By nature, copyrights and patents exist to protect creators and inventors. Giving individuals and corporations a monopoly on their work incentivizes them to share their creations with the public, thus contributing to the general enrichment of society. However, the general enrichment of society is often left out when thinking about existing copyright and patent laws. If a creator wants to declare that a certain work is their intellectual property (IP), but doing so would undeniably harm the public or is contradictory to common sense, then it should not be allowed. This mode of reasoning needs to be used as the legal system considers how software fits into the current intellectual property regime. Due to its wide scope, software needs to be considered as a distinct entity from existing creative works because it challenges the existing conception of what constitutes creative expression.
Applying existing definitions of copyright to technology can fail even in the most simple cases. Consider a fantastically designed webpage built in HTML and styled with CSS. The webpage was the product of hours of work by a designer and a front-end web developer. Under US copyright code, this webpage can be, and should be, copyrighted because it is an “original work of authorship fixed in a tangible medium.” If someone were to use that web-page and create a new one which looks reasonably similar, this should be regarded as a copyright violation because the individual effectively copied an entire creative work. It is equivalent to someone painting and selling a piece of art after staring at it in a museum.
This seems perfectly unambiguous until one considers the specific components which make up the webpage. What would happen if instead of copying the entire webpage, someone copied the functionality of a dropdown menu? Since a drop-down menu is a “method of operation”, it can’t be copyrighted, but it can be patented. However, something as simple as a drop-down menu should not be patentable. If it was, it would severely limit the capabilities of websites by removing a conventional method of navigation, thus inhibiting more websites from being created because drop-down menus are a obvious mode of navigation.
In hindsight, such a judgement is easy because drop-down menus are now ubiquitous, but the example holds in principle as well. If there are only a limited number of ways to accomplish a task in a logical manner as often is the case with certain aspects of software (such as how to navigate a user interface), then the specific components which can accomplish that task should not be protected under IP law. Therefore, while the design aspects of user interfaces should be protected, the individual components should not.
Outside of user interfaces, IP law is also inadequate when considering programs and code. The most notable instance of this is the lawsuit between Oracle and Google. Google rewrote Oracle’s Java API to use in Android, and Oracle sued them for it. The primary issue in this case is not that code was copied, but rather the mode of interaction with built-in functions was copied. For example, one of the functions in both API’s is declared as “public static int max(int x, int y)” (Source).
This is an entirely logical and obvious way to name a function which computes the maximum of two numbers, and it accordingly shouldn’t be copyrightable on the account that it is a logical and obvious approach. Moreover, if it does become copyrightable, it would severely alter the current tech landscape. For example, if someone wrote the same method in a different programming language, it would logically follow under current law that it should not be allowed since it is merely a translation, and translations must be conducted with the permission of the original author (Source). For example, the same method signature in Swift would be “public static max(x: Int, y: Int) -> Int.” It conveys the exact same meaning, and thus is effectively a translation.
If translation copyrights applied, new programming languages would each need to be radically different from each other because nothing could be named the same as the first language developed. Although the case between Oracle and Google is more complex than this simple example, if copyright legislation were to protect APIs in general, its application to the simplest cases would lead to a fractured and incoherent software landscape, inhibiting the development of new technology. Accordingly, API’s should not be copyrightable to prevent this situation.
On an even more fundamental level, current IP law is unable to make the difficult distinction between algorithms and applications of an algorithm. Most computer algorithms are extensions of mathematics or logic. For example, a neural network is fundamentally a series of matrix operations. Since math can not be patented, the concept of a neural network should not be patentable either because this would make it harder to make use of something which is arguably a fact of nature.
This is most likely covered properly under current law, but it becomes ambiguous when considering a machine learning model in its entirety. A model is more than just the math which it runs on. Hyper-parameters, the format of input data, the network structure, and the training mechanism are all choices made in creating a model and influence its success. Therefore, while the concept of a neural network is mathematical, each specific model has synergistic properties, such as being able to perform well on a given task, which make it more than an abstract concept.
In this way, a certain applications are more like a machine or other tangible invention than a mathematical concept, and thus should be patentable. It is unclear how IP law would handle a case about neural networks or other applications which are more than just the algorithms which constitute them. However, resolving such ambiguity would require an addendum to current law because the ambiguity between math and invention would likely divide the judiciary if it is left to merely interpret the current laws.
In general, legislating new IP laws to deal with the ambiguities created by software would reduce the uncertainty in the tech community between legal and illegal practices. However, any legislation which is passed will only be beneficial if it is meant to promote innovation and enhance public knowledge and not to set up unnecessary paywalls and prevent industry standards from being reached. This is a high standard to be met, but it can only be accomplished by reconsidering current perspectives on intellectual property.