Court Says You Can Copyright Numerical Ratings

Legal Issues

from the this-seems-troubling dept

The US, thankfully, still hasn’t allowed copyright on databases. There’s been a push over the years to allow such a database right, but this is problematic for a variety of reasons — mainly that it’s effectively a way to allow the copyrighting of facts, so long as you put a few of them together. But a bigger issue is that there’s empirical evidence that the ability to copyright a database doesn’t create more databases or actually help businesses. It does the opposite. It limits business — exactly the opposite of copyright’s stated purpose. If we actually had an empirically driven copyright system, there wouldn’t be database rights, because the evidence that they do more harm than good is quite clear. Anyone actually pushing for a database right is either ignorant of the evidence or is hoping to profit off such a right by limiting the market.

So it was of great concern to me when I saw a blog post from a lawyer suggesting that the US courts may be effectively allowing a database right on rating data. Beyond the troubling nature of the case, the lawyer who wrote that post also claimed that database holders might finally be “getting the protection they so desperately deserve, and need.” Neither of those points is true. Database creators neither deserve nor need protection — and the evidence on that is quite clear. The database industry in the US has been thriving without such protections, while places in Europe that do have such a right have seen significantly limited growth in those industries, with the data being a lot more expensive and a lot less useful. There is simply no compelling reason why such protectionism is needed unless one wants to simply ignore all empirical evidence.

Then, as you dig into the details of the actual case, you begin to realize what a poorly reasoned and dangerous decision this is, effectively allowing copyright to be expanded to data itself. Eric Goldman does a masterful job detailing the many, many, many problems with this decision. His description of the basics of the case make it clear how ridiculous the outcome is:

A Colorado judge has reached the remarkable conclusion that a hospital publicizing its star ratings and other recognition from a third party rating service in its marketing material might be committing copyright and trademark infringement. This is a little like saying that it could be copyright and trademark infringement for a law school to include its US News rankings in its marketing material or for a book publisher to issue a press release announcing its ranking on the New York Times bestseller list. CRAZY.

Goldman goes on to break down exactly where and how the judge went wrong on every single aspect of the ruling, touching on copyright, trademark and breach of contract. You should read his whole discussion, but here’s the excerpt on copyright:

Let me start with a basic proposition. A single numerical value can never be copyrighted. Ever. I don’t care what formula produced the value; I don’t care how many digits the number has; I don’t care what explanatory text is used to describe the value. I know cases occasionally have reached the absurd result that individual numerical values can be copyrighted, including one of my least favorite copyright cases of all time, the CDN v. Kapes Ninth Circuit case. They are wrong wrong WRONG.

Courts can reach this erroneous conclusion by treating a numerical output as a “compilation” of underlying data values. If you squint, you can almost see how this makes sense. The publisher chooses the underlying values to include, uses editorial judgment to build the algorithm crunching those values, and sometimes layers subjective judgments on top of the algorithm’s output. However attractive this logic is, I think fundamentally misreads the copyright statute’s definition of “compile.” Under the copyright act, a compilation must represent a “collection and assembling of preexisting materials or of data that are selected, coordinated, or arranged.” When a single number distills but obscures the underlying numerical values, the single number cannot reflect a selection, coordination or arrangement of the underlying numbers. Thus, according to my argument, numerical values cannot be compiled unless the reader can see those underlying values directly.

In this case, the judge gets led astray by contemplating the idea/expression dichotomy as a spectrum with “discoveries” on one end and “expression” on the other. Because the ratings aren’t discoveries, the court concludes they should qualify as expression. But the court’s dichotomy is fatally incomplete. Instead, the inquiry is whether a single numerical value can represent an original work of authorship because it expresses an idea. A single numerical value cannot express an idea any more than a single word ever could.

Even if one reaches the incredible conclusion that a single numerical value is an original work of authorship, then surely it is preempted from copyright coverage by the merger doctrine, which says that if there are a limited number of ways to express a fact or idea, then the idea and expression merge into a single uncopyrightable whole. It seems like the star ratings in a 1-3-5 star rating system would, by definition, be subject to merger. Sorry to state the obvious, but how many ways are there to express that someone is rated one star??? Nevertheless, this court distorts the merger doctrine by saying the idea being expressed here is the rankings of healthcare providers. This is too high a level of conceptual generality. If every judge used this level of abstraction, the merger doctrine always would be a null set.

Goldman also explains why such a dreadful ruling, which seems to go against common sense and the law in nearly every way, may set a terrible precedent and get used in other cases:

If other courts follow this judge’s “logic,” the potential for mischief from cases like this is enormous. Think of every reputational system that spits out a numerical assessment of the subjects it evaluates. Now, assume each and every one of those numbers is copyrighted. Individual eBay feedback scores? Individual FICO scores? Individual Billboard rankings of songs and albums? All possibly copyrighted and requiring the initial publisher’s consent to republish. Add in potential trademark claims, and the crazy-o-meter goes off the charts.

Ugh. This is a bad and dangerous ruling all around. And as you can see from our original link to the case, there are some other lawyers who think it’s exactly on track. That’s a bad, bad sign. Hopefully this ruling doesn’t stand for very long.

Filed Under: copyright, database rights, numbers