In a recent post, I reviewed the way Nassim Nicholas Taleb analyzes common reasoning errors in his recently published Skin in the Game. My recent post focused on ‘the nth effect’ and the reasoning problems it can cause. Today, I want to look more into a second type of error – focusing more on actions instead of interactions.
This is a very similar concept to the nth effect problem but there are a couple of minor differences. One subtle difference is that nth effects tend to rely on a more traditional ‘cause and effect’ mode of thinking while actions-interactions consider the way an unseen or unconsidered factor might influence the way a cause or an effect would manifest itself. If the nth effect is a way to ask ‘then what?’, actions-interactions is a way to ask ‘what else?’
A good thought exercise that illustrates the difference in these kinds of thinking is to consider what might cause automobile fatalities to fall. The reasons centered on lowering the rate of automobile fatalities are good examples of nth effect thinking. A new driving instructor who is far superior to a predecessor will improve the overall skill level for all his or her students. This, in turn, leads to an increase in the general skill level among drivers that leads to safer driving overall and a lower automobile fatality rate. Other similar examples could start with a different initial event – a governor introduces a new bill to fix potholes or someone at Honda invents a better airbag – and asking a series of ‘then what?’ questions in line with nth effect thinking would lead to the same overall result – a lower rate of automobile fatalities brought about by improved road quality or better crash safety features.
A less obvious set of reasons keeps the rate of automobile fatalities constant and focuses instead on explaining why people might drive less than they did in the past. In other words, these reasons show that keeping the likelihood of a crash constant for any given trip can be irrelevant from the perspective of total automobile fatalities so long as the total number of trips is decreasing. These reasons are more about interactions than actions. If investment in public transportation rises significantly and people start trading in their cars for train passes, the number of car trips taken overall will eventually decrease. This, in turn, would lower the total number of automobile fatalities because there are less people on the road to begin with.
The tricky part of these two concepts is that it can be hard to tell when nth effect thinking ends and actions-interactions begins. This brings me to magnitude, the second difference between the two concepts. Nth effects tend to work linearly in comparison to actions-interactions and this means magnitude has a consistent effect on the outcome.The airbag example above works here because each improved airbag fractionally lowers the fatality risk. If the public benefit of installing improved airbags in every car is a 1% reduction in total fatality rates, then each improved airbag moves the total fatality rate closer to that 1% target in equal and proportional measure.
On the other hand, increased investment in public transportation does not necessarily have such a smooth impact. The magnitude of the investment is crucial because until a certain amount is invested there will be no change to driver behavior and therefore no improvement in the fatality rate. However, different levels of investment will cause the transit system to seem like a good idea for people in stages. A helpful way to think of this is by extending a train service - for each additional station build into the line, more people who live or work near the new station will trade in a car for a train pass. Unlike with the airbag example, however, there is no linear progression to these changes. If a station costs $200 million to complete, it isn't until that 200 millionth dollar is spent that commuters will make new decisions and the fatality rate will see any improvement.
Decision makers that consider the difference between nth effects and actions-interactions can find endless opportunities to maximize their time, money, and efforts. Though instinct suggests that the best investment of resources is always the one with the most direct impact on the desired outcome, recognizing the influence of hidden or subtle interactions can present much better alternatives for certain situations.
