Industry is where the research data is

Over on Facebook, Lionel Briand commented on how hard it is for an academic research group to compete with industrial organizations on research. This is certainly true for hot topics. Right now, with a burning hot AI summer, the research world is split into the have and have-nots (in terms of GPU resources), and this is a key factor in who gets to innovate and how fast. It does not hold true in general, though.

Competition vs. partnering

A first comment is to partner, not to compete. We have different goals: Industry needs solutions, typically for enhanced products, while academia wants insights for publishable papers. In software engineering, we are performing research to the benefit of industry, among other things, but we can’t recreate industrial situations in a research lab. There are no 1000 developers that I could employ to develop a product line. I have to go to industry to learn about such situations, hence, we need to partner. Industry is where the research data is, as I like to say.

Academia is in a good position to partner with companies. Often, we can work in an anonymized fashion across multiple companies, building a theory better than any of the participating organizations could do themselves. Our independence can afford us a position of trust.

Inner source, the use of open (source) collaboration principles for firm-internal development is a good example for this: To understand how inner source works, we need to collaborate with many different case study partners, and no single one could have a final individual answer.

Point solution vs. theory

Industry’s goal is to win in the market and hence the focus is on better products, not theory. Theory is only relevant to the extent that it helps building better products. For that reason, industry is often satisfied and stops when something works; a deeper theory is rarely of interest. The following graphics from my research course Nailing Your Thesis illustrates this.

Industry is happy once it knows how to get from A to B. Academia or research for the sake of general knowledge, however, needs to figure out the coordinate system x and y, i.e. build a proper theory. The theory is of interest to industry too, to learn to go from A to B in a straight line, but often industry does not have the time to develop the theory themselves once they have a point solution in place.

Innovation vs. research

A variant of the previous argument is that industry wants to innovate and create something new, typically in support of new or better products or services. Performing the analytical work of theory building is an afterthought. Inventing something is not the same as performing research to understand why the innovation actually works, how it can be generalized, and where the limits are. The later is often left to academia.

Inner source is a good example here as well. Process research requires broad input, but does not provide product innovation.

How industry focuses

Even the largest companies are resource constrained. They therefore prefer to focus their resources on where they expect the biggest bang for their buck. This is typically in product development. New or better products make money, saving costs is often less relevant. As an academic researcher, you can therefore avoid the competition by focusing on topics industry needs and wants but prefers not to invest in.

Companies want inner source, but don’t want to develop theory in-house but rather hire consultants, or action researchers.

Off mainstream

Finally, if you don’t like competition and reviewers who shoot you down just because they think you are trampling through their garden, you should simply avoid where industry and others focus their work and choose something different. That’s the benefit of academia. Of course you should choose something that will be relevant, but that others’ haven grasped yet.

Did I mention inner source?