1 I

Reflections on Crime and Culpability is the sequel to Crime and Culpability and stands somewhat in relation to the first as the second volume of Schopenhauer’s World as Will and Representation stands to the first. The first volume develops a general schema for looking at the criminal law and the second fills out that schema by dealing with all the complications it gives rise to. They should thus ideally be read together, as one work, at least to get the maximum out of them, and that maximum is well worth extracting. That may well look like a daunting undertaking, and indeed it is, because although in sheer physical bulk these are ordinary-size books, they are extraordinarily dense with argument. There isn’t really an issue in substantive criminal law on which something new and surprising isn’t said. But none of this yet gets to the heart of the reason why such an investment of effort is worth one’s while.

What the two books seek to do is to revise our system of criminal law top to bottom. It would be a mistake, however, to think that they are therefore of interest principally to would-be reformers. To us at least, being of a more Burkean, tradition-respecting cast of mind than the authors, its interest lies mostly in the purely intellectual realm, for the unexpectedly bright light it sheds on the existing criminal law and for the novel approach to doing legal theory more generally that it suggests.

So what exactly is it that they propose by way of not just rethinking but redoing criminal law? They start sedately enough by pointing out that their point of departure is the same as that of most criminal law theorists: moderate retributivism, although here already they add a subtlety to that self-description that takes them beyond what criminal law scholars typically say when describing their grounding premises: punishment should correspond to desert for wrongdoing—that’s the retributivist part—but punishment requires resources and those need to be shared with competing uses—that’s the moderation part. That means, they point out, that there is actually a spectrum of possible retributivisms, depending on how that balance is struck. But while acknowledging the need and the difficulty of striking that balance, this is not where their main interests lie.

There are innumerable points in the criminal law in which criminal scholars have noted in passing that the law might perhaps have gone the other way; indeed they have sometimes gone so far as to acknowledge that common sense might suggest that the alternative not taken would be the more sensible one, but, alas, strangely enough, it’s not the turn the law has taken. Occasionally, they go so far as to peer in the direction of what the law might look like if it had taken the deviant turn. Causation is probably the most common place where this occurs: Why attach such significance as the criminal law does to whether an outcome that was intended or risked actually occurs? Is there really a difference between two actors who fire a bullet, or drive under the influence, and one produces a deadly outcome and the other does not? Interestingly, but to many somewhat inexplicably, no legal system has chosen to do that, though the Model Penal Code (MPC) has at least proposed doing so to some limited extent, without finding any takers for that part of its otherwise welcome proposals. There are, however, many more such fundamental cross-roads where the law could have taken one turn, the deviant turn we will continue to call it, but did not, and moreover has pretty consistently refused to—and when we say “the law,” we mean not just Anglo-American law, but virtually all known legal regimes. This stands in sharp contrast to less fundamental cross-roads, such as where the line between manslaughter and murder is to be drawn, whether there should be a merger doctrine, how intoxication should be treated, and exactly what should be criminalized.

What if instead of merely noting the existence of these cross-roads, and perhaps briefly peering down the road not taken and asking why, one adopted the view that if no obvious reason stood in the way, and common sense actually supported it, the deviant turn should be taken, and one did so in every instant in which the opportunity arose. This is the bold experiment Alexander and Ferzan run.

Well, to begin with, what are some of the other deviant turns one might take beyond trying to make outcomes irrelevant? The most familiar, and once again the MPC actually advocates this, is the abolition of negligence (as opposed to recklessness) as a basis for liability, and Alexander and Ferzan endorse that proposal. But that really is just the least of it. Everything else they propose has rarely been contemplated singly, let alone collectively, as a possible way to structure a criminal code.

Start with mens rea. It has struck many criminal law teachers in passing that maybe knowledge is not a mens rea the criminal law strictly requires because it is essentially subsumed under recklessness: it is recklessness with a perceived probability of 100% of harm, and inasmuch as someone who knowingly harms can escape liability if he has a good defense, such as necessity, the whole concept is really already covered by recklessness. To be sure, the person who harms knowingly and unjustifiably may be more blameworthy than someone whose probability of harm is lower, but that suggests simply that in addition to abolishing knowledge as a separate mens rea, we should do away with the uniform treatment of all instances of recklessness and grade them instead according to the ingredients that go into the seriousness of the harm, its probability, and the reasons for undertaking it. And that’s what Alexander and Ferzan propose to do.

They then extend this strategy further by pointing out that purpose can, first impressions notwithstanding, also be folded into recklessness, or at least those species of purpose we care about: typically the person who intends harm and has no defense is in fact being reckless. There is a possibility of harm on the bad side of the ledger and a disapproved reason on the justification side. Ergo, recklessness. Dropping purpose from the index of relevant mens rea states is thus another deviant turn they take. What remains is just one mens rea, recklessness, the display of insufficient concern, assessed according to the familiar formula of weighing probable harm against the reason for which it was undertaken, now graded, however, in a manner left open, according to the amount of each of those components present.

What about the actus reus? It really just comes down the creation of risk. Risk creation is now not merely the central, but the only crime. The risk part is clear—recklessness—but what about the creating part? That has to be somewhat like the current law of attempts, but only somewhat. Since purpose or knowledge are no longer required, the focus is not on any particular kind of wrongdoing, but the creation of a diffuse mix of untoward possibilities. That is one conspicuous difference from the current law of attempt. The other is how far the defendant must have progressed in his risk creation. The answer they give is: quite far. They reject all the various tests that have at times been used alternately or cumulatively, to determine whether someone has done enough, in favor of the most demanding one, the last act test. That is an unexpected turn, but not an unfamiliar one of course, and we will explore later just why they need to do this.

So much for what happens to the doctrines relating to principal liability for completed and attempted offenses. What about complicity? As they see it, there is no longer any need for the doctrine. People are liable if they create risks recklessly and that applies in group situations as well as others, period. No special doctrines are required for collective actions.

What happens to defenses? They too virtually disappear as a distinct category. Let us see why. Start with the justifications: self-defense, necessity and related ones. These now simply become reasons in the calculus of recklessness. The same actually happens to excuses such as duress. They too become reasons that justify risk creation when they apply. Insanity, immaturity, incapacity and the like do get a separate place: those who qualify are simply not subject to the criminal law in the same way that animals and plants aren’t. They lack the kind of reasoning capacity that makes judgments about behavior that lacks justifying reasons relevant.

One of the most radical moves they make is the abolition of the special part. There no longer are any crimes of either the malum in se or malum prohibitum variety. Malum prohibitum predictably has to go because it doesn’t really fit with retributivism (not that some haven’t tried to reconcile it). Malum in se doesn’t so much disappear as get folded into the calculus of recklessness. In other words, they obviously don’t want to fail to punish theft and rape and murder, but they cover them through the one and only remaining crime, that of impermissible risk creation. The criminal code, as they envision it, would still list the various interests risk to which must be taken into account, i.e. life, sexual autonomy, property etc. and anyone who acts with insufficient concern as to them is guilty of reckless risk creation.

One last doctrinal reform is important to mention here: Ignorance of law. This doctrine, too, mostly disappears, because the only really punishable conduct is of the malum in se variety, in other words, stuff that everyone can be trusted to know is impermissible, and if he doesn’t he is either to be judged the worse for not knowing it or perhaps such an extreme psychopath as to deserve to be put into the “exempt” category along with infants, dogs and trees.

Also worth mentioning: Problems of counting offenses, which current law deals with under a multiplicity of headings such as double jeopardy, merger and the like, are no longer needed either. Every act of risk creation registers and is added up for the desert-determining total.

What we have just described will of course strike many people as wild and immoderate and not something they would ever dream of signing on to. But it is well worth asking oneself why. Every individual deviant turn they take is fairly commonsensical, with at least as much going for it as against. And even if the total looks very unfamiliar, so what?

To repeat: We are actually not particularly up for implementing their scheme and putting it in place of the current one, out of nothing more than a wariness of any radical experiment, but we think that it raises compelling theoretical questions such as: (1) What are the logical properties of their novel system? (2) Why has it never been tried before? (3) What can it teach us about the system we have? (4) What can it teach us about doing legal theory?

A word about what we mean by those questions, before trying to at least partially answer them.

Regarding the first—What are the logical properties of this system? Logical properties may actually not be quite the right expression for what we are looking for. Rather it is this. The current system harbors numerous oddities, unexplained counterintuitive features—namely those that prompted the deviant turns that Alexander and Ferzan take. Are there oddities in the new regime? What relationship do they bear to the old oddities? And how do they compare? That’ something we can only do in the most cursory fashion here, but it is well worth pursuing further.

Regarding the second question—Why has this never been tried before? It seems like a striking fact that it hasn’t. On most issues on which it seems possible to hold different opinions, and neither seems clearly superior to the other, one finds that some jurisdictions go one way and some the other, just as one would expect. Not so with most of these doctrines: no legal system known to us reduces mens rea to recklessness, abolishes harm as a factor, or does without complicity, or defenses, and so on. We note this not as a criticism, but as a genuinely interesting and open question.

Regarding the third question—What can their experiment teach us about the system we have? Many years ago Robert Fogel caused a stir with a novel approach to the longstanding question in the economic history of the United States, how much railroads contributed to American economic growth in the nineteenth century. He did so by constructing a hypothetical alternative economy in which shipping was used to do the work of the railroads. He discovered that if that had been done, although America would look very different in that cities on waterways would have become far more important, economic growth would hardly have been different. Railroads were really unimportant. Constructing an alternative criminal code can be seen as essentially pursuing the same kind of approach and is thus interesting at least as much for the light it sheds on the current system as for the reforms it proposes.

Regarding the fourth question—What can it teach us about doing legal theory? The approach they have taken is one that invites generalization to other areas of law. What if there too, say in contracts or torts or property, we took a commonsensical deviant turn whenever the opportunity arose? Tax law scholars have done a little bit of that by playing around with alternative tax systems. Tort scholars have at times played around with different schemes for dealing with automobile accidents. Elsewhere though there is relatively little of that. (No doubt we are overlooking some fields whose practitioners are instantly going to bristle with annoyance. But surely it says something about the relative rarity of the approach that other examples don’t readily come to mind.)

2 II

Our main strategy for shedding light on the above questions will be to narrow our efforts to one very specific task that relates to all of them: ferreting out the counterintuitive implications of taking the seemingly commonsensical deviant turn. That there are such implications need in no way be fatal to their enterprise, but it is critical for answering all of the above questions.

Now with some of the doctrinal changes they avow, the counterintuitive cost of the deviant turn is well-understood. That is principally true of shedding the causation requirement. Michael Moore has pointed out at length how eliminating outcome luck creates an exceedingly odd asymmetry between the treatment of outcome luck and opportunity luck and circumstantial luck. That’s not in our eyes a knockout argument, and Alexander and Ferzan don’t find it even mildly disturbing. We do. But that doesn’t mean that their cure is worse than the disease. It just means that their cure has some undeniable side effects that one has to consider. Indeed this is our position with regard to all the other side effects we are about to take up. And as to some of them, there will be a legitimate question, as there might be here, whether the side effect, though disturbing at first, is in fact undesirable.

The Unleashing. “One of the most controversial implications of our central claim is that nothing can be culpable prior to culpably unleashing risks. To put this point in the language of the criminal law, we reject the culpability of so-called incomplete attempts.” (p.5, Reflections.) This is their way of telling us that they are adopting what is commonly called the last act test for criminal attempts, and since attempts—or rather their version of it, the creation of excessive risk through insufficient regard—is all that is criminalized in their regime, it is the central defining attribute of their actus reus requirement.

This is indeed a controversial-seeming position. It seems especially surprising for someone to adopt who is committed to a retributivist view of the criminal law. Hasn’t ample blameworthiness been displayed long before one has taken the last act? They offer two reasons for their insistence on this unleashing requirement. One is that the actor might yet change his mind. Far more intriguing is a second reason: the fluid boundary between outright intentions and conditional intentions and the fact that it would in many circumstances be exceedingly odd to punish someone for conditionally intending to commit a crime. Do we want to hold someone who says “your money or your life” for attempted murder? Do we want to charge someone with the crime of acquiring illegal drugs with intent to distribute, if he acquires them for his own use but would be willing, if offered a high enough price, to sell them to someone else? Those are valid and interesting worries about neglected complications of attempt law, which they have pursued at greater length and with much ingenuity elsewhere. But that doesn’t quite explain why under their schema they feel compelled to adopt the unleashing/last act rule The considerations they adduce apply equally to the current regime as they apply to theirs and have not been found compelling by advocates of the current regime. Since the rule seems such a counterintuitive encumbrance for a retributivist, seemingly inconsistent with the desire to make punishment commensurate with blameworthiness, why do they insist on embracing it?

Our surmise is that that they are right to think that they have to, but the real reason is one they only hint at. They say: “Although it seems reasonable and possible to make predictions of another party’s actions in the future, one cannot predict his own willingness to act in accord with the balance of reasons.” (p.40, Reflections.) And why not? What we suspect they meant to say is that one cannot be expected to predict and act in accordance with one’s predictions of one’s own future actions. But again, why not? Here is why; or rather we surmise that what guides their intuition is the following fact. Imagine someone who puts himself in a situation of great temptation—or perhaps just stumbles into it—one where the payoff is high, the failure to do the wrong thing (think embezzlement) is truly costly (bankruptcy) and the ostensible chance of being caught nil. The probability of yielding to temptation is great. Under the current regime, that is irrelevant unless and until the defendant actually tries to do the illegal thing, or at the very least intends to. Under the Alexander/Ferzan scheme, he would be guilty—if, that is, they did not block that outcome with the unleashing rule. So they seem right in thinking that they need this rule. However, there is a cost: the cost that drives the current regime away from the last act test. The tie between blameworthiness and punishment seems severed. To be sure, there may be a way to dodge this last bullet: Even retributivists don’t consider all types of blameworthiness to be the sort that are eligible for punishment. Villainy of the kind that populates novels, bad Samaritans owing no duty who let the easily rescued baby drown, for instance, seem plenty blameworthy but not of the type that we want to criminalize. One could argue that bad actors shy of crossing the unleashing threshold are in that category.

Complicity. One of the doctrines of the current regime to be thrown out the window is that of complicity, and the closely related doctrine of intervening voluntary actors. Both concern the responsibility one has for wrongful actions taken by others. The intervening act doctrine disappears along with all of factual and proximate causation because actual consequences no longer matter and the complicity doctrine disappears because the question Alexander and Ferzan would have us ask in every case is simply whether the defendant acted with insufficient concern vis-à-vis a certain subset of interests of others, whether that is in a group context or not being quite irrelevant.

But the problem of complicity and intervening actors then returns by the backdoor, because there is a set of intuitions that those doctrines sought to handle that the proposed schema is going to have to handle as well. Whether or not we concern ourselves with what consequences the defendant ends up bringing about, there is the sense that not all actions by others that we help precipitate, or rather, in their schema, risk precipitating, we are required to worry about. They do not, as they put it, belong on our ledger. Identifying those with exacting care and ingenuity is the focus of the longest chapter of Reflections, “Risking Other People’s Risking.” By this route, the distinction embodied in the voluntary act principle comes back to the fore, but in the course of reintroducing it, they nicely sharpen it, both by focusing the issue with a neat juxtaposition of hypotheticals—one person risks sparking a fire that the wind unexpectedly fans into a conflagration, the other does the same but it is his companion who does what the wind did—and then put their finger on what other commentators have mostly overlooked, that the appeal here has to do with the fact that in the second case the intervening actor “reduces the available options of the first. Although we may all have our options limited by chance, Al’s options are being limited by another actor’s possible decision to act wrongfully, thus imposing on him costs he has a right not to bear.”

There are many other configurations of downstream harm to consider, most of them sharing the feature that to force this particular risk to be part of the defendant’s calculus is to limit his rights in some way similar to the above case. A particularly intriguing configuration they consider is one they call “moral blackmail,” an ingenious intertwining of the trolley problem with the problem of what goes on someone’s ledger. “A runaway trolley is currently hurtling down the track, and on its current track, the trolley will harm no one. There are three possible tracks on which the trolley can travel: the track it is currently on, a track with one trapped worker, and just beyond that one, a track with five trapped workers. Mary tells you that if you do not turn the trolley from its current track, where it will cause no harm, onto the track where it will kill the one, she will turn the trolley onto the track where it will kill the five.”(Reflections, p. 27.) What makes the case so ingenious is that it is exactly halfway between the trolley scenario and the utilitarian transplant scenario with which it is usually contrasted, and our intuitions veer as they do between competing views of an optical illusion—like the outward and in the inward bulge of a Necker cube.

When all is said and done, what they have replaced the doctrines of intervening actors and complicity with is a complex and understandably incomplete taxonomy of downstream actors and a more general question to be asked about every potential downstream actor the defendant is aware of: does he deserve to be part of the calculus of recklessness that the defendant is expected to engage in. Current doctrines have thus been replaced by something that is scarcely less complex. But that need not be a drawback, because the complexity may get much closer to what is really going on and reflect much more precisely the moral distinctions we should make. But it is worth noting that what might at first appear to be a welcome simplification really is not; what it really offers is “precisification.”

If there is a problem with this treatment of complicity, it lies elsewhere, namely with the idea that the recklessness calculus can allow some risks to be, as it were, “off the books.” Why that is a problem, though, we will delay exploring till the end when we turn to the role recklessness plays in their schema.

The Special Part. A major feature of the Alexander/Ferzan schema is the abolition of the special part. The closest their model code comes to having something resembling the special part is the list of interests that might be affected: life, bodily integrity, sexual autonomy, loss or damage to property, etc. Rather than checking to see whether the defendant has committed a particular offense of the standard variety, we are to ask whether he has created an unjustified risk to some subset of those interests, somewhat in the way tort law does, but without the concern for actual harm, or a remedy measured by that harm.

What is it that gets lost by doing so, and should one be concerned about it? It is one of the virtues of their approach that yet another underappreciated feature of the current regime becomes salient. Compare homicide with property offenses. Homicide statutes have a very simple structure. Loss of life caused intentionally, knowingly, or recklessly. That’s it. We could have a property law just like that. Property loss caused intentionally, knowingly, or recklessly. Instead we have distinctions based on the manner of deprivation, whether by destruction, embezzlement, fraud, robbery, theft or a multitude of other means, with varying mens rea requirements, such as whether the defendant intends to merely borrow or keep the property for good. Defenses, too, operate quite differently in regard to each. Assault and battery and rape have yet another structure. Consent in all three contexts operates quite differently. Advance consent usually being a defense with regard to property offenses, contemporaneous consent being required where bodily invasions are concerned. What happens to these distinctions? Possibly they come back in by the back door through the definition of interests; there are hints of that in their model code; or maybe they mean to toss most of them out. Either way, it is a notable counterintuitive feature of their schema, far from fatal to it, of course, because the current regime has its own bundle of counterintuitive features, but it is a notable tradeoff.

Omissions. An extended chapter in Reflections is devoted to omissions. Why? Omissions play a special role in their schema for a reason they make explicit and another perhaps more fundamental reason they leave implicit. The explicit reason arises when someone culpably omits and a harm ensues. The omission extends over time and since every time that someone unleashes a risk counts as an offense, it becomes tricky how exactly to evaluate this accumulation of omissive unleashings. But there is a larger reason why omissions will in their schema present more of a headache than in the less granular, more coarse-grained approach of the current regime: from a granular perspective, many acts become omissions. The person who fails to brake in time and runs someone over is from their perspective more likely to be guilty of a potentially culpable omission. That makes all sorts of vexing questions about where and how such duties arise and how extensive they are both more important and more vexing than under the current regime.

Moral Uncertainty. Moral uncertainty plays a role both in the current regime and in the Alexander/Ferzan schema, but its role in the current regime is comparatively minor and straightforward, at least in comparison to the formidable role it comes to play in the Alexander/Ferzan schema. This presumably is the reason why the subject gets an entire chapter to itself and crops up again and again for cameo appearances in other chapters.

Moral uncertainty in the current regime surfaces in two ways. First, when somebody wants to be excused because he mistakenly thought something that is illegal to be legal, which is handled by the ignorance of law is no defense principle, which when applied to the abundance of malum prohibitum crimes is judged to be harsh and unjust. With regard to malum in se crimes, however, it is thought to pose not much of a problem. Second, when someone wants to do something he believes to be illegal when it isn’t. These are cases of so-called legal impossibility and for the most part not punishing someone in this kind of case feels unproblematic. (Very very broadly speaking, and neglecting quite a few qualifications.)

Now what happens to these two kinds of mistakes under Alexander and Ferzan? The ignorance of law problem, they rightly point out, more or less vanishes because only malum in se wrongdoing is now going to be criminalized, to be more precise, the risking of malum in se wrongdoing, so that few will be able to plead ignorance there. Nevertheless they devote an entire chapter to the subject, and the reason, again left somewhat implicit, is what happens to the problem of pure legal impossibility in their system: it turns into the problem of moral recklessness, which is as perplexing a problem as one is likely to find anywhere in ethics. Once one decides to put recklessness centerstage, as they do, the question then arises what to do with the person who is uncertain whether an action he is about to take is moral or not. Just why this is so problematic becomes apparent once one looks at what happens when one does what the pioneer in this area, Ted Lockhardt, did in his book Moral Uncertainty, which is to adapt standard decision theory to the purpose. Take a problem like abortion—his central example. A woman for whom carrying a pregnancy to term would for various reasons involve considerable sacrifice is considering an abortion. She is uncertain whether what she is about to do is moral or not. Suppose now we tried to apply standard decision theory. We would then probably be led to reason thus: If she has the child, that would clearly be a moral course of action. If she does not, it might or might not be. That situation looks analogous to the question of whether I should blow up a building that might potentially have a child in it when not doing so would be quite burdensome for me, or not do so because that is sure to avoid a calamitous outcome. We would presumably regard blowing up the building as reckless whether or not a child is actually in it. So should we regard having the abortion as reckless? If that seems at all tempting, consider what else would have to be regarded as reckless: eating meat or not giving most of your income away to the poor. In both cases, we are certain that not eating meat is moral and doing so might or might not be. Ditto giving away most of your income. So are you reckless there? But if not, a host of other difficulties await you, several of which are sketched out in their chapter on moral ignorance.

Recklessness. The concept of recklessness has a frankly consequentialist structure: it is essentially a kind of probabilistic cost–benefit analysis. That mixes uneasily with deontological restrictions. The difficulty of course is present in the current system as well but it becomes much weightier when recklessness is put centerstage as it is by Alexander and Ferzan. They are hardly unaware of it and enormous attention is devoted to it in several chapters, most notably one on the question of how to assess so-called “optimific wrongs.”

An example of an optimific wrong: “Alex, a large man, has fallen asleep on some trolley tracks. The presence of his body will prevent a runaway trolley from hitting and killing five trapped workmen farther down the track, but Alex will die. Betty sees what is going to happen to Alex and has time to wake him and get him off the tracks, but she cannot free the five trapped workers.” May she awaken Alex? The rest of the chapter considers a truly Mozartian set of ingenious variations and the answer they give for the most part is that this is something Betty may do on deontological grounds, even though it is not the loss of life minimizing strategy. So far so good. It seems they have found a relatively harmonious modus vivendi between the consequentialist orientation of recklessness and deontological principles.

But let us consider a different kind of “optimific wrong” case. Imagine two courses of action of the following kind: The first does a lot of harm, but not harm that appears on the defendant’s ledger under the principles Alexander and Ferzan lay out in their chapter “Risking Other People’s Riskings.” The second does much less harm, but all of it counts. The second course of action is thus more blameworthy than the first, the first not being blameworthy at all. What if one has a choice between the two? Do we compare them according to the metric of blameworthiness or harm? What they say about Alex and Betty would suggest we go with the blameworthiness metric. Now complicate the matter a little further. Suppose a cost needs to be incurred, some burden or risk imposed on a third party, to ensure that the more harmful, less blameworthy path is chosen. How much of a cost may one impose before that choice is deemed reckless? Here we are forced to somehow weigh utilitarian and deontological commitments against each other and it isn’t at all clear how this is to be done. What is clear is that, in whatever way it is done, there are going to be highly counterintuitive consequences.

3 III

There is a difference between criticism and critique. What we have sought to offer is critique not criticism. We have tried to draw out some of the counterintuitive consequences of taking the deviant turn. They are by no means fatal or disqualifying to the Alexander/Ferzan schema. After all, they took the deviant turn to avoid some genuinely counterintuitive features of the current system. It’s a tradeoff. A tradeoff well worth studying almost regardless of whether one is sold on their reform effort or not. For two reasons: first, because it sheds tremendous new light on the regime we have whose real raisons d’etre one hasn’t really understood until one has explored the implications of doing things differently. (Also, we suspect that there are many, many more such counterintuitive and surprising features to be discovered in the Alexander/Ferzan schema.) And secondly, what they have done—to repeat what we said at the outset—suggests an entirely new way of trying to understand a legal domain, something a clever scholar will want to imitate by trying it out on another body of doctrine. We just might embark on that ourselves. A colleague once told us that a break-through piece should suggest, but leave open, an optimal number of questions for others to explore. Alexander and Ferzan’s two books, beyond their various other virtues, do certainly do that.