Little Black Boxes

What is a black box and why should you care?

The concept is simple, we input data into a “black box “and we receive an output. We then take these outputs and apply them to planning, analyses, or designs. These black boxes are often represented as models. They can be as simple as an excel spreadsheet, or as complicated as a three-dimensional hydrodynamic model. The question is, do you know what that little black box is doing?

I believe there is a responsibility to understand “the man behind the curtain”. Engineering is essentially applied physics. When working on stream restoration problems we are practicing applied geomorphology, among other disciplines (ecology, biology, ect.). Do we need to know how to derive the two-dimensional flow equations or how to do a Taylor Series? Probably not. When I am applying hydrodynamic models should I understand that flow equations are made up of terms used to describe physical forces like gravity, Coriolis force, viscosity (turbulence)? Probably yes, but of even more importance is to understand what models use what terms and why those terms may be important or unimportant for my needed application.

Another way we encounter black boxes is in the form of empirical equations themselves. Take bend scour for example. Which equation are you using – Zeller, Maynord, Thorne, NEH? Was the equation developed in a lab or was it derived from field data? Is the equation based on a best fit or is the curve adjusted to envelope all the data points? What are the applicable parameter ranges for the equation?

I also think of little black boxes as a single point analysis - one method gave me one answer. This can be highlighted by thinking about channel forming discharge, or bankfull as it is often described. How did you determine it? Was it based on a day in the field looking at vegetation, ordinary high-water mark, landscape changes? Did you also look at nearby gages and recurrence interval? Do you understand the watershed conditions which may make the gaging station applicable or not?  Did you look at effective discharge? Do you understand which sediment transport equation to use to develop a rating curve? Do you understand how hydrology and flow record can affect effective discharge? Which methods correspond with one another and which is more trustworthy and applicable?

Peeling back the layers to understand black boxes and to understand first principles is not easy. Our jobs and the projects we work on are not simple. We deal with complex natural environments. Because of this, it is important to avoid the black box, to dive deeper and to gain an understanding of natural processes and an understanding of the tools and methods we use to understand or modify those natural processes.

With computer tools becoming easier and simpler to apply it is easy to fall into this trap without even knowing it. I am constantly reminding myself to avoid the black box and to dig deeper to understand what consequences my choices (inputs) have and if those choices are even applicable for the setting. Sometimes it can be as simple as RTDM (read the damn manual!). Regardless of the how, we all owe it to ourselves and our profession to avoid that little black box.