In the world of software patents, there are no clear borders. This is especially true when writing a single software application of any utility whatsoever; because of the borderless nature of software patents, such an application may unwittingly trespass dozens of software patents, and the authors of the application have no efficient way of determining that they are trespassing until they are served with a cease-and-desist letter or notified of litigation proceeding against them for infringement.
Slate has this to say on the subject:
The "measurement costs" of software's boundaries—defining where one algorithm begins and another ends—are inescapably high.The article also delves into judicial activism, revealing the little-known fact that software and business method patents came about not because Congress authorized them, but because the Court of Appeals of the Federal Circuit decided to allow them in the 1980s and 90s.
This point about measurement costs is borne out in practice. As anyone in the industry will tell you, and as Ronald Mann has documented, most programmers or firms cannot figure out whether they're infringing software patents or not and simply always assume they are. The point is simple: Property without discernable borders brings all the costs and none of the benefits.
In this debate it must be remembered that the regime of easy patenting of software is not natural law but an experiment—a judicial flight of fancy in an area of difficult economic policy. To remain healthy, the U.S. legal system should experiment, but it also needs to recognize when its experiments have failed. Both Congress and the Supreme Court have a chance to do something about the problem this year. The rise of the patent trolls is, in this sense, telling us something—that it may be time to end a system that is doing no one any favors.