P2P No. 39 — What industry can tell us about good science
Reflections on the ELLIS Industry Roundtable@EDS2023 Helsinki
At the end of August, I was participating in the ELLIS Doctoral Symposium, which is the student-organized annual event of the European initiative ELLIS (European Laboratory for Learning and Intelligent Systems—if you would like to pursue a Ph.D. in AI, robotics, and the like, the most recent call for applications can be found here). Besides indulging in photography, reconnecting with friends, and getting to know the new cohort of ELLIS students, I was after (as usual) to figure out how to be a good scientist.
Below, I report the insight from the industry roundtable. Yes, even industry can teach us to become better scientists. With the increasingly many research and applied scientist roles, the knowledge sector highly relies on PhDs since many roles. But what do they want?
What are the traits most sought after in the knowledge sector?
Though the incentives and structure of how scientists conduct research vary between academia and industry, science is the same. Thus, highly wanted traits in the industry can guide us on what to improve on. As the industry experts elaborated, they want candidates who:
have learned how to think
can communicate clearly to a broader audience
have their own ideas
Learn to think
During my studies, I often wondered why people with a degree (maybe even Ph.D.) in field X are hired by companies in field Y. What I did not realize is that those degrees signal at least as much how you work (i.e., transferable/soft skills) as what you work on.
Nonetheless, the art of doing science is usually only a byproduct of doctoral education. Completing a PhD will probably correlate with above-average thinking capabilities, but it does not necessarily imply that. You might say that learning to think is often only an implicit part of the curriculum. Thus, choosing the right people to work with becomes even more important, suggesting that by choosing your colleagues wisely, you can learn skills far beyond those related to the specific field.
You should look for transferable skills if you do not know what life throws at you. This leaves the options open—even if you are entirely sure that you have found your true calling at a given point. In a disruptive field such as AI, you cannot be so rigid as to fall in love with a single field or method.
Learning to think also helps you in life. Let it be a back-of-the-envelope calculation to filter out too-good-to-be-true opportunities or to question accepted but potentially outdated everyday practices.
Have your own ideas
When you start research, you most probably join a project or work on an idea passed on to you by your supervisor. This provides the context to develop expertise in a field and simultaneously learn how to research without the burden of developing a feasible idea. With an analogy, in the beginning, you are given fish (research question) to prepare as dinner. In the process, you will learn how to make a net (expertise) and figure out where to cast it (formulating your ideas).
This leads to developing autonomy, which is essential to leading a research effort—whether as a postdoc or a research group leader in academia or as the chief scientist in industry.
Communicate clearly
Academic projects often revolve around small groups of experts, where everyone is immersed in the state of the art. In that bubble, everyone speaks in jargon and knows the same (nontrivial) facts, often leading to the curse of knowledge. When you wonder why the reviewers did not understand your paper or are confronted with a sea of blank stares, remind yourself that you need to spell out things that seem trivial to you. Because probably they are not to a more general audience.
Professor
, writer of , proposes the mental model of The Audience Onion to make yourself aware of the curse of knowledge and to devise a strategy to counteract it. Being aware is half the success; the other half is simply making your points explicit and appropriate to your audience.Insights from negotiation theory can also be helpful here. Take the case of a court trial: as a plaintiff, you could try to convince the defendant, but that's not the point! Depending on the legal system, your arguments should target the judge/the jury.
It is attributed to Nobel prize-winner physicist Richard Feynman to advise people explaining complex ideas to five-year-olds. If you succeed, then (almost) everyone will understand it.
On a side note, people often mistake simple for easy. I have heard stories about struggling to publish simple research ideas because they seemed easy. They were not. They were simple, elegant, and insightful, but not easy. If they were, people would have already thought about them.