« The "Anti-Foreign Law" Craze--Bills, Amendments and Decisions | Main | Road Trip »

Thursday, November 06, 2014

The Injustice of Bumpy Laws

Whether a person has been sexually assaulted can turn on the reasonableness of a defendant's belief that the alleged victim consented. Presumably, the more reasonably one believed consent was given, the less culpable one is for proceeding with sexual intercourse. Similarly, the more reasonably one believed consent was given, the less dangerous one is likely to be. For both these reasons, the amount of punishment a person receives should be at least in part a function of the reasonableness of his belief in consent. In my terminology, reasonableness is an input into a legal decision that likely ought to have a smooth relationship with the pertinent output: amount of punishment. As the input gradually increases, the output gradually decreases. At some point, a person's beliefs are sufficiently reasonable that he no longer warrants any punishment (because he is not sufficiently culpable or because the costs of punishing him exceed the benefits).

In this instance, we seem to prefer smoothness in theory, but the criminal law's treatment is in fact quite bumpy. One person deemed just reasonable enough that consent was given may have no criminal liability while a similarly-situated person who was just a bit less reasonable may be sentenced to at least the statutory minimum. Even though both defendants are virtually alike in terms of culpability and dangerousness, they are treated quite differently under our bumpy treatment of beliefs about consent: gradual changes to the reasonableness input do not affect conviction, except at a critical threshold. At that threshold, a gradual change in reasonableness has enormous effects. 

So there appears to be a deviation between our normative theories of criminal law and what the law actually does. Now, criminal law may actually be less bumpy than many other areas of the law. At sentencing, judges can smooth the relationship to some degree by giving more reasonable offenders shorter sentences than less reasonable ones. But judges are often limited by statutory minima. Such minima may have certain advantages, too, by controlling the allocation of sentencing discretion. But the value of such discretion must be weighed against the harms of deviating from our best theories of just punishment. And many retributivists are committed to never knowingly overpunishing an offender, allocations of discretion be damned.

Even in the absence of statutory minima, most of us--judges included--are simply not closely attuned to the smooth-bumpy distinction. So many would look askance at a judge that gave a very light sentence for rape, even though there must be cases that closely straddle the line between guilty and not guilty. And other bumpy features of criminal justice cannot be mended by eliminating statutory minima. For example, for every day defendants spend in pretrial detention, they usually receive one day of credit against any punishment they may subsequently receive. But what if offenders spend pre-trial time in drug rehabilitation facilities that are less confining than jail but more confining than, say, house arrest?  Jurisdictions treat such issues in a bumpy manner. They either give full credit or no credit, even though there is a much smoother approach: give partial credit. We're surely used to thinking of criminal law in bumpy, all-or-nothing terms, but when you look for ways to smooth the law, you find many opportunities.

Posted by Adam Kolber on November 6, 2014 at 11:46 AM | Permalink


Thanks Brian! Perhaps the example I gave wasn't detailed enough to answer your question clearly one way or the other. One possibility is that there is direct precedent on the matter in this jurisdiction. Precedent A precisely coincides with the defendant's behavior and was deemed unreasonable, while precedent B reflects a defendant who exercised just a bit more care and was deemed reasonable. (Perhaps we need to add the assumption that something about the jury instructions clearly distinguishes these two cases for jurors.) If the background law were such, then we might expect a deterrent effect that was nevertheless still bumpy.

Of course, the far more realistic scenario and presumably the one you had in mind is one where no one is really sure in advance how the law applies to particular facts. I'm a little puzzled how to play things out in this variation because it happens to be a case of negligence. So I'm not sure how we analyze deterrence behavior in cases aimed at those who were unaware of a substantial risk that they were violating the law. I guess our focus, then, is on matters of general deterrence.

Let's assume that the law increases the deterrent force of the law as reasonableness decreases. Then, I'm still not sure how the overall consequentialist analysis plays out. If we're deciding the kind of return on our investment in incarcerative resources, it would seem that there is little to distinguish the quality of the investment in the defendant who was just over the line versus the defendant who was just under it. They're both about equally dangerous. (Ditto any hopes to rehabilitate the person and ditto the interest in specifically deterring offenders who might re-offend.)

In other words, perhaps the consequentialist would still have qualms on grounds of bumpiness, even in the case you likely had in mind. Perhaps one way to see it is to take my oversimplified example where the precedent is very clearly. Then, add some random probabilistic factor to it with the role of dice (which, I think, is how you get to the smooth kind of deterrent function you had in mind). Would that please the consequentialist? Or does the addition of randomness simply weaken the law's deterrent signal?

On top of all of this, I have questions about the harms caused in sexual assault cases, and how they relate to the reasonableness question. I think we'd want to know more about that in assessing how much we want to deter at particular levels of unreasonableness. So if it's not clear already, you've asked a rich, interesting question, and I hope I've addressed it to some degree. But I welcome further insights from you and the rest of the Prawfs readership! Thanks again.

Posted by: Adam Kolber | Nov 7, 2014 2:08:20 AM

Adam, a great project. To pick a nit with the post, isn't your example one of a law that is discontinuous ex post but continuous in expectation? That is, each actor's risk of conviction is increasing as reasonableness drops. From a deterrence perspective, the discontinuity doesn't seem that troubling, at least if probability of conviction depends on the value of the underlying, smooth, distribution. In focusing on the ex post perspective, you seem to be adopting a retributivist view that the incremental punishment does not reflect the incremental harm.

Have I got that wrong?

Posted by: BDG | Nov 6, 2014 5:22:28 PM

Thanks, Bruce, for your very thoughtful comment, and I'm glad that you like the article! I couldn't agree more that the bumpy part of a legal relationship can indeed be quite predictable. For example, sexual intercourse with a person one day shy of age eighteen may be a serious crime while sexual intercourse with a person who just had an eighteenth birthday is perfectly legal. The location of the discontinuity will usually be obvious. So I didn't mean for the term bumpy to carry the "turbulence" resonance: a bump in a road can stay there for years and be quite predictable. The term "bumpy" does contrast nicely with a "smooth" relationship, and when I look at the phase diagram you link to, I do see a bump. But I acknowledge that the term "bumpy" is a somewhat arbitrary term to describe the relationship I have in mind. If it's any consolation, the terms are more evocative than say, calling them "relationship #1" and "relationship #2."

As for what I take to be your more substantive point, I think you've identified a particular benefit of bumpy laws (that I acknowledge in the paper). Namely, bumpy laws tend to be cheaper to administer. It's easier to determine whether someone crossed a particular threshold than to determine where they fell along a spectrum. So, consistent with your comment, we may invest a lot more resources determining whether or not a tort or criminal defendant was negligent than determining just how negligent he was.

Given that judicial resources are indeed limited, bumpy laws may sometimes be a worthwhile trade off. But I hope the smooth-bumpy distinction helps to clarify the nature of the trade off and brings to light in the criminal law context how the system is essentially guaranteed to get punishment wrong in at least some cases.

Posted by: Adam Kolber | Nov 6, 2014 5:00:59 PM

Adam, this is a fascinating article. I'd like to quibble with one thing, and that's the choice of the word "bumpy" to describe discontinuous jumps. "Bumpy" conjures up an image of turbulence -- unpredictable and essentially random jolts produced by the uneven road surface or eddies in the air currents, both of which are hard or impossible for drivers/pilots to detect and avoid. But I don't think the discontinuities you're talking about are the result of turbulence. They are entirely predictable events where something changes it's state, e.g., not liable to liable. I think they are more like phase transitions. The transition from ice to water is not unpredictable, even though it involves a discontinuous jump. If all you are monitoring is the amount of energy invested to produce a rise in the temperatures, at 0 there is suddenly a huge jump in the amount of energy (and time, if the energy is being added at a constant rate) it takes to go from 0 to 1. This is what I have in mind: http://guweb2.gonzaga.edu/faculty/cronk/CHEM105/images/heating_curve_ice-water.gif

I don't think legal decision-making works like a physical substance but I suspect there is an equivalent amount of both complexity and predictability in the jump from "not guilty" to "guilty." It's more like thawing ice than air turbulence because observers know roughly (but not precisely) where the line is. And at that line, I think there is significantly more decisionmaking activity than on either side of it. Police will use their discretion at that line, as will prosecutors, trying to separate out the cases that should fall on one side versus the other. (Of course they are flawed, but your pot doesn't heat evenly either.) Jurors and judges will also spend (hopefully) more resources in determining whether the facts and the law fall on one side of the line or the other. It may be that, like water, it's not the case that one unit of carelessness all of a sudden pushes you from no negligence to a full negligence state -- it could be that it takes multiple units to get you across that threshold, one to be sued, one to lose a motion to dismiss, one to lose a motion for summary judgement, etc. Or perhaps the probability of conviction/liability increases linearly as you proceed through the phase transition. Or the severity of the sentence/amount of the damages awarded. Or all 3.

Posted by: Bruce Boyden | Nov 6, 2014 4:05:57 PM

Post a comment