Fair use and the Internet frequently clash. One common practice on the webone that I engage in, for reasons that will become apparentis so-called "deep linking." By linking directly to material that is not on a website's front page, one enables a reader to jump directly to the content he or she is interested in. Some content providers really, really don't like this. The purported excuse is that it keeps readers from seeing the oh-so-valuable advertisements that some of these sites are convinced provide their ultimate revenue stream. (They don't, but that is for another time.) The purported rationale is that deep-linking creates a derivative work, and thus cannot be fair use. The Online Journalism Review has gone so far as to print a seven-page article on its website. However, the whole argument misses the point.
Instead, the problem is not with creation of unauthorized indices or copies. It is the difficult distinction among direct, contributory, and vicarious infringement. If I copied an entire article verbatim from the Los Angeles Times, or linked to the article inside a frame that made the article appear as if it was on my own site, I would be directly infringing the article. (We'll leave aside fair use for the moment, since it is a defense that becomes relevant only after one finds an infringement.) The copy on my website, or that I feed to the user through the frame, is a direct infringement because I, the website operator, have made the copy. If I limit my activities to, say, a centralized index of authorized MP3s, I have not directly infringed the copyright (at least not if I have not downloaded all of them myself!). I have, however, contributorily infringed the copyright, "[L]iability exists if the defendant engages in personal conduct that encourages or assists the infringement." A&M Records, Inc. v. Napster, Inc., 239 F.3d 1004, 1019 (2001). A contributory infringer must also know of the infringements that take place using his or her system. Id. at 1020. Where knowledge is absent, though, one might still be liable for vicarious infringement when "a defendant 'has the right and ability to supervise the infringing activity and also has a direct financial interest in such activities.'" Id. at 1022.
So, then, what is deep linking? There remains one key issue, and to my mind it is a critical one. Individual MP3s are individual, stand-alone copyrighted works. A link to such a file, when used, creates a complete copy. A deep link, however, may not. The deep link may be to, say, a sidebar within an article; it may be to quotations from a government document; it may be to any number of things. Functionally, though, deep-linking is just an instantaneous index. So long as deep linking does not retrieve a stand-alone copyrighted work, it is merely a technological advance on the card catalog, or the Reader's Guide to Periodical Literature. It must instead be analyzed under the (rewritten) Sony standardwhether the particular instance of deep linking has a substantial defensible use. (The language in the case law is "noninfringing," but that is not entirely accurate; fair use, for example, is a defense relevant only after finding a prima facie case for infringement.)
With few exceptions, the purpose of deep linking is to enable commentary on the linked material. That was my purpose in, for example, linking to the OJR article in the first paragraph. This is functionally no different than if I had provided a search that would pull up the article, or mentioned it in footnote 217 of a law review article on fair use. This question really isn't even a close one. So long as the pointer is to an authorized source, deep linking is nothing more than a hyperefficient index. If the pointer is to infringing material, though, it is at minimum vicarious infringement.