Posted by: Jamie Winterton
In my role with the Global Security Initiative at Arizona State University, I work closely with colleagues in the law school. At law school events, I often introduce myself by saying:
“I’m a physicist by training… so I work with laws, but they’re in a much different jurisdiction.”
The lawyers usually get a laugh out of it. But there’s an element of truth there.
One thing I’ve always loved about physics is the precision with which we can describe our world. Why do the planets move the way they do? They’re obeying Kepler’s Laws of Planetary Motion. Why can’t you cool your house by leaving the refrigerator door open? That would violate the second law of thermodynamics. Our equations may be long and our explanations interminable, but the laws of physics are precise.
Coming from the field of physics, law often feels turbulent, or even “fuzzy”. The joke often told about law school (and the best jokes are based on truth!) is that everything in law boils down to the phase “it depends”. I hear it a lot from my law school colleagues. Richard Feynman, rock star physicist, once said “Imagine how much harder physics would be if electrons had feelings!” In the same vein, imagine how much harder physics would be if there were conflicting interpretations on how to apply the law of gravity. “The space mission was going well until the probe hit the 7th sector, which tends to favor a interpretation…” Or to riff on the phrasing from one famous copyright decision (which I’ll talk about more below): “So far, in the laboratories, the second law of thermodynamics has received a mixed reception. While some materials have adopted its reasoning… others have rejected it”.
To quote an ASU professor in Cyberspace Law: “You see the problem.”
The case I proxied in the example above, Whelan Associates v Jaslow Dental Laboratory, is one of the classic decisions on the topic of copyright. Namely, to what extent is copyright protection provided to software? We know from the copyright act that copyright covers the expression of an idea, not the idea itself – but how this ideal translates to software has been addressed for years, and not always with the same results.
In Whelan v Jaslow, Elaine Whelan, an experienced computer programmer, was hired by Jaslow Dental Laboratory to create software that would help the owner, Rand Jaslow, manage his business. The program was called Dentalab, originally written in a programming language called Event Driven Language (EDL) that was compatible with Jaslow’s machine, which was at the time an IBM Series One.
A few years passed, and Rand Jaslow realized that many smaller dental companies were using newer computers – computers that weren’t compatible with EDL. So Jaslow created a program in the BASIC programming language – one with the same functionality as Dentalab, called “Dentlab”.
Having created a new company, Dentcom, Jaslow terminated his relationship with Whelan’s company, and formed a new company that sold both the Dentalab and Dentlab programs. (To avoid confusion due to the nearly identical names, I’ll refer to the former as ‘Whelan’s code’ and the latter as ‘Jaslow’s code’).
Shortly after founding the company, Jaslow sued Whelan, claiming that she had misappropriated his trade secrets. Whelan countersued, claiming copyright infringement of her original EDL code.
One of the expert witnesses for the plaintiff testified that “The person who constructed the Dentcom system must have been familiar with the series one system, because the same file structure and same program steps are followed, same overall flow takes place in both systems.” However, the code must have looked significantly different. Here’s a simple example of the same function written in two different (albeit more modern) programming languages – Python and Java:
Example image from Steve Ferg’s blog ‘Python Conquers the Universe’: https://pythonconquerstheuniverse.wordpress.com/2009/10/03/python-java-a-side-by-side-comparison/
While both code snippets do the same thing – print the phrase “Hello, world!” to the screen – the way in which they do it is obviously different. Java requires a lot of specific instructions; Python makes quite a few assumptions. The code on the left looks very different than the code on the right. The same is likely true for code in EDL and BASIC (although as EDL is deprecated, I was unable to find a sufficient example.) Are those differences sufficient to deny copyright protection to Jaslow’s code? The court decided no – that the structure of the code, not the literal lines of code themselves, were copied, and this was sufficient infringement of Whelan’s copyright.
With Whelan v Jaslow decided in the Third Circuit, were questions of copyright and source code settled? Hardly. A turbulent six years passed, which saw some courts agreeing with Whelan and some rejecting it. Then in 1992, along came the Computer Associates v Altai case. Again, questions over the extent to which non-literal elements of software are protected by copyright are raised, this time in the Third Circuit. This case centered on how tasks are assigned and executed automatically by a computer. Computer Associates (hereafter ‘CA’) created software to schedule tasks, and a support function that would translate the scheduler between operating systems for broader use. Altai created a job scheduling software package, and in an effort to translate it to a different operating system, employed a former programmer from CA. Unbeknownst to Altai, the programmer leveraged 30% of the CA code to do the job. When Altai found out (because CA sued them for copyright infringement and misuse of trade secrets), they ordered a “clean room rewrite” and re-release of the scheduler, to ensure that none of the lines would be identical to the CA product. Even with the rewrite, CA appealed, using the same logic as Whelan – that the functions of the products were substantially similar and that Altai was infringing their copyright.
In its decision, the Altai court rejected Whelan “because it is so closely tied to what can now be seen—with the passage of time—as the opinion’s somewhat outdated appreciation of computer science”. They follow with a new test for substantial similarity – the Abstraction-Filtration-Comparison test. This test takes into consideration the fact that a single piece of software may consist of multiple ideas and expressions, and that they must be sorted accordingly. Once the fundamental levels of the software have been determined, and the ideas have been separated from the expressions, there may be core expressions that remain, which are protectable. As the court stated, “In terms of a work’s copyright value, this is the golden nugget,” and the court will evaluate infringement based on these portions (if they are found to exist).
Even though the Altai case established a test for substantial similarity in software, assessing non-literal infringement isn’t cut-and-dried. More than thirty years later, the courts are still faced with these issues. Consider the case of CSS v Herrington – a case from 2017 which cites Whelan and Altai in its decision. The CSS story includes some very familiar elements – Chris Herrington, originally a programmer with CSS (his father’s company), left the company after several years to start a competing business. Both companies sell land management software to county clerks, and CSS filed for a preliminary injunction against Chris’ new company, claiming copyright infringement and theft of trade secrets. CSS’ petition failed, but the case itself is interesting in that it contains a paragraph of contemplation from the judge on how to assess non-literal infringement, starting with the understanding that the “Fourth Circuit has not yet explicitly endorsed a test for non-literal copyright infringement for computer source code”. The judge elects to apply the Altai test, but only after an assessment of the other Circuits’ behavior, and it’s very clear that the application of Altai is a choice. A popular one, but still – “it depends”. The relevance of the Altai test must be continually re-evaluated. At some point, it may no longer suffice.
One project I worked on as a physicist used data from a particle accelerator. During our analysis, we applied the laws of relativity. We did so not because it was a popular interpretation – but because it was correct. To do otherwise would simply be wrong. The machines would not have worked under any other paradigm! Technical disciplines can’t function when “it depends”. So it can be difficult for people from technical disciplines to immediately understand flexibility in the law. Wouldn’t it be easier if there were immutable ways to define what can be protected under copyright, for example? Easier, perhaps. But would it be more correct? Would inflexible guidelines fulfill the intention of copyright law in the Constitution – “to promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries”? Certainly not. Unlike physical laws, legal precedent must be revisited and reinterpreted continually within context, to serve a legal purpose.
This is not to say that physical laws are never updated. If that were true, we’d never have created particle accelerators! But the way in which the laws of physics change is significantly different. As physicists, we take these laws as “truth”, until our experiments (physical or theoretical) break down. When that happens, the field experiences upheaval, until we are able to update the law such that old and new experiments are reconciled. But in the meantime, we accept the physics we have and build on top of it. New accelerator experiments don’t include a re-evaluation of relativity, in the way that new court cases must re-evaluate precedent.
It’s important to note that “It depends” doesn’t mean outright chaos. It doesn’t mean that judges are free to make decisions based on whims. It does mean that each case has a different context, and that context is important. If electrons weren’t literally identical, and if they were constantly evolving as technology does, we’d have to consider their context as well. As technology advances, often in ways that can’t be predicted, we need flexibility of interpretation in the law to ensure that things are fair.
The original decision in Whelan reflected how software was created and understood at the time. Altai reflected not just a new understanding of software, but of the effects Whelan had on the industry. Evolution of technology, and its place in society, required a re-interpretation. New facets of these topics have been explored, and in the future, the courts will certainly continue to evaluate these topics. The Altai test may eventually fall by the wayside for some courts. Technology changes, people change, the relationship between the two continues to evolve. What will happen next in the realm of copyright and software? I suppose the answer is: “it depends”.