Tuesday 18 January 2011

Straw man rationality

Dan Ariely analogizes:
Here’s one way of thinking about this. Imagine that you’re in charge of designing highways, and you plan them under the assumption that all people drive perfectly. What would such rational road designs look like? Certainly, there would be no paved margins on the side of the road. Why would we lay concrete and asphalt on a part of the road where no one is supposed to drive on? Second, we would not have cut lines on the side of the road that make a brrrrrr sound when you drive over them, because all people are expected to drive perfectly straight down the middle of the lane. We would also make the width of the lanes much closer to the width of the car, eliminate all speed limits, and fill traffic lanes to 100 percent of their capacity. There is no question that this would be a more rational way to build roads, but is this a system that you would like to drive in? Of course not.
Assume rational drivers who have to exert effort to keep a perfectly straight track and with heterogeneous driving ability that affects the amount of effort that has to be exerted to keep a perfectly straight track. Add in random shocks to attention, like a kid that darts out from the side, wildlife, sun glare, a screaming toddler in the back seat or whatever, such that even the best driver devoting full attention to driving occasionally veers from a perfectly straight track.

The drivers are perfectly rational but have constraints. Optimal road design will incorporate things like paved shoulders and buzz lines to give more margin for error and feedback about errors. We could even want speed limits if drivers vary in ability and risk tolerance (both rationally) and if variance in speed contributes to accidents. All of these things build in room for error; without them, driving would be a horribly nervewracking experience: the slightest error could mean death. We'd all have to devote far too much effort to driving. I don't know why we'd need to assume irrationality to get any of the road features that Ariely seems to think can't be explained without it. Rather, I think it's a straw-man version of rationality that insists that we can never err.
What I find amazing is that when it comes to designing the mental and cognitive realm, we somehow assume that human beings are without bounds. We cling to the idea that we are fully rational beings, and that, like mental Supermen, we can figure out anything. Why are we so readily willing to admit to our physical limitations but are unwilling to take our cognitive limitations into account? To start with, our physical limitations stare us in the face all the time; but our cognitive limitations are not as obvious. A second reason is that we have a desire to see ourselves as perfectly capable— an impossibility in the physical domain. And perhaps a final reason why we don’t see our cognitive limitations is that maybe we have all bought into standard economics a little too much.
It's not irrational to err if there are information costs. It's not irrational to use heuristics if there are computation costs. We might sometimes pick the wrong heuristic, but we shift if we find a better one. It's no more irrational to pick the wrong heuristic when there are search costs over the domain of heuristics than it is to buy a car that costs $200 more than the one across town if there are search costs when buying a vehicle. The existence of price dispersion lets entrepreneurs profit by building price comparison websites; heuristic dispersion lets entrepreneurs profit by writing self-help books or working as life coaches.

And it's not irrational for folks to do things in maximizing their own utility functions that would be downright silly if they were trying to maximize mine.

5 comments:

  1. There's a bit of tension between this post ("All of these things build in room for error; without them, driving would be a horribly nervewracking experience: the slightest error could mean death. We'd all have to devote far too much effort to driving.") and your masthead. (I'm not saying they're incompatible, mind!) Safety margins are basically good, but with rapidly diminishing returns as the margins exceed the constraints for which they were designed to compensate.

    ReplyDelete
  2. I don't actually advocate putting in the spike, Blunt. Although I might, were I exclusively a pedestrian.

    ReplyDelete
  3. On the way south from Christchurch they have the road rumbles [the rumbles go brrr sound] so that if you drift off to the left you get a warning.

    Stay close to the centre, drift off to sleep, cross the middle line, hit a truck, never wake up, no warning.

    ReplyDelete
  4. Lesson is to lean left. US interstates have nobly things between lanes, reflecty too.

    ReplyDelete
  5. Ariely is terrible when it comes to picking a definition of rationality and sticking to it. I read "Predictably Irrational" and wanted to puke.

    ReplyDelete