The data science website 'FiveThirtyEight' just published a new 'What's the Point' podcast about the limits of Drone Warfare, and especially the difference between tactical tools and strategic thought. Their guest is William Arkin, a former intelligence minder and current journalist. The prompt for the discussion is Arkin's new book, Unmanned: Drones, Data, and the Illusion of Perfect Warfare. an awkward title with an important theme.

The gist of the discussion is that the U.S. security forces have, with the aid of drones, amassed mountains of data which they comb through to find targets to destroy. The United States has devoted tremendous resources to this ongoing project, and it has led to a large number of deaths, and a comparable amount of controversy among ethics watchdogs. Drone strikes though, compared with 'boots on the ground' soldiering, consumes relatively few resources, does not immediately risk the lives of U.S. soldiers, and so receives little media attention. The film Good Kill and the play Grounded explore the moral costs of these issues for those who pilot and command drone operations, but little time is spent understanding the conflicts that we enter.

Arkin argues that an investment in drones cannot displace an intensive investigation of the purpose and end-goal of the conflicts we enter. Good data can lead to good knowledge, but when that data lacks a coherent framework to understand it, it may simply help perpetuate the slack fighting that we have been involved in since September 2001. It's true that drones have created a tempestuous legal debate, and discussions about the morality of assassination. But these debates lack the purposeful depth of strategic discussion.

Why do our leaders want us to engage in violence on the far side of the world? Can the violence we are capable of actually help us achieve the objectives laid down by our leadership? Is the public fully informed of the costs and dangers of entering such a conflict? Can we achieve better outcomes using different tools, including diplomacy, or an alternative military action? How do people interpret our drone strikes, both those that succeed, and those that fail?

When data displaces debate, we silence the most important questions, and replace strategy with rote task management.

This is not, of course, a new trend. It can occur whenever technology or workforce expertise permit rapid gains in data gathering and data analysis. As Lawrence Freedman describes in Strategy: A History (a wonderful read and door-stopper of a book), former Secretary of Defense Robert McNamara led a similar data-driven revolution during the 1960s, one that encouraged tremendous introspection and self-justification within the Pentagon, but also one that failed to consider the dark spots in American knowledge and strategic thinking. Pentagon leaders spent more time justifying their beliefs than they did altering them. Here in lies the hard problem: smart thinking does not necessarily lead to strategic thinking. Strategic thinking, even when it takes place, may very well fail due to the overwhelming complexity of enforcing one's political will against an opponent. 

Given that strategy is, at best, a coping mechanism rather than a master stroke, it is tempting to fall into the bureaucratic trap of pursuing further drone attacks as a good unto themselves. Commanders, not knowing how to win the war, may therefore settle for continuous attacks via drone.

Drones (of course) will not go away. Within the military, I suspect that in a few years time they will be  relegated to company and battalion command: a more efficient version of mortar fire. Drones will also likely assume a dominant role in domestic policing--a quiet, efficient accompaniment to traffic patrol and police helicopters.

Drones may also encourage a new crop of useful close air support aircraft to replace the A10. Ground troops have been clamoring for years for air support that flies low and slow, and (unlike Apache helicopters) are not such expensive toys that they cannot afford to take risks in heavy ground combat.

But tactical weapons are not the same as strategy: the pursuit of strategy requires a steady, moderate march disciplined with human dignity and a reasonable hope for peace.

Kenneth R. Koedinger, Julie L. Booth, David Klahr. 'Instructional Complexity and the Science to Constrain It.' Science. Vol 342, 22 Nov 2013, Pg 935-937. <Link: https://www.sciencemag.org/content/342/6161/935.full>

My review of James Lang's On Course emphasized the practical attributes of successful teaching. My review of Mark Edmundson's Why Teach? examined current trends in academic life, and pleads for a more spiritually challenging curriculum. On Course uses some references to empirical data and peer-reviewed studies, but seeks mainly to provide firm guidance to young educators. Why Teach? eschews peer review to trace one intellectual's Walden-like musings on the modern intellectual wilderness. My sympathies lay with Edmundson, but my need for practical guidance brought me closer to Lang. 

As luck would have it, this week's issue of Science included an article on education research. As with most articles in Science, the purpose of the piece is to establish what exactly a particular field can and cannot say based on the data available, and then to suggest new avenues of research. When it comes to human learning, these researchers find the entire field a chaotic mess. The authors attempt to build a few categorical fences to clarify things for both researchers and educators, but also use a little math to show the complexity of the issue. 

Here is the math they use to demonstrate complexity. "If we consider just 15 of the 30 instructional techniques we identified, three alternative dosage levels, and the possibility of different dosage choices for early and late instruction, we compute 3^(15*2) or 205 trillion options... the vast size of this space reveals that simple two-sided debates about improving learning--in the scientific literature, as well as in the public forum--obscure the complexity that a productive science of instruction must address" (936).  

In other words, the potential for research is almost overwhelmingly vast, and conducting it might prove overwhelmingly expensive. To encourage research rather than surrender, the authors suggest a framework within which future studies might operate. You can read their article if you wish to fully understand their suggestions, but I will briefly examine one aspect of their paper.

The authors produced a helpful table that reviews some of the current findings on 'instruction.' Much of the table draws on Koedinger, Corbett, and Perfetti's learning-to-instruction theoretical framework. Below, the authors of the Science article fit thirty nifty principles within three categories the authors call 'the functions of instruction' (936). 

The reason I find this table helpful is that one can interpret its arguments as practical advice. (Aside: While the authors do not explicitly limit the applicability of the graph to teaching mathematics, I suspect that such a limitation might exist for at least some of the principles). To my remembrance, the authors' findings do not greatly vary from the methods suggested in a typical mathematics textbook. That's probably because human educators have been in the business of understanding human learning for a very long time; when an educator fails, it can be because either they or the student do not care whether learning occurs--or perhaps they are not in a position to care due to environmental factors. Anyhow, the chart is comforting, it feels like smoking a cigarette in snowfall. 

Some books should be articles, and some articles should be books. The lead author, Koedinger, has published extensively in his particular field (he earned an MS in Computer Science and PhD in Cognitive Science), but has not written a book on the subject. A book could include practical examples of his work, and show how he stumbles into and out of research problems. It would also give him the room to demonstrate that his research is useful to educators, as opposed to researchers that scientifically examine education. It is common scientific practice to laconically claim one's work has 'policy implications,' and yet have absolutely no significant evidence of one's ability to implicate, change, or effectively argue policy. Scientists should perhaps engage in politics more often than they do. 

God knows I probably could have benefited from some of Professor Koedinger's knowledge when I myself haltingly studied geometry. "To the field, Koedinger, to the field."

                                                                       *****

Clearly, we are far from 'perfecting' college instruction. Human limitations represent a permanent obstacle to ideal outcomes. In the face of those limitation, it is sometimes helpful to step back and use qualitative empirical studies to think through the purpose and effectiveness of college education. In this vein, the books of Ken Bain (president of the Best Teachers Institute) are helpful. Bain wrote a book entitled What the Best College Teachers Do (2004), and another called What the Best College Students Do (2012). The former book reported the findings of a fifteen year study that followed the habits and practices of some of the most successful educators in the United States. Not all together surprisingly, he found that the best instructors combined a deep knowledge of their subject with a belief in student learning. These two principles seem easy, but they are not easy to accomplish. Many (most?) teachers are out of touch with the most challenging questions in their discipline, and many others lack sympathy for the importance of undergraduate education.