smokerblog

Acceptable Risk

I had an interesting discussion today at work with our IT Director about our email spam filter. It seems that people were complaining when critical emails weren't getting through because of stringent filtering settings. The perception is that email should be 100% reliable at all times, but this alternative would lead to inboxes stuffed with potentially virus-laden spam, itself an extreme inconvenience. So where to draw the line between an acceptable amount of delayed or deleted important emails vs. the convenience of spam- and virus-free workstations? Should we strive for an email system email that is safe, secure, and 100%-reliable? What does 100%-reliable mean? Are you willing to risk your life based on the outcome of an email transfer? How about $1,000,000? How about a 1% chance of losing $1,000,000? How much are we willing to spend in order to make it so?

Actuaries literally make a career out of these kinds of statistics-crunching risk evaluations. In everyday life, we all have an inner actuary buried somewhere in our brains that also makes these decisions for us, so that we don't have to sweat the details. I have no evidence that tells me that the tap water flowing into my kitchen sink is not laced with poisons, yet I drink it when I get thirsty. My inner actuary determines this is okay based on previous experience and that to test every single glass I drink would be an inordinate hassle for an infinitesimal risk. Trouble is, that inner actuary is often easily swayed by emotion, mob mentality, or fermented beverages.

My favorite example along these lines is from a book written by one of my college professors. In the appendix of his book Retreat from Doomsday: The Obsolescence of Major War, John Mueller poses a hypothetical situation:

Suppose an engineering firm came up tomorrow with an amazing new form of transportation. People would step inside a booth, dial a location, and then be taken apart atom by atom and transmitted over wires to the desired location where they would be reassembled. After thoroughly testing the new device for safety, the firm has concluded that the overwhelming majority of trips would be utterly without incident...Unfortunately, in a very tiny percentage of trips, things would go wrong and the traveller would never rematerialize. Injuries from minor contusions to paralysis would also occasionally occur. The total: probably not much more per year than 50,000 deaths and 2 million or so disabling injuries...Should we install a system with costs like that?

Of course, except for the "transmitted over wires" part (how quaintly 1980's), Prof. Mueller is talking about the private passenger automobile. Every day people risk their lives by climbing into their cars to go to work, visit friends, or pick up some milk, yet more people are afraid of flying than driving in a car despite the best statistics. We are also regularly reminded of the tragic human cost of the war in Iraq, yet more Americans die in a two-week period driving their cars than died in Iraq since the war started.

This is where our inner actuaries fail us. We tend to ignore statistics and often follow emotional cues rather than logic. We Americans love our cars, we love the freedom cars provide and we love to drive fast. And all that is fine. But as long as we continue to insist on this freedom, we should be mindful of the price we are paying.