Imagine two poor people, P1 and P2. P1 decides to play Russian roulette once and wins a fortune, while P2 passes and stays poor. Was playing the right decision? P1 would say yes, but if we replay history many times, shifting the bullet in the revolver’s cylinder every time, P1 would die five times out of six. The odds are quite bad. P1 got rich in this universe, but is dead in five out of six possible ones. I call this being contingently right but “necessarily” wrong. Is there an inverse example, being contingently wrong but “necessarily” right? Yes: P2, who stays poor but at least is alive in all universes.
But “necessity”, meant here as the opposite of contingency, is too strong a word and could be misleading. I am not implying strict necessity but dispositional invariance: a stable tendency to behave in certain ways when conditions allow it. In that sense, P1’s decision can be thought of as robustly suboptimal across counterfactual variations. We could say P1 is dispositionally, or structurally, wrong.
Imagine now two scientists, S1 and S2. Given the ideas of their time, both are about to formulate an important theory; only S1 gets there first, not necessarily due to better skills. Decades later, no one remembers S2 and everyone deifies S1. Even worse, S1’s biography retroactively presents childhood details as signs that success was already inevitable. No one writes the biography of S2, but if it existed, everything would be interpreted through the lens of failure.
If we replayed history many times, each time with slightly different initial conditions, we could plausibly see S2 succeeding a substantial fraction of the time, though we cannot know this exactly. Perhaps the S1/S2 success ratio would be closer to 2/5, or 4/3. In our universe, S1 is a contingent genius and S2 a contingent failure, but perhaps we should examine the robustness of their merits under historical variation. We would probably conclude that neither was a genius in the sense of creating something out of nothing, and that both were geniuses in the sense of being excellent listeners and players of their time.
One cannot understand risk without counterfactual thinking, and one cannot understand history without considering other histories. As N.N. Taleb would say (paraphrasing), “always visit the cemetery”. Otherwise we get a biased picture of the world.
I want to apply this framework to abuser-victim scenarios. Consider a completely fictional example with two nations, N1 and N2. Both begin their south-directed expansion of an asymmetric peninsula that is wider on its western side. Both nations cross the neck of the peninsula simultaneously, but N1 happens to take the western part and, acquiring more land, becomes more powerful. As a consequence, N1 abuses N2, consistently marginalising it.
People from N2 resist and make victimhood the trademark of their resistance. They are, of course, right to complain, since they are being abused. But now consider what would have happened if, in similar histories, N2 had taken the western part first. In all probability, N2 would have abused N1. History would look almost identical, with the labels swapped, and N1 would be right to complain.
Returning to our initial history, with N1 as abuser and N2 as victim, we can say, after applying the counterfactual reversal test, that N1 is a contingent abuser and N2 a contingent victim. The important point, however, is that both N1 and N2 become abusers when the opportunity arises. In this sense, they are both structural abusers, regardless of how their dice have happened to roll.
How can we be confident about how counterfactual histories would develop? It is easier than it seems. Take the contingent victim nation N2 and examine how it applies power within its own territory. Does it exhibit strong centralism and use central power against peripheral regions? If so, it seems reasonable to infer that similar behaviour would emerge under larger-scale power.
Consider now two women, W1 and W2. Both have suffered the patriarchy all their lives and are right to consider themselves victims of it. But there is a crucial difference: W2 is vegan, whereas W1 is not. This means that W1 abuses non-human animals when this power is available to her, while W2, given the same opportunity, refuses to do so. Even if both are contingent victims, W1 is clearly a dispositional abuser, whereas W2 probably is not. We can be confident about the former but not the latter, although W2 earns a degree of trust as a non-abuser. If both W1 and W2 were men, it is reasonable to assume that W1 would take advantage of the patriarichy, whereas W2 would challenge it.
This example illustrates the core point. A non-abuser is an agent or group that actively refuses to abuse others when this power is available. There is no need to appeal to parallel universes: our own universe already offers countless opportunities for abuse, even to contingent victims, and these opportunities reveal true dispositional character. The counterfactual reversal test does not require literal role symmetry; it relies on dispositional evidence drawn from any context in which comparable power over others is exercised.
The counterfactual reversal test can therefore be stated in weak and strong forms:
Weak form: If an agent or group, when exercising limited or local power, exhibits patterns of domination or abuse, this provides defeasible evidence that they would also abuse under reversed global power conditions.
Strong form: An agent or group is a structural or dispositional abuser if, across available contexts in which they hold power, they systematically exploit that power, such that role reversal would predict abuse independently of contingent historical outcomes.
This test does not invalidate criticism made by contingent victims, but it weakens one side of their moral position if they themselves are abusers. Abusers have no justification for their actions; victims who abuse others undermine part of their own defence. This incoherence does not absolve the original abuser, but it does implicate the victim as well. In defending themselves, they partially accuse themselves. As dispositional abusers, they would likely behave as their abusers if positions were reversed.
It is easy to find people who abuse non-human animals but would never abuse another human; they are contingent and dispositional abusers. This case highlights the final angle I want to explore. When abusers seek to avoid being labelled as such, they degrade their victims’ status. Abuse is condemned when inflicted on equals, so abusers either lower their victims’ status or elevate their own, until victims no longer appear as victims but as objects, livestock, or ungodly creatures.
In conclusion, risk, merit, and especially abuse depend crucially on how they appear under counterfactual reversal inspection. Although this exercise may evoke multiverse or science-fiction imagery, it can applied in practice through two mechanisms. The first is the inspection of abuse within nested power structures, where conduct at smaller scales can be extrapolated to larger ones. The second is the observation of historical role reversals, examining how former victims/abusers behave once they gain/lose power.
Only those who renounce all forms of power when it is available can reasonably be trusted as dispositional non-abusers, and even this is not proof. Only sustained empirical behaviour over time counts as evidence. The rest, those who denounce colonisation after previously appropriated land at the expense of ecosystems, those who decry genocide while later committing one, those who terrorise a country for decades and then complain about the violence that follows, are no better than their claimed enemies.
As a final remark, true dispositions are revealed only through behaviour in one’s own historical context. Almost everyone alive today condemns nineteenth-century racism, yet almost no one did so at the relevant time, when abuse of slaves was widespread and profitable. There were, besides the courageous victims, only a handful of people among the dominant group who actively denounced slavery and fought it. Those individuals were true moral gems of their era, and there are many such people today as well. Do not confuse mere victimhood with genuine moral disposition, ascribed only to those who pass the counterfactual reversal test.