Hegel famously observed that the owl of Minerva takes flight only at dusk. Among Hegel’s dicta this one stands out as being not just importantly correct but broadly comprehensible (two qualities not often combined by the German master). Because the goddess Minerva symbolizes wisdom, Hegel was suggesting that philosophical thought properly begins only after the day is done—that is, only in reflecting on what has already occurred. Were philosophy’s goal solely to make sense of the world retrospectively, that would be good enough. But we also turn to philosophy in the hope that it might make our behavior more morally responsible, our institutions more just, our political decisions more legitimate, and so on. We seek from it not only a coherent account of the world we’re in now, but a set of principles, ideals, and guidelines that will help us make good choices for the future.

The best political theory (and this includes Hegel) brings together these two goals—the retrospective and the prospective. Avishai Margalit’s On Compromise and Rotten Compromises is in this respect exemplary. Margalit wants to clarify issues in political morality that have tremendous urgency today, and he seeks to do so partly by reflecting on events in our past. His book is an uncommon example of philosophical argument informed by acute historical awareness.

Margalit’s main interest is in establishing when it is and isn’t appropriate for one political community to compromise with another it deems morally deficient in some important way. Among political theorists, many of whom seek to articulate ideal principles (of justice, punishment, democracy, etc.), the topic of compromise is seriously underexplored, and Margalit’s contribution is very welcome. After all, most of our lives play out in nonideal scenarios, where we encounter others sadly lacking our own moral clarity and judgment, and moral theorists should be in the business of helping us navigate such moments intelligently. This need is especially strong in the United States today, where concerns over threats against our citizens have led many to argue that we must compromise both our own principles (by relaxing the absolute prohibition on torture, for example) and our disdain for foreign governments whose policies we deem unjust (by allowing them to aid us in fighting our enemies).

When should we compromise, then, and when must we not? For Margalit, the main reason to compromise, even when it involves accepting injustice, is to achieve peace. Too often the sectarian’s staunch refusal to compromise his principles impedes the achievement of lasting peace and allows instead only a temporary truce, a cessation in hostilities likely to erupt again when either side senses an advantage. The costs to human welfare of such intransigence may not be dramatically visible when the issue is, say, tax cuts for the wealthy, but in other scenarios—think of the endless cycle of violence in the Middle East, for example—they are profound. Here the virtues of compromise are unmistakable.

When he turns to the compromises we must never make, Margalit’s position is both more provocative and, I think, less convincing. His central claim concerns “rotten” compromises, those that “establish or maintain regimes of cruelty and humiliation” (the latter he dubs “inhuman” regimes). Margalit states his thesis strongly: “Rotten compromise must be avoided, come what may.”

This simple-sounding thesis, however, hides multiple ambiguities. To begin with, Margalit notes that not all inhuman regimes are equally bad. Some, like Hitler’s Germany, exclude certain human beings from moral consideration entirely and thereby attack morality itself. Others, like Stalin’s Soviet Union, may be complicit in cruelty and humiliation without similarly violating the principle of shared humanity, and so fall short of the “radical evil” Nazism presented. The qualified evil of Stalin’s regime makes compromising with it, Margalit suggests, more acceptable than compromising with Hitler would have been. Readers may wonder whether this distinction can bear such weight. Is it so much worse to kill an innocent Jew because one thinks Jews don’t deserve moral consideration than to kill an innocent bourgeois because one thinks he stands in the way of revolution?

A second question arises from the fact that Margalit’s central thesis does not prohibit making deals with inhuman regimes; it just prohibits agreements that establish or maintain them. But what does “maintaining” mean here? As Margalit himself notes, a respected country’s decision to conclude a treaty with some renegade group or foreign power, even to negotiate with it at all, confers legitimacy on the latter. Does this help maintain it? If so, the “establish or maintain” clause is misleading, and the thesis is simply “Don’t make agreements with inhuman regimes.” But that can’t be right, for Margalit thinks that Stalin’s Soviet Union was an inhuman regime we were nonetheless right to align with to defeat Hitler. Perhaps, then, his thesis is “Don’t make agreements with inhuman regimes unless doing so helps combat even worse ones.”

Comparisons between Nazi Germany and Stalin’s Soviet state point to a general concern about Margalit’s overall strategy here. The book deserves high praise for largely bypassing the thought experiments philosophers favor and instead relying on careful, subtle discussion of actual historical events. Those events mostly come from World War II, which Margalit presents as an important laboratory from which to derive more general moral principles. (The war is to morality, he says, what the supercollider is to physics.)

But I wonder how far the extraordinary, perhaps unparalleled features of that experience—including the sui generis nature of Nazism itself—can yield general moral lessons to help us face our current challenges. Questions of how to deal with China or Iran, with Hamas or moderates in the Taliban, require a far more complicated moral response than did the stark confrontation with Nazism. The supreme importance of defeating Hitler gave us, we might say, a kind of moral permission slip to do virtually anything, to make virtually any agreements, that might advance that goal (even if it did not, as Margalit sharply notes, justify the craven agreement on the part of the allies to repatriate Soviet soldiers after the war, where most met grisly ends). Without that permission, we must instead fall back on careful attention to details and particulars, doing our best to weigh the petty grievances, claims for mutual respect, moral shortcomings, and a thousand vexing factors that complicate the pursuit of lasting peace in any serious conflict. In the end, it is in providing a model of how such nuanced deliberation should proceed that the value of this deeply humane book lies.