Illustration: Greg Mably

The four pillars of ­intellectual ­property—patents, copyright, trademarks, and trade secrets—all play roles in ­protecting software. No ­wonder this is one of the most slippery subjects for an engineer’s lawyer.

Luckily, the courts have established a framework that matches the way the real world works (up to a point). The framework begins by considering three milestones in software development: function, design, and code.

At the highest level of abstraction stand the functions that software performs. These functions are often specified formally. For the first word-processing program to include a cut-and-paste function, the spec might have read: “The program shall allow a user to highlight a section of text and move or copy that section of text elsewhere.”

Design refers to the plan by which the software will carry out the functions— which in patent applications often appears as a flow chart showing how the various subroutines, modules, and algorithms interact.

Finally, there is the code—the instructions that carry out the functions according to the design.

It is in moving among these three levels that the disparity between patents and copyright becomes clearest. By law, copyright can never protect functionality. And, there is a great uncertainty concerning even how much of the software design copyright can reach. Older court cases touting ideas like the “look and feel” of a software interface and the protection of “selection and arrangement” constituted attempts to protect software design. Those cases have now mostly been discredited. Even the notion of copyright protection for the code is limited, because it doesn’t apply to code taken from the public domain or mandated by hardware or software constraints.

The good news about a copyright, though, is that it’s relatively simple and inexpensive to register one. The code doesn’t have to be inventive, new, or unobvious; technically, it doesn’t even have to work. So in spite of the limits, you are well advised to register a copyright for all important software code.

Patents, in sharp contrast, are designed to protect functionality, usually not by covering specific lines of code but rather by covering a software “engine” that takes certain inputs, manipulates them, and provides certain outputs. The bad news is that patents are expensive, and you have to prove that the functionality you want to protect is new and unobvious.

Patents can also ­protect software design. An ­example: suppose someone else has already invented the cut-and-paste function, but you have designed a faster or more ­intuitively ­comprehensible way to perform the ­function. The “and ­improvements thereof” allowance in ­patent land would ­protect your refinement.

Next on the list of IP weapons is the trade secret, which is encumbered with a lot of conventional wisdom that happens to be dead wrong. Trade secrets do not offer free protection, they do not provide a fallback position, and they include several severe restrictions. They cannot protect function if it is publicly known—that cut-and-paste function, for instance, is not only not a trade secret, it’s an advertised feature. Trade secrets can, however, protect design or code, provided that you take active (and sometimes expensive) steps to invoke a company-wide program that will hold up in court.

The final weapon is ­trademark, which has little to do with function, design, or code but protects only the commercial name of the program. “Windows” is a trademark; so are the names of the software products advertised in this magazine. Trademark registrations are only ­moderately expensive and can be obtained fairly easily.

The function, design, and code demarcation should be kept in mind as we continue to debate any meaningful IP reform. Meanwhile, for any important software product, the general consensus is that all species of intellectual property should be explored and their various levels of protection exploited.