The practical value of basic PhD research

There is a growing awareness about the downsides of pursuing a PhD. During my time as a PhD student, I was regularly reminded about the challenges of the academic job market and about the doctorate’s poor financial returns on investment.

Though there are also contrary views about the degree’s value, these views often point to the practical research skills that PhD students develop and their relevance to industry. Today, PhD students are often encouraged to conduct applied research, so that they can easily communicate and transfer the results of their work.

One of the overlooked practical benefits of the PhD degree is that it offers students the opportunity to conduct curiosity-driven basic research. PhD students often have the option to focus on answering fundamental questions about why or how the universe is, the answers of which might not be easily applied to a particular industry. Counter-intuitively, I argue that basic research has economic benefits, and that these benefits should be considered when assessing the practical value of PhD studies.

Risk and uncertainty

My argument is rooted in the difference between risk and uncertainty, as originally articulated by economist Frank Knight. Risk concerns situations when we do not know the outcome of a situation but can measure the odds. For example, researchers might not have discovered the best artificial intelligence algorithms for predicting air quality, but can predict that a solution is possible; the odds of success are known.

Uncertainty on the other hand occurs when the odds of success are unknown. Basic research pursues questions with uncertain outcomes and unpredictable practical value. This often leads to the perception that money spent on basic research is wasted. However, the impacts of basic research are occasionally great, and paradoxically have huge practical applications.

One example was the research of Canadian-English computer scientist Geoffrey Hinton at the University of Toronto. In the 1980s, Hinton conducted research in artificial neural networks, what were then seen as curiosities with little practical value. Today, neural networks are the backbone of a new generation of artificial intelligence behind technologies ranging from Apple’s Siri to automated cancer detection systems.

A second example is Canadian physicist Donna Strickland, who recently won a Nobel Prize for her work in chirped optical pulses. Strickland, whose Nobel Prize winning work begun as a doctoral student, later reflected on how it took at least a decade for practical applications of her research to come into view.

Strategies for fostering basic research

Both Hinton and Strickland conducted basic research early in their academic career which did not yield practical applications until decades afterwards. In both cases, the future practical benefits to society were unknown and could not have been known or quantified at the time.

PhD studies offer the benefit of full-time research, which can become scarce later in a scholarly career. The PhD can be a vehicle for conducting and cultivating curiosity-driven research—a privilege that is not experienced elsewhere in society, yet has unique benefits.

The potential practical value of basic research, especially at the PhD level, should therefore be considered by research policymakers. However, policymakers may find it challenging to finance research activities when we cannot easily quantify the outcomes.

We can take steps to maximize the effectiveness of basic research by drawing on the ways entrepreneurs and angel investors deal with uncertainty. For example, policies can emphasize supporting a large number of promising or interesting PhD projects, rather than offering more financing for the best ones. This is a common approach taken by successful angel investors.

Alternatively, policymakers can attempt to maximize serendipity, which has been identified as aiding unintended discovery. Encouraging collaboration of PhD students across or between disciplines could increase the chance of discovery. Interdisciplinary PhD programs might further offer a way to encourage students to tackle some of the pressing issues of today’s world, which are often interdisciplinary in nature.

Regardless of approach, we must stop viewing the value of a PhD degree as purely intangible and recognize the economic and practical value it plays in basic research. Though not all PhD students can be expected to become Hinton or Strickland, basic research at the PhD level enabled their success. It can continue to enable future generations of researchers too, if we allow it.

Reflections about PhDs on the eve of a defence

Despite my noble intentions with respect to this blog, I have been largely unsuccessful at adding content. There are many reasons for this, such as the aggressive teaching and publishing deadlines from last semester. However, a big part of the reason for this is my looming thesis defence, due to happen tomorrow at 10:00 am. I thought it would be fitting to reflect a bit on what it means to do a PhD and what the last four years have entailed, in hindsight.

Having done more than my fair share of university degrees, I can attest to how the PhD is quite different from the others. There are clear financial reasons for pursuing Masters or professional programs. However, in many disciplines PhDs often come with few financial rewards. According to the US Census Bureau, in many fields with robust professional programs (i.e. Business or Law), median earnings among PhD holders are lower than their professional counterparts (i.e. MBA, JD). When one considers that the opportunity cost of doing a PhD 3 to 7 years of productive labour, it becomes clear that the motivation for doing a PhD is often not financial.  Realistically, the PhD only equips students to do one thing: make a substantial contribution of research to the academic community in the discipline students have decided to pursue. Though some PhDs also require students to gain teaching experience, this is not mandatory in all PhD programs.  When it is mandatory, students could expect to spend hundreds of hours teaching a course or working as a teaching assistant. This is small compared to the thousands of hours spent cultivating research.

The best analogy of a PhD that I have yet encountered was written by Tad Waddington in Lasting Contribution. His quote reads:

The last step of the [education] process is to contribute to knowledge, which is unlike the previous steps. Elementary school is like learning to ride a tricycle. High school is like learning to ride a bicycle. College is like learning to drive a car. A master’s degree is like learning to drive a race car. Students often think that the next step is more of the same, like learning to fly an airplane. On the contrary, the Ph.D. is like learning to design a new car. Instead of taking in more knowledge, you have to create knowledge. You have to discover (and then share with others) something that nobody has ever known before.

When I first read this quote three years ago, it stuck with me. I had the good fortune of having pursued two master’s degrees before starting my PhD and had originally thought that the PhD would be like a more advanced version of the previous two. Looking back, I don’t believe that was the case, and agree with Waddington more than ever. If you are considering ever doing a PhD, I recommend that you should be the sort of person who enjoys spending a ridiculous amount of time and energy to make a difficult, small, yet very real contribution to human knowledge.

It’s been a pleasure to have the opportunity to pursue and cultivate interdisciplinary research which I feel truly does break down barriers between disciplines. It has been challenging and at times grueling, but also rewarding, and I believe I have come out a better person. I would like to thank everyone for supporting me through the journey. I wouldn’t do it any differently if I could do it all again.

Why MOOCs are bad and what we can do about it

Perna, L., Ruby, A., Boruch, R., Wang, N., Scull, J., Evans, C., & Ahmad, S. (2013, December). The life cycle of a million MOOC users. In MOOC Research Initiative Conference (pp. 5-6).

In 2011, Sebastian Thurn and David Evans had the bright idea of recording their Stanford University lectures and posting them on the internet. The initiative was incredibly successful and hundreds of thousands of students flocked over the broadband highways to learn about artificial intelligence from two of the greatest minds in the field. In fact, their original initiative was so successful that Thurn later left his job to fund Udacity, today one of the most innovative and disruptive forces in university education. By 2012, it seemed that the whole world was talking about Massive Open Online Courses (“MOOCs”) and that MOOCs were on the path to transforming how teaching was done forever by providing the highest quality education to everyone, everywhere, for free.

This story has a deeply personal element for me. Then an unemployed (or at times underemployed) philosophy grad, I had to make a decision about whether to go back to school to pursue yet another graduate degree, this time in computer science. I sometimes wonder what my life would have been like if I had decided to take a year off and gorge myself on an intellectual mash of Stanford videos on machine learning. I usually conclude that it would have been for the worse. I am very thankful that I ended up going back to school because I probably learned a lot more than I would have otherwise. In late 2013 and early 2014, a number of quantitative studies were published that were like a wet blanket over silicon valley’s burning fire for MOOCs. Researchers at Penn, for instance, found that as few as 5% of MOOC registrants actually finish their courses, while only a fraction of those attain high grades. What’s worse is that MOOC users were found to disproportionately come from educated, male and wealthy backgrounds, largely in the USA. So much, then, for the fad that was the MOOC revolution. Or so the story goes.

Why do MOOCs suck so much at teaching the people that they are trying to help? One of the many reasons is that they are not well-designed. Robert Ubell from NYU has been doing e-learning a long time and thinks that MOOCs suck because they were not designed to keep users engaged, like a good teacher would. Ubell points to active learning, a theory that getting students deeply involved in the learning process will produce better outcomes. For example, active learning holds that asking students questions during a lecture would produce better results because students are more deeply engaged in the process. By involving students, we can better keep their attention, which is one of the fundamental brain mechanisms governing learning. MOOCs suck at knowing when you are paying attention. Good teachers know this by the glazed look in their students eyes as their attention drifts into the mental netherworld between the classroom and PewDiePie’s latest embarrassment.

If we had a good way to measure attention, we would have a way of improving MOOCs. The problem is that scientists do not yet have reliable ways of measuring attention through a computer. Sure we can look at clicks or scrolls, but do clicks really tell you much when you are rewarded by faking it? Alternatively, we could ask you whether you are paying attention, but this will disrupt the course experience. This is why I am looking at brain data. If we watch people’s brains, we can reliably understand when they stop paying attention, and maybe build MOOCs that teach better. It’s a bold idea, but if it works, we could develop technologies that achieve this original vision: quality education for everyone, everywhere, for free.