When is it ok to be “descriptive”

I presented a taxonomy of federal litigation today to a terrific audience at Rutgers-Camden. As I’ve covered in exhausting detail, the paper sets out to describe how lawyers organize causes of action together into complaints.  It uses a method called spectral clustering to illustrate the networks of legal theories that typically are pled together.  (It does some more stuff, but that’s the gist.)  As often happens when presenting this particular paper, it was pointed out to me that the project lacks a clearly defined normative “so what”.  This is basically correct. The “so what” of the paper is “this is a different, more-finely grained, way to see how attorneys think and produce cases. With pretty pictures. How do you like them apples?

As I said, I tend to get the so-what objection quite often when presenting this paper, and it’s pushed my co-authors and I to make the paper clearer about the implications of the method. At the same time, it has made me even more aware of the bias in legal writing to come up with papers that do more than taxonomize, or describe. This is a well-known problem with the legal academy.  True, taxonomies can be highly successful – Solove’s Taxonomy article is just one recent hit in a long parade of exceptionally good papers that basically try out different ways to organize legal concepts.  But those papers generally pitch the contribution of taxonomies  as systems to harmonize doctrine, or because they illustrate something about the world that needs fixing, or they uncover a missing category that is novel and interesting.

What’s less common is work that is no more than descriptive – this is what the world looks like; this is what happened – and doesn’t go on to fix or recommend a single thing.  Often such work is derided as mere reportage, a practitioner’s piece, or (worse) an uninteresting collection of facts, put together without a synthesis of why we should care.  (Actually, some papers are attacked on all three grounds.)  But other times, descriptive work is seen universally to be immensely important and valuable, even if it doesn’t advance any prescriptive agenda. Some of the middle-period Law and Society papers have this feel, though of course L&S generally is quite ideological.

You may be wondering: what’s the so-what of this post?  Here it comes:

what is your sense of the appropriate criteria for deciding that purely descriptive scholarship makes a contribution?

relatedly, if you were advising a first-time scholar, would you advise against writing a paper that is missing a policy solution in Part IV?  

My answer to the first question is that schools and faculties vary widely, and consequently I’d say the risk averse response to the second question is very, very clear.  Discuss.

You may also like...

14 Responses

  1. Orin Kerr says:

    In my view, the main criteria should be the significance of the matter described. If lots of people are wondering how X actually works, a paper that explains how X actually works is very valuable. The trouble comes when a purely descriptive paper answers a question that no one was actually asking: That’s what legitimately prompts the “so what” question.

  2. Orin Kerr says:

    Make that “criterion,” not “criteria.”

  3. Bill Henderson says:

    Dave, I think the right answer, “what do you think?” You wrote the paper. It seems to me that understanding how lawyers think and produce cases in a descriptively accurate way is a contribution in a scientific vein. If you agree, you are done. It is okay.

    When you ask, “when is it okay to be descriptive?”, I think you are asking, what is our collective criteria for good scholarship? This is really about gaining academic acceptance among one’s peers. But after tenure, shouldn’t this be an internal scorecard?

    Instead of the proverbial Part IV, stick to you guns and explain why you think this is valuable. Academic norms shift occasionally because some go against fashion. Thank goodness!

    Max Planck once said, “Truth never triumphs—its opponents just die out. … science advances one funeral at a time.” It that was true for physicists, it is certainly true for law professors.

    Keep doing your empirical work. Describe the world as it truly operates. The pieces will fit together over the long run. It is okay to describe what we don’t understand.

    My 2 cents. Bh

  4. Lawrence Cunningham says:

    – when it is interesting, important and true;
    – depends on context (e.g., field, training, school, expected productivity).

  5. Marc DeGirolami says:

    I wish I could simply “like” this post and be done with it. But I’m afraid there would be no payoff or, even worse, no normative takeaway; and anyway, no one would care: http://prawfsblawg.blogs.com/prawfsblawg/2009/11/who-cares.html

  6. It’s funny, but in many other disciplines, the question would be flipped: “when is purely normative scholarship acceptable?” And the answer, in some cases, would probably be “never.”

    There’s nothing wrong with legal empiricism — understanding what is happening descriptively is the best basis for making recommendations about legal policies. Far too often, we engage in the latter without the benefit of the former.

    Sorry I missed the talk, but I’m glad you deemed the audience terrific! 🙂

  7. Dave Hoffman says:

    Great responses so far, keep ’em coming. To Bill in particular: I’ve no worries about the piece, and feel pretty confident about its contribution. Was looking for more general norms & views about purely descriptive scholarship.

  8. Purely descriptive work is fine, even great, as long as its normative significance doesn’t need to be made explicit in the paper. Thus, if lots of people are asking, “Is X actually happening” because they have normative theories about X being good or bad, a descriptive paper about whether X is happening is valuable. Sometimes, a descriptive paper answers an important question that no one is asking; if so, it needs also to explain why the question is normatively important. Sometimes, a descriptive paper is relevant to a wide range of normative debates; if so, it can leave the implications unstated because there are so many of them.

    It sounds like your complaint-clustering work fits best into this last category. There is no specific normative payoff, but it feeds into a wide variety of conversations about civil litigation. Think of it as basic research: the goal is to understand the legal equivalent of scientific fundamentals, rather than to produce the legal equivalent of useful inventions.

    Of course, this can be flipped: purely normative work is fine, even great, as long as its descriptive dependencies don’t need to be made explicit in the paper.

  9. Dave,

    You have really hit the nail on the head here.

    One of the biggest weaknesses of our enterprise is the unjustified obsession with developing normative theories when the underlying positive/descriptive properties are so poorly understood. The balance is tilted way to far in the normative direction.

    Usually those “policy solutions in Part IV” are the WORST / LEAST justified parts of papers. You can usually see folks stretching / bending over backwards and it is entirely unnecessary in most instances (although the pressure to do so is always present).

    If folks just want to do advocacy that is fine but lets not pretend (in most instances at least) like it is anything other this.

    Serious people understand the usefulness of taxonomies, typologies, dataset papers, etc.

    I am with Bill H. — The Science of Science /Ecology of Ideas teaches us that cutting against the grain and exploring the broader intellectual terrain is what generates the real scientific breakthroughs … (although it is more likely to be ignored at the time)

    Check out here for more on the topic from scholars in complex systems / sociology following in the Kuhn/Planck debate tradition:

    So push on and let your opponents maintain their challenges from the retirement home …


  10. TS says:

    I doubt the pretense of social scientists that there is much scholarship which is “purely descriptive” in the sense that there is a theory ladenness of observation. After all, theories shape our observations. How do we operationalize the variables and what variables do we choose to observe? Surely that represents certain normative judgments.

  11. Matt says:


    I don’t think the necessity of specifying a model or operationalizing variables implies normative judgment, at least not in the way people are using the term “normative” here. In the physical sciences you must also have a theory and operationalize variables in a way that influences what is observed or measured. Does every chemistry paper represent “certain normative judgments”? I don’t see how what you call the “theory ladenness of observation” has any special significance for social science relative to physical science.

  12. John Coates says:

    If you’re writing for a legal academic audience, then the question is when is the purely descriptive going to be true, new and important to them, which may equate to what is going to change the way they think about some important aspect of law in a significant way.

    A built-in advantage of normative work is that by critiquing existing law, it tries to change the way the audience thinks about law. Even if the focus of a normative paper is narrow, it is still trying to change what readers value, and US legal audiences like plausible argument for its own sake.

    Description can change the way people think, too, particularly if is trans-substantive (like your paper) or presents new information about a part of law that has clear implications for other areas of law.

    But if description is narrow — e.g., the details of a particular cause of action in recent cases in a state with no national influence — then even if the description is powerful and elegant, it isn’t likely to impact anyone other than practitioners and the few academics in in-state schools who teach that particular topic. A normative point from the same narrow range of cases, however, can be often be generalized more readily.

  13. Some or most (thus I hope not all) of that which follows may be more or less obvious to many CO readers: in which case, I’m speaking to the few.

    While I believe in the necessity and value of a descriptive enterprise, I think we do need to be clear about the fact that that which aims to be “purely descriptive” is no less engaged in (normative) judgments as to what “counts,” what is relevant, what has bearing on the questions and enterprises that are important and meaningful to us, whether in the law or elsewhere (the ‘context of application’ so to speak, which may be hypothetical, predictive, explanatory, educational, edifying, interpretive, and so forth). In other words, “what, precisely, is this knowledge for?” Or, “this data set may prove useful for the following reasons or purposes…..” We circumscribe a particular descriptive domain of facts because we believe it has function, purpose, and value for the kinds of questions we (or others) are asking (we have contestable assumptions as to what are the interesting and relevant questions that would motivate this particular descriptive task). It is a map designed to guide us to getting a grip on what is relevant: it is a RELEVANCE MAP. The act of description is at the porous boundary betwixt and between the domains of information and knowledge, it aims to enable comprehension of something significant. We need to justify this taxonomy rather than that one (i.e., ask ourselves ‘why this, and not that?’). Any description much select from a virtually innumerable number of possible “facts” and perspectives on such facts, so much so that much hinges in the first instance as to the initial choice or act of circumscription with regard to what we decide to objectively or impartially describes as far as possible or practicable (this ‘objectivity’ may be simply or largely consensual in nature). And any particular descriptive endeavor involves (presupposes, assumes, and/or posits) any number of values (some would prefer here the less helpful concept of ‘interests’) and must exemplify certain cognitive virtues (coherence, elegance, abstractness, functional simplicity, or perhaps economy or parsimony, and the like) to be a worthwhile description. With regard to values, Robert Nozick wrote:

    “Values enter into the very definition of what a fact is; the realm of facts cannot be defined or specified without utilizing certain values. Values enter into the process of knowing a fact; without utilizing or presupposing certain values, we cannot determine which is the realm of facts, we cannot know the real from the unreal.”

    Or, as another philosopher (one, unlike Nozick, still with us), Hilary Putnam puts it, our knowledge of the world presupposes values, indeed, what comes to count as the real world depends upon our values (and these need not—and I believe should not—be construed in merely emotivist, subjectivist or conventional terms, nor should they be viewed as irrational or non-rational). This is evidenced in the “implicit standards and skills on the basis of which we decide whether someone is able to give a true, adequate, and perspicuous account of even the simplest perceptual facts….” It is Putnam who also reminds us that insofar as facts (or truth) and rationality are interdependent notions, a descriptive statement of fact entails “criteria of relevance as well as criteria of rational acceptability, and…all of our values are involved in our criteria of relevance.” So, should we want to proffer a description that is factual and thus true (that is, ‘true by our present lights, or “as true as anything is”’), we will be answering the relevant questions that motivate the descriptive enterprise, and at the same time revealing (intentionally or otherwise) our values or system of value commitments. Putnam elaborates:

    “The way in which criteria of relevance involves values, at least indirectly, may be seen by examining the simplest statement. Take the sentence ‘The cat is on the mat.’ If someone actually makes this judgment in a particular context, then he employs conceptual resources—the notions ‘cat,’ ‘on,’ and ‘mat’—which are provided by a particular culture, and whose presence and ubiquity reveal something about the interests and values of that culture, and of almost every culture. We have the category ‘cat’ because we regard the division of the world into animals and non-animals as significant, and we are further interested in what species as given animal belongs to. It is relevant that there is a cat on the mat and not just a thing. We have the category ‘mat’ because we regard the division of inanimate things into artifacts and non-artifacts as significant, and we are further interested in the purpose and nature a particular artifact has. It is relevant that it is a mat that the cat is on and just something. We have the category ‘on’ because we are interested in spatial relations. Notice what we have: we took the most banal statement imaginable, ‘the cat is on the mat,’ and we found that the presuppositions which make this statement a relevant one in certain contexts include the significance of the categories animate/inanimate, purpose, and space. To a mind with no disposition to regard these as relevant categories, ‘the cat is on the mat’ would be as irrational as ‘the number of hexagonal objects in this room is 76’ would be, uttered in the middle of a tête-à-tête between young lovers. Not only do very general facts about our value system show themselves in our categories (artifacts, species name, term for a spatial relation) but, our more specific values (for example, sensitivity and compassion), also show up in the use we make of specific classificatory words (‘considerate,’ ‘selfish’). To repeat, our criteria of relevance rest on and reveal our whole system of values.”

    Putnam also reminds us that norms and standards of a kind are intrinsic to our descriptive projects:

    “(1) In ordinary circumstances, there is usually a fact of the matter as to whether the statements people make are warranted or not. [….]
    (2) Whether a statement is warranted or no is independent of whether the majority of one’s cultural peers would say it is warranted or unwarranted.
    (3) Our norms and standards of warranted assertibility are historical products; they evolve in time.
    (4) Our norms and standards always reflect our interests and values. Our picture of intellectual flourishing is part of, and only makes sense as part of, our picture of human flourishing in general.
    (5) Our norms and standards of anything—including warranted assertibility—are capable of reform. There are better and worse norms and standards.”

    Any piece that aims to be “purely descriptive” is of course subject to subsumption within, or manipulation by, or enlistment within, works of normative scholarship, or any number of possible research agendas, and it would seem that the decision to construct this or that taxonomy must at least bear this in mind. In other words, no description is merely an island of data set apart from the bulk of explanatory and theoretical research, even if it remains the case that (after Hume), logically at least, the normative or prescriptive “ought” does not follow deductively from the descriptive or empirical “is” (and yet, after Marx, we should not confuse the logic of things for the things of logic, hence there are, after all, other forms of reasoning: inferential, evaluative, practical). Scientific taxonomies and factual classification have long been basic to the natural and social sciences, and philosophers of science do see them as steeped in theory: indeed, they acquire scientific respect or significance through theory. And our “purely objective descriptions,” that is, those worthy of the adjective “rational,” are indexed to conceptual schemes, indeed, facts are internal to conceptual schemes, so lucidity as to such conceptual schemes is yet another desideratum, and this is perfectly consistent with the (minimally realist) notion that truth is the correspondence of (relative) propositions with (relative) facts (truth itself being a value we care about: all truths are relative, but that need not mean that our conception of truth is a relative concept). Michael Lynch, who has well explained truth as “one and many,” elaborates:

    “[T]he conditions under which a proposition is true are partly determined by the conceptual scheme in which the proposition is expressed. But what makes a proposition true is not its relation to a scheme but whether or not the conditions in question obtain. For a claim to be true (or false), the conditions must be relative to a scheme. Yet the reason that the claim is true is not because it is relative to a scheme (as the truth relativist must hold); it is true because it is the case. [….] A fact, in the human sense, is simply what is the case.”

    Lastly, Nicholas Rescher speaks to other relevant epistemic aspects of our aims to be purely descriptive, aspects that necessitate an appreciation of

    “the diversity in people’s experiences and cognitive situations; the variation of ‘available data;’ the underdetermination of facts by data (all too frequently insufficient); the variability of people’s cognitive values (evidential security, simplicity, etc.); the variation of cognitive methodology; and the epistemic ‘state of the art.’ Such factors—and others like them—make for an unavoidable difference in the beliefs, judgments, and evaluations even of otherwise ‘perfectly rational’ people.”

  14. erratum (second para.): “…what we decide to objectively or impartially describe as far as possible or practicable….”