Software can be a primary product of scientific research

Once again, I’m writing in response to Titus Brown, who answered Is software a primary product of science? with “no” in his blog and on Twitter:

Q: Is software a primary product of research? (A: No.) http://t.co/f5y13rAFj2 @PBarmby #longread #sorry cc @fperez_org @danielskatz

— Titus Brown (@ctitusbrown) April 22, 2015

Note that Titus cleverly changed the question from asking about research to asking about science. Perhaps we can combine these and think about scientific research.

So, what are primary products of scientific research? I think most would agree that books and journal papers qualify, potentially along with book chapters, and conference papers, depending on science domain. So, writing text seems to be accepted as a primary product of research.

But this isn’t quite right: it’s not any text that counts; the subject of the text also matters. Generally, the text should be about a hypothesis and proof about the truth or falsity of the hypothesis, though it can sometimes also be a survey of previous work (adding new insight) or a position statement.

So what is the product? Is it the text, or the underlying knowledge gained? I think it’s the knowledge, as represented in the text.

How else can we represent knowledge?  In data? In software? In tools or instruments? The answer is maybe. Much like text, all data, all software, all tools, all instruments, etc. are not products of scientific research. But some are.

It may be worth considering the (translated) words of Alfred Nobel who left a large part of his estate to be given as, “prizes to those who, during the preceding year, shall have conferred the greatest benefit on mankind.” Reading further, the language that describes the criteria for the prizes in different fields varies, as follows:

• physics: “most important discovery or invention”
• chemistry: “most important discovery or improvement”
• physiology or medicine: “most important discovery”
• literature: “most outstanding work”
• peace: “most or best work”

So, could a Nobel Prize be given based on software? It appears that the answer would be yes in physics or chemistry, but no in medicine. (This is assuming that the software is not a discovery itself, which might be the case in computer science, but that’s not a field Nobel knew about.)

If it’s worthy of a Nobel Prize, does that make it a primary product? Probably yes.

Is all software potentially worthy of a Nobel Prize? Certainly not.

Is all software a primary product of scientific research? No. But a lot of it, particularly software that is intended to be used by others, is. This type of software encapsulates knowledge and methods, and implicitly demonstrates the proofs of hypotheses.

Is this a clearcut distinction? No.

I admit that I mostly agree with Titus’s post about software’s importance and role in scientific research, including “better software, better research!” but I’ve weighed the same evidence and come to a different conclusion.

Is software a primary product of science? Overall, I think there is enough software that is a primary product of scientific research that I can answer yes.

Disclaimer

Some work by the author was supported by the National Science Foundation (NSF) while working at the Foundation; any opinion, finding, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the NSF.