Taxonomies and Commercial Reputations

By Eric Goldman

This coming Saturday, the Information Society Project at Yale Law School is sponsoring a very attractive event entitled “Reputation Economies in Cyberspace.” I’m especially excited about this event because I think my next big project will focus on reputation topics, so this should be a fantastic learning experience. I’m on the last panel, which is a precarious time because of the high preemption risk. As a result, I’ve picked a less-than-mainstream topic with a low preemption risk, although I may move into more mainstream topics depending on what gets discussed earlier in the day. Here’s the short summary I provided to the conference organizers:


Taxonomies and Commercial Reputations

A “taxonomy” is a structure for organizing content. It provides the anchors that allow topically relevant content to be grouped together in a logical fashion. Among other benefits, taxonomies can provide a system for designating unique identifiers for marketplace offerings. These unique identifications are crucial for the development and management of commercial reputations. For reputational mechanisms to work properly, objective and subjective data about offerings need a place to be associated uniquely with the offering. Without this, the data has no place to attach, distorting the reputational mechanism.

Proprietary rights threaten the ability to optimally taxonomize marketplace offerings in at least two ways. First, taxonomy developers can assert a proprietary interest in their taxonomies. Second, trademark owners can use their proprietary rights to distort the taxonomy or content attached to it.

Taxonomy Developers’ Rights

Taxonomy developers may be able to claim copyright in their taxonomies. See American Dental Association v. Delta Dental Plans Association, 126 F.3d 977 (7th Cir. 1997); but see Southco, Inc. v. Kanebridge Corp., 390 F.3d 276 (3d Cir. 2004); ATC Distribution Group, Inc. v. Whatever It Takes Transmissions & Parts, Inc., 402 F.3d 700 (6th Cir. 2005). Even if they cannot, online taxonomy developers can restrict access to their taxonomies through server protection doctrines (such as trespass to chattels, Computer Fraud & Abuse Act, computer tampering doctrines, etc.).

Excludable taxonomies create two problems. First, competitors need to recreate taxonomies. This leads to duplicative efforts that are socially wasteful, and implicitly it increases barriers to entry by new intermediaries. Second, and perhaps more importantly, it hinders consumers’ abilities to do apples-to-apples comparisons between marketplace offerings, because consumers must do extra research to determine if taxonomical nodes in two different taxonomies are the same offering. As a result, consumers miss valuable reputational information because they cannot find it.

There is a licensing market for taxonomical data in many (but not all) product verticals. These licensing programs can be expensive for new entrants. Some licensors provide unique identifiers that can enable consumers to make apples-to-apples comparisons, but in other cases, catalog standardization/normalization remains a challenge. Furthermore, if there are competitive licensors in a particular vertical and they use different identifiers, then consumers may face a cacophony of identifiers.

This could be solved through a comprehensive and non-excludable product taxonomy with unique identifiers for all marketplace offerings. The exemplar is the ISBN taxonomy, which has done a remarkable job of allowing consumers to find and compare books. Because this economy-wide taxonomy hasn’t developed via private efforts, and because any developer will likely assert proprietary interests in the taxonomy in ways that would hinder its functioning, government sponsorship may be necessary (and appropriate) to develop the uniform taxonomy.

Trademark Owners’ Rights

By definition, trademarks should act as unique identifiers for marketplace offerings. Presently, trademarks provide the main taxonomical structure for marketplace offerings in most industry verticals. However, the proprietary interests of trademark owners limit the utility of trademarks as a taxonomy in at least two ways.

First, trademark owners can use trademark law to limit the use of their trademarks as a taxonomical node. Retailers can generally use trademarks for the products they sell under the trademark exhaustion doctrine. However, other intermediaries (such as product review sites) have no such defense, and their usage may qualify as a trademark use in commerce, meaning that any trademark inquiry becomes messy and unpredictable. Furthermore, search engines provide consumers with access to unstructured databases and use user-initiated search keywords—which may be trademarked—as a type of “dynamic taxonomy,” and this has exposed search engines to potential trademark liability as well.

Second, trademark owners may also use trademark law to strip out content that has been anchored at the trademarked taxonomical node. For example, in a group of product reviews, trademark owners could attack negative reviews as violating their trademarks. Once again, those reviews might satisfy the trademark use in commerce doctrine, leading again to messy and uncertain analysis. Indeed, trademark owners have every incentive to use trademark law to produce “lopsided databases” where favorable opinions remain and unfavorable ones are excised.

Some solutions to address these problems include:

1) We should provide a legislative safe harbor allowing search engines to use trademarks to create dynamic taxonomies for unstructured databases.

2) We should use the innocent printer/publisher safe harbor (15 U.S.C. §1114(2)(A)-(C)) more extensively to curb efforts to produce lopsided databases.

3) We should categorically exclude all referential trademark uses (i.e., uses of a trademark for its referential value), even if made by commercial actors in commercial settings, from trademark scrutiny. I build this argument out here.