Indeed, implementing the algorithm from the non-code information should not constitute a copyright infringement. As a an example, albeit of potential patent infringement, you can read the story of the FFT, as recalled in the oral history from James W. Cooley on FFT and Numerical Analysis.
Excerpt (click to open)
Writing the 1965 Paper, Patents
Goldstein:
When did you start working on the FFT paper and can you tell me something about writing that paper, the 1965 paper?
Cooley:
After I wrote the program and gave it to Dick Garwin, he went about publicizing it and they still didn’t think it was very important. We had a seminar here in the math department, as we often did. We were talking about computing and so on, and a couple of the people who were developing APL, Ken Arborson. Are you familiar with APL?
Goldstein:
Sure.
Cooley:
Well, they were giving some lectures on APL and then they said, “are there any other algorithms that we can illustrate with this?” Of course it wasn’t the programming language then, they were talking about it as a way of describing algorithms. Nobody had implemented it on a computer. So then I volunteered to give a talk on this algorithm, which I did. Then it gained some interest, they programmed in APL and did a very good job and I think one of their later programs could do the whole thing in one line of APL.
But in any case, there was a lawyer in the back of the room there named Thomas, but I can’t remember his first name. But he heard my talk and said afterward, “This has patent possibilities.” So they called a meeting with some attorneys, the patent attorneys, and they decided yes, it has patent possibilities. There are several considerations, however. One is that John Tukey isn’t an IBMer, the other is that we better get the thing - instead of patenting, put it in the public domain so we protect the right of IBM to use the idea, before someone else patents it.
Cooley:
So they suggest that I write a paper on it.
Goldstein:
Yea.
Cooley:
So I just wrote a paper, if you have seen it, it is rather short, simply describing the basics of the method and of course you could have N, any number of factors and then show you can get the smallest number of operations of N, as a power of 2 or 4, but any set of factors will do, which John pointed out right from the beginning. And the paper went back and forth from me to John Tukey several times. Finally he agreed on it and we got it published. As I said, it was only because the lawyers told us to do so. Now the method for protecting this as a patent was this. The lawyer suggested that somebody design a device, because at the time you could not publish or patent an algorithm. You have to describe a process on a device, so they suggested that Ray Miller and Wienegraad in our department design a circuit which would do the Fourier transform by this algorithm, and they told me to put a footnote in the paper saying they did it. That would put the whole thing in the public domain and prevent anyone from patenting it.
Goldstein:
So they did an analog circuit that realized the FFT?
Cooley:
It was digital because it is a digital method. Well, it turned out to be a very good idea, because many years later, someone did get a patent, a fellow named Ron Bracewell, when the Hartley Transform came along, which has a lot in common with the FFT, a basic idea of doing the factoring is there, and he described his algorithm and of course most of the description was simply the FFT. Would you like me to go on with another part of the story about the patent?
Goldstein:
Sure, sure.
Cooley:
Well, this was much later, it must have been around ’87, I worked on a software package [for] an engineering and scientific subroutine library, for the new vector machine which was coming out. The day before the machine was to be announced, the lawyers called up and said, “Hey, you might be infringing on somebody’s patent.” The subroutine package, which was a very important part of the vector machine to have this library available. We had a very quick meeting and we went over there and they asked us what we had done before and what this has to do with the patent. Now, Ramesh Agarwal and I who had - we worked on this and in fact Ramesh did all of the programming for the FFT. We described what the original FFT and what we had done and how we had put it in the public domain and in fact one of the lawyers, Frank Tezerdian, was the lawyer who actually suggested doing the publication of the FFT idea.
Goldstein:
You mean originally or - ?
Cooley:
Originally, yes.
Goldstein:
Oh, I thought you said that was someone named Thomas.
Cooley:
No, he was the fellow who said that was patentable and called this meeting. He had since passed away.
Goldstein:
I see.
Cooley:
But Frank was on that committee too, and Frank was the one who suggested the strategy for putting it into the public domain. Ramesh and I had already read the patent, somebody had sent it to us, and we were well-prepared and we were able to show the lawyers and of course since Frank Tezerdian was familiar with it, we were able to respond to the lawyers who were setting up and issuing the software package. So this saved the day. We just probably would have held up the release of the vector machine, which was a pretty strategic thing for IBM.
In this case because the FFT algorithm was put into the public domain (through publishing in Math. Comp. 19 (1965), 297-301 (PDF available)) and implemented on a digital circuit (so they had a “device”), they could avoid patent infringement.