No topic is more frequently debated with less resolution than hate speech. This is made abundantly clear in a new collection of essays written by some of the leading contributors to the debate. The volume is called "The Content and Context of Hate Speech: Rethinking Regulation and Responses," and it is edited by Michael Herz and Peter Molnar.

What you learn in the course of reading this book is that there is no generally accepted account of (1) what hate speech is, (2) what it does (what its effects are) and (3) what, if anything, should be done about it. To be sure, everyone agrees that it is hate speech when words are used to directly incite violence against a specific person or group of persons. But as Arthur Jacobson and Bernhard Schlink point out in their contribution to the volume, on such occasions the words are instrumental "to an incipient assault," and it is the assault, not the words, that the state criminalizes. (It is, say the courts, "speech brigaded with action.") There need be no debate about what to do in the face of that kind of speech because it is already being done by extant laws.

The rest are all hard cases. Is it hate speech when, in a paper with scholarly trappings, someone says that the Holocaust never happened and was invented by Jews in an effort to induce guilt and gain money? Is it hate speech when a pamphlet explains how Muslim Americans plan to impose Shariah law and subvert the traditions of this country? Some who consume such statements will certainly feel hatred for Jews and Muslims, and no doubt those who make such statements intended that result. But no call to violence is issued and one might say -- in the United States it will always be said -- that while it is hate speech disguised, the disguise is good enough to remove it as a candidate for regulation.

Then there is what we might call genteel hate speech -- casually produced anti-Semitic and racial slurs in conversation and in countless British novels. It is certainly hateful speech, but it reflects less the intention of the speaker or writer than the cultural background of the society he lives in. If it is hate speech, it is so distant from any specific design to wound that a legal remedy against it seems quixotic and unenforceable. And yet, demeaning speech that flies under the legal radar because it is an extension of what people regularly say and even more regularly think may in the end be more harmful than the direct, frontal insult.

And what about hate speech that is never uttered, but is implied by the structure of institutions? When the law in many states reduced black Americans to the status of property, wasn't it being said, in the most forceful way imaginable, that blacks were more like animals than humans? When women were denied the vote for so many decades, wasn't it being said that they were perpetual children and unworthy of an independent existence? And, to reference an example invoked by Peter Molnar in his essay, isn't the statue of Teddy Roosevelt in front of the American Museum of National History that shows him on a horse flanked by the figures of a Native American and an African-American a form of hate speech declaring the natural subservience of the "red" man and the black man?

Let's suppose that we could sort through these versions of hate speech (and there are many more) and come up with a baseline definition of what it is; we would still have the problem of specifying its effects. You can't cogently debate whether to regulate something unless you have first identified the harms it produces. Bhikhu Parekh, political philosopher and a member of the British House of Lords, is quite confident in his account of those harms. Hate speech, he says, "lowers the tone of public debate, coarsens the community's moral sensibility, and weakens the culture of mutual respect that lies at the heart of a good society." In addition, hate speech "violates the dignity of the members of the target group" who lead "ghettoized and isolated lives with a knock-down effect on their children's education and career choices."

Not necessarily, says Nadine Strossen, a professor of law and a past president of the A.C.L.U. We are not, she insists, "automatically diminished just because some bigot says something negative about us." Indeed, we are better off knowing about the hateful things being said, first because it provides "valuable information," second because it gives the targeted individuals "an opportunity to respond" and third because it "highlights issues that can be addressed in other ways, for example through education."

Behind Strossen's and Parekh's assertions are two very different views of human beings. For Strossen, the hate speech recipient is (or can and should be) a resolute individual standing up for herself in the face of verbal assault and emerging stronger from the encounter. For Parekh, the hate speech recipient is the vulnerable victim of forces that rob her of dignity and deprive her of the resources necessary for human flourishing, resources that therefore must be supplied by the state in the form of hate speech laws. Strossen's view follows from a strongly libertarian reading of the First Amendment and is pretty much official doctrine in the United States. Parekh's view reflects the more communitarian concerns of other Western democracies that balance free expression rights against the right of the society to order itself in decent and humane ways.

Michel Rosenfeld draws the relevant contrast with respect to the United States and Canada: "Under the American view, there seems to be a greater likelihood of harm from suppression of hate speech than from its toleration." But from a Canadian perspective, "dissemination of hate propaganda seems more dangerous than its suppression as it is seen as likely to produce enduring injuries to self-worth and to undermine social cohesion in the long run."

Rosenfeld observes that the two countries "differ in their practical assessments of the consequences of tolerating hate speech." Not quite; what the two countries differ in is their respective assumptions concerning what must be protected: on the one hand, a rights-based individualism that can take care of itself and would be diminished by nanny-state intervention, on the other, a psychological and societal fragility that must be shored up by law. In one vision, hate speech is an opportunity; in the other it is a virus. Given such two different accounts of the effects of hate speech, it is not surprising that there would be two different accounts of what to do about it, or what not to do about it, and no hope of reconciling them.

Taking note of differences like these, some contributors to the volume propose a contextual analysis sensitive to cultural and political circumstances. They say that in a country like ours where there are strongly established democratic institutions, hate speech (short of incitement to violence) can be pretty much tolerated in the confidence that no drastic, long-term consequences will ensue. But in countries where democratic institutions are fragile or just emerging, and where tribal, sectarian conflict has been the order of the day for a long time, the state may have to move against forms of speech that threaten to rend the fabric of society. (A bit paternalistic, don't you think?)

One must also, it is said, take into account the recent history of the nation. In Germany, Parekh observes, banning Holocaust denial is "part of reparative justice, a public statement of the country's acknowledgment of and apology for its past, a way of fighting neo-Nazi trends in German society." In France, Julie Suk asserts, criminalizing Holocaust denial is not negatively directed at hate-speakers, but positively directed at publicly establishing the moral legitimacy of the state. Whereas in America it is thought "that the state would be undermining its democratic legitimacy were it to discriminate against certain viewpoints" -- an argument made by Ronald Dworkin in his essay -- in France the state wants to make it clear that "racism has no place in collective self-determination." Thus "the fight against racism ... becomes a collective national project" in which the state plays a major role.

The contextual approach - regulation for this society, but not for that - will be resisted by those who insist that it's a matter of principle: it's just wrong to suppress and/or punish viewpoints with which the state disagrees, and it is wrong, indeed immoral, in every place and at every time. In response, pragmatic-minded commentators will say that the problem is not a moral but a managerial one. You've got to look at what you have, where you've been and where you want to go, and figure out the best means to get there. Regulating hate speech may be one of those means, and it may not be, depending on the facts of your particular situation. Why commit yourself to either course in advance? Why not allow yourself the flexibility to devise strategies that will work on the ground?

These questions will always circle back to the original -- and unanswerable -- question of what is hate speech and what does it do anyway. It is unanswerable because hate speech is a category without a stable content. As Strossen says, "One person's hate speech is another person's deeply held religious belief." The response of Enlightenment liberalism to this difference -- to the depth and intractability of substantive disagreement -- is to seek a common, usually procedural, ground on the basis of which lines can be drawn without putting the state on the side of anyone's viewpoint. If we can all just agree on a minimalist definition of hate speech and its harms, we might get somewhere and bring everyone along. But every effort in this direction founders on the fact that you can't be minimal enough. No matter how low the bar is set, some will feel, and feel with reason, that it has been set to exclude them.

Alon Harel, for example, proposes in his essay to draw the line demarcating protected from unprotected hate speech by according protection to arguably hurtful speech if it is part and parcel of "a comprehensive and valuable form of life"; if, that is, it is "deeply rooted" in "long term customs, ways of life, and ideological commitments." There is, however, a caveat, already more than implied in the word "valuable": the tradition to which the speech is attached must include "prominent humanistic components," a requirement that would exclude "Nazism and the Ku Klux Klan," for they are "hatred based traditions." But who is to say that Nazism and the Ku Klux Klan are all hatred and have no humanistic components or ideological commitments? Surely members of those groups would not say so, and on what basis are they peremptorily eliminated from the roster of viewpoints accorded protection?

On the basis, obviously, of the substantive preferences (which in this case I certainly share) of the one doing the line-drawing. But from the perspective of the liberalism within which Harel writes, neither the state nor the political theorist can make in-advance substantive judgments on what does and does not have a place in the marketplace of ideas. Excluding Nazis and Ku Klux Klanners and gay bashers from the zone of speech-toleration may feel good, but it rests on no principled ground, and attempts, like Harel's, to specify such a ground are merely attempts to occlude the entirely arbitrary nature of the classification.

In the end, none of the alternative ways of dealing with hate speech is entirely satisfying. Allowing it all leaves unanswered the question of what to do about the harms it causes. Banning it all reopens the question of just what it is and what it isn't. Selectively banning this but not that only reanimates the divisiveness that hate speech regulation promises to diminish. We'll be at this for a long time, going in exactly the same circles.