Line drawing is inherent in law, and even though it is frequently arbitrary, this doesn't mean it shouldn't be done; of necessity, when there is a case or controversy, a court must do the best it can. Did defendant take too much, for example. On an ad hoc basis, line drawing has little systemic effect. There is, however, a strong pull toward creating dichotomies as a short-cut: the idea/expression dichotomy and the fact/expression dichotomy are two well-known examples. Ideas and facts aren't protectible but expression is. Once given material is classified as one or the other, the court's job is done because the dichotomy does the rest. The dichotomy is preexisting, objective. this posting deals with a third, the fact/value dichotomy, by which we say we won't protect compilations of facts, but we will compilations of things expressed as a value judgment.
Logicians have long pointed out that one problem with dichotomies is that they create two mutually exclusive categories: A and not A. These two categories cannot share features. Because they are different, one must develop fairly rigorous ways to define the characteristics that define the two categories. The messy nature of facts throws a monkey wrench in many attempts to create tight dichotomies in law, and that's one reason a number of judges, such as Judges Easterbrook and Jon Newman have rejected Nimmer's assertion that there is an "abstractions" "test" (derived from a misreading of Learned Hand); they reject the test approach and favor the continuum concept, by which courts don't dichotomize the inquiry but instead examine the degree of protection.
In the area of compilations, though, identifying idea and expression has proved particularly difficult because the basis for protection is shaky, certainly since Feist threw out sweat of the brow. One concern is that the idea of a compilation might be defined in a way that will eliminate protection, as when we say the idea is to compile data on all households in Ohio who voted for George Bush in 2004 and who had family members killed in Iraq. Certainly we don't want to give a monopoly for that alone, and perhaps there will be protection in arrangement of the data.
But here, there are verifiable facts: the household had members who either voted for Bush or didn't, and if so, either had members who were or were not killed in Iraq.
Some compilations contain different types of data, and here is where the fact/value dichotomy comes into play. If I compile a book about estimated car values, as in CCC Information Services, Inc. v. Maclean Hunter Market Reports, Inc., 44 F.3d 61 (2d Cir. 1994) are those estimates facts or the exercise of judgment? The same applies for coin prices, CDN Inc. v. Kapes, 197 F.3d 1256 (9th Cir. 1999). Perhaps more easily, what if I select my favorite restaurants, as in Adventures in Good Eating, Inc. v. Best Places to Eat, Inc., 131 F.2d 809 (7th Cir. 1942). Am I out of the realm of objectivity and into the realm of judgment and therefore protectibility?
Wikipedia gives a summary of the fact/value dichotomy, while British philosopher Hilary Putnam in his November 2000 Rosenthal Foundation Lectures at Northwestern Law School, "The Collapse of the Fact/Value Dichotomy attacks the very basis of the dichotomy as predicated on unsustainable views of facts as always being objective and value judgments as always being subjective. This is the problem of accurately defining the characteristics of A and not A. Putnam also attacks the dichotomy as being a "thought-stopper," by which he refers to the phenomenon I noted at the outset: once a dichotomy is regarded as being preexisting, courts may regard their job as merely one of classification.
This doesn't mean the problem of protecting some material and not other material goes away; it only means that we should think things through for ourselves, perhaps based on the underlying purposes for protection. When we get to compilations, what is the goal, the purpose? Do we want to base protection on predictions of prices or valuation (in the different sense of monetary worth)? What about compilations of assessments of taste or aesthetics, what makes them so different that they become acts of authorship?