The Dice Person: Inconsistency, Toxicity and Restraint
Some people seem to answer the same remark with a different sentiment each time, as if they throw a die to decide how to respond. I present several ways to classify people by their types of reaction and their types of inconsistency.
The dice roll
You share an interest with someone. You ask: “Did you see that update?” One time the answer is long and enthusiastic. Another time it is “who cares,” with a look that says the question was stupid. Same remark, same person, and nothing you can see explains the difference. From where you stand, it feels as if they throw a die to decide how to respond.
Two kinds of toxicity
The remark. A cutting answer to a plain question is toxic on its own. Dutch has a word for it: a kutopmerking, a remark that is unnecessary and hurtful by definition. Which remark qualifies can be argued. What the word means cannot. Not all negativity is unnecessary: “why do you keep asking?” is earned the third time the same question comes, and so is a sharp answer to a question that opened with an insult.
The pattern. The second kind sits in the pattern, not in any single remark. It can only be spotted when the swings happen repeatedly and structurally: one cutting answer after a bad night is a bad day, not a pattern. And even once the pattern is clear, the next reaction stays hard to predict. Knowing that someone is a die does not tell you the next throw.
The pattern does its damage through the warm throws too. When the person does respond with enthusiasm, two things arrive at once: the conversation, and relief that the hurtful version did not show up. That second part is the relief effect. It makes a good outcome from an inconsistent person feel bigger than the same warmth from a consistent one, and the more the bad outcome was dreaded, the bigger it feels.
The milder die
Some people never hurt, but swing between enthusiasm and flatness: a playful deflection, minimal engagement, distance even when physically together. No single reaction is hostile, so the toxicity of the remark is absent. The toxicity of the pattern is not. Not knowing which version will show up, and bracing for it every time, costs more than it seems (unnecessary unpredictability). It weighs most when nothing comes with the flat answer: no “sorry, I’m tired,” nothing that turns the throw into information.
Three people
- A is always pleasant.
- B is enthusiastic 80% of the time and hurtful 20% of the time.
- C is always negative.
Assume each of them can be avoided or reshaped. Reshaping is not ending: you can play padel with someone, skip the conversation and go home.
Whether B or C is better to have around stays open here. What is easy to miss is how much harm B does. C’s negativity is known in advance, so contact can be shaped around it. B’s 20% cannot be planned around, and the 80%, with the relief effect on top, pulls you back every time. B may well do more harm than C.
The relief effect also distorts how the three get ranked. B can feel like the most rewarding of the three, and be the most popular, while A, the steadiest, is taken for granted as the least exciting.
The claim
Predictability is worth more than the dopamine system signals. An inconsistent person is a larger risk than the average of their reactions suggests, and intermittent warmth makes a relationship feel more rewarding than it is.
Two claims in one:
- Undervalued risk. The cost of an inconsistent person is not the average of the throws. A bad throw cannot be planned around, so no adaptation reduces it.
- Overvalued reward. A good throw carries relief on top of enjoyment. The feeling measures the contrast with what was dreaded, not the quality of the relationship.
The variables
In the order they occur:
- Unnecessary negativity. The tendency to add negativity the interaction did not call for. The source of the toxicity of the remark.
- Inconsistency. How much a person’s response to the same approach varies. The source of the toxicity of the pattern. It comes in two kinds:
- Engagement volatility: the swing runs from enthusiastic to flat or distant, and stops at neutral.
- Emotional volatility: the swing crosses into unnecessary negativity.
- Restraint. How much of the first two reaches the surface. It rests on awareness (noticing the urge), equanimity (letting the urge exist without being pushed by it) and the choice of what to do anyway. Two people can behave identically while one of them spends far more effort doing so. That effort deserves respect.
- Predictability. The same behavior seen from the other side of the conversation: how well the person approaching can anticipate the response. A cause the observer cannot see, such as bad sleep or stress, makes a person look like a die. Finding the cause raises predictability. It does not make an unnecessary remark necessary: an explanation is not a justification.
Unnecessary negativity and inconsistency are potentials. Restraint decides how much of them becomes behavior, and predictability is how that behavior lands on the other person.
Six types
The model labels people who carry one or both potentials. Engagement volatility makes a dice person (coined), emotional volatility a toxic dice person (coined).
| Unrestrained | Restrained | |
|---|---|---|
| Dice person | Every approach is a throw: enthusiastic, flat or distant. | The swings exist but rarely surface. |
| Toxic person | Regular unnecessary negativity, known in advance. | The urge to make the remark is there. The remark mostly is not. |
| Toxic dice person | Warm one time, hurtful the next, and the relief effect pulls you back. | The volatile, hurtful potential stays below the surface. |
Someone with neither potential gets no label: restraint has nothing to work on.
Scope
- Avoidable relationships only. The model assumes the relationship can be avoided or reshaped: friends, colleagues, acquaintances. A child cannot leave a parent. A cat that nurses abandoned baby squirrels next to her own kittens is touching to watch, and the squirrels are entirely at the mercy of a cat’s whims. Had she ignored, abandoned or eaten them, that would have been cruel, but expected. Where leaving is not an option, the same pattern does more damage, and the three-people comparison no longer holds.
- Controversial in the open. Applied to named people, the labels sting. Their use is to name a pattern that is under-discussed and rarely recognized.
Related research
- Intermittent reinforcement (Ferster and Skinner, 1957): rewards on an unpredictable schedule sustain behavior longer than steady rewards do.
- Reward prediction error (Schultz, Dayan and Montague, 1997): dopamine neurons respond to a reward that beats expectation, not to the reward as such.
- Relief as a reward (Leknes et al., 2011): the omission of an expected painful event is experienced as pleasant, and the more it was dreaded, the more pleasant the relief. The closest lab analogue of the relief effect.
- Ambivalent friendships (Holt-Lunstad and Clark, 2014): participants who brought a friend who was both helpful and upsetting showed higher heart rate reactivity and anxiety during a stress task than participants who brought a supportive friend.
- Traumatic bonding (Dutton and Painter, 1993): strong attachment that grows under intermittent abuse plus a power imbalance. The toxic dice person carries the intermittency without the imbalance.
- Anxious attachment (Ainsworth et al., 1978): associated with inconsistent caregiving, the case in which leaving is impossible.
Limitations
- The builder’s bias. The model is built from the side that pays a high price for unpredictability. Someone who enjoys spontaneity and handles emotional uncertainty well may find a dice person worth the throws, and that enjoyment is not necessarily a dopamine illusion.
- Restraint is invisible. From the outside, strong restraint and no potential at all look the same. The six types therefore sort one’s own patterns better than other people’s.
- Research status. The mechanisms above are documented. That an inconsistent person can do more harm than a consistently negative one, when both can be avoided, is this model’s claim, not a finding.