There are moments in life which define every moment thereafter. Such moments may seem important and consequencial even as they happen, but the scope of that river delta isn’t apparent until you’re stuck in the fly and leach infested swamp downstream. One such moment in my life entailed a technical dispute which informs all my evaluations of “expert” opinion, including my own.

At the time, I was working my first job. I had only scratched the surface of the technical knowledge necessary to practice my profession (which, due to its breadth, I will never master), when I found myself arguing with a slightly more senior employee the correct loading for a steel girder forming the hip of a roof. While there are numerous complications to calculating the design load, our dispute concerned the load magnitude of the vertical component of the varying uniform load along the length of the girder.

Structural engineers usual approximate the weight of individual closely spaced joists upon a supporting girder as a continuous load – as if the joists did not apply load at discrete locations along the length of the girder but uniformly and evenly without regard for the unloaded distance between joist connections to the girder. This, many joists spaced two feet apart which each impart four hundred pounds of weight onto a girder are approximated to impart a continuous load of two hundred pounds per foot of girder length.

If the span of the joists is oriented at a right angle to the span of the girder, then the approximation follows geometric patterns structural engineers repeatedly observe, and so the load calculation becomes, over time, second nature. However, complications may occur.

A common, and fundamental, variation occurs when the girder is oriented at an other than right angle to the joists. When that happens, the right angle spacing of the joists does not correspond to the distance between connections of the joist to the girder. Thus, the four hundred pounds load at the end of each joist may be spaced four feet apart at the girder, imparting a uniform load half that if the girder were at right angle to all joists.

Such was the dispute between us at my first job. My colleague didn’t recognize the difference resulting from the girder orientation, and so we escalated the argument to our employer, an engineer with as much experience as I have now. He also failed to recognize the consequence of the girder orientation, and asked me, “Which approach is most conservative?” I replied, “His [my coworker’s] approach.” And so that is how the design was performed.

The flaw with this arbitration is that it ignored the correct assessment of reality for a compromise which rewarded the incorrect idea at the expense of the customer. That was the day I quit my first job – not literally that day, but the moment I knew I would soon be employed elsewhere.

I recalled this moment from more than a quarter century ago while reading an “expert” opinion I know to lack correlation with prior proven theory and available data. The “expert” opinion advocates a conservative approach, ostensibly to avoid risk. But the dirty secret about life is that we are unable to avoid risk. One duty of engineers is to understand the world so that they can predict the future. I’d say that’s why we’re licensed: we predict the future. Yet we all predict the future, else nobody would buy breakfast for tomorrow.

My boss, decades ago, chose an option less risky because he wouldn’t (or couldn’t) predict the future. Some people will respond, “Nobody knows the future,” and I’ll admit, I don’t know the future, nor do I trouble myself worrying whether the sun won’t rise over the horizon tomorrow, or how I’ll cook my eggs one hundred years hence; neither is likely. Practical risk assessment isn’t accomplished by preparing for every worst scenario – even at the aggregate where the actuarial tables tell us somebody will misuse the welding torch or succumb to an innocent toddler’s cough.

Assess risk and act according to what you anticipate will result, not according to what you fear or hope will result.