhether it is persistent and aimed at the partner’s memory2 (epistemic destabilization), if it undermines the partner’s trust in their own memory
Outside contacts steadily discouragedJealousy; genuine conflict with those contacts; retention through dependenceWhether the discouragement restricts or merely objects11 (isolation and dependency), if it reduces access to reality-testing or exit
A third party repeatedly made salientA real friendship; carelessness; deliberate jealousy inductionWhether the partner’s standing is made conditional9 (belonging pressure), if belonging is made conditional on compliance
An affair and its concealmentMany motives for the affair; for the concealment, fear of consequences, shame, or wanting to keep both relationshipsWhat conduct, beyond not telling, kept the partner from finding outNo primitive established by the affair or concealment alone. See the note below the table.
Fewer people the partner can talk toA move; falling-outs; exhaustion; discouragement by the partnerWhich acts, by whom, produced the narrowing11, only if traceable to imposed acts

Note on the affair row. Conduct used to maintain a concealment may meet a primitive even though the concealment alone does not:

  • 1 (frame control): attention is steered to a substitute question or comparison.
  • 2 (epistemic destabilization): the partner’s confidence in their own perception or memory is persistently undermined.
  • 13 (staged demonstration): genuine observations or successful checks are deliberately selected or arranged to support a misleading inference. Fabricated evidence is deception, not this primitive.

Sources and further reading

  • American Psychiatric Association. (n.d.). “Personality disorders.” psychiatry.org
  • Back, M. D., Küfner, A. C. P., Dufner, M., Gerlach, T. M., Rauthmann, J. F., and Denissen, J. J. A. (2013). “Narcissistic admiration and rivalry: Disentangling the bright and dark sides of narcissism.” Journal of Personality and Social Psychology, 105(6), 1013–1037. DOI
  • Claing, A., Brassard, A., Dugal, C., Savard, C., Daspe, M.-È., Péloquin, K., and Godbout, N. (2025). “Dark Triad personality traits and empathy as explanatory factors in pathways from cumulative childhood trauma to intimate partner violence perpetration.” Journal of Aggression, Maltreatment & Trauma, 34(12), 1911–1932. DOI
  • Jonason, P. K., Li, N. P., and Buss, D. M. (2010). “The costs and benefits of the Dark Triad: Implications for mate poaching and mate retention tactics.” Personality and Individual Differences, 48(4), 373–378. DOI
  • Jones, D. N., and Paulhus, D. L. (2011). “The role of impulsivity in the Dark Triad of personality.” Personality and Individual Differences, 51, 679–682. DOI · Author-hosted PDF
  • Jones, D. N., and Weiser, D. A. (2014). “Differential infidelity patterns among the Dark Triad.” Personality and Individual Differences. DOI · UTEP ScholarWorks
  • Kislev, E. (2023). “The longitudinal effect of narcissistic admiration and rivalry traits on relationship satisfaction.” Social Psychological and Personality Science, 14(7), 865–874. DOI
  • Massar, K., Winters, C., Lenz, S., and Jonason, P. K. (2017). “Green-eyed snakes: The associations between psychopathy, jealousy, and jealousy induction.” Personality and Individual Differences, 115, 164–168. Western Sydney University
  • Patrick, C. J., Fowles, D. C., and Krueger, R. F. (2009). “Triarchic conceptualization of psychopathy: Developmental origins of disinhibition, boldness, and meanness.” Development and Psychopathology, 21(3), 913–938. DOI
  • Paulhus, D. L., and Williams, K. M. (2002). “The Dark Triad of personality: Narcissism, Machiavellianism, and psychopathy.” Journal of Research in Personality, 36, 556–563. Author-hosted PDF
  • Rogoza, R., and Cieciuch, J. (2020). “Dark Triad traits and their structure: An empirical approach.” Current Psychology, 39, 1287–1302. DOI
  • Seidman, G., and Chopik, W. J. (2026). “From spark to strain? Changes in relationship satisfaction as a function of narcissistic admiration and rivalry.” Journal of Personality. Advance online publication. DOI
  • Stark, E. (2007). Coercive Control: How Men Entrap Women in Personal Life. Oxford University Press.
  • The Hotline. (n.d.). “Why didn’t you say anything?” thehotline.org
  • The Hotline. (n.d.). “Building your case: How to document abuse.” thehotline.org
  • Tortoriello, G. K., Hart, W., Richardson, K., and Tullett, A. M. (2017). “Do narcissists try to make romantic partners jealous on purpose? An examination of motives for deliberate jealousy-induction among subtypes of narcissism.” Personality and Individual Differences, 114, 10–15. DOI
  • Wurst, S. N., Gerlach, T. M., Dufner, M., Rauthmann, J. F., Grosz, M. P., Küfner, A. C. P., Denissen, J. J. A., and Back, M. D. (2017). “Narcissism and romantic relationships: The differential impact of narcissistic admiration and rivalry.” Journal of Personality and Social Psychology, 112(2), 280–306. DOI

Further reading

  • Machado, C. M. (2019). In the Dream House. Graywolf Press. Testimony about one experience of psychological abuse, not evidence about personality traits.

In this series: The Grammar of Manipulation · The Hands on the Frame