I went for a long run on Saturday. I was tired of my normal routes, so I explored some new paths by running down roads I have never visited. This is sorta like an online longest path algorithm with a side constraint of enjoyment. Since running up and down roads that dead end are no fun, I specifically look for “lanes” and “drives” rather than “courts” when trying to identify through streets. I quickly learned that most subdivisions in my area have just one or two through streets and all other streets dead end as cul-de-sacs off of the main thoroughfare, whether the streets were labeled as drives, lanes, avenues, or courts. This was really frustrating. I repeatedly ran down “drives,” optimistically wondering where they would take me, only to discover that they took me nowhere, and I had to turn around and run back. What a running buzzkill.
As I was mentally cursing at the “lanes” and “drives” that should have more accurately been labeled as “courts, ” I thought about the importance of labels, classifications, and naming conventions.
I also recollected a conversation I once had with a psychologist friend of mine about cheesecake. In this conversation, I said something about cheesecake that implied that it was pie. He replied, “Most people feel that cheesecake is cake, not pie.”
I was surprised and argued, “But it has a pie crust in it, it is served in wedges like pie, and it is often served with sauces and fruit in the same manner as pie.” As I was arguing my case, it suddenly dawned on me that cheesecake has “cake” in its name, and that is precisely why it is often classified as cake rather than pie.
My friend explained that this is well-understood by psychologists. People classify things according to name rather than structure. I was classifying by structure in the case of cheesecake (yeah!) Not that I am more evolved, as I indicated with the case of my run, I kept falling for the meaningless labels of streets in my area.
How do you think that names and labels impact operations research? Quickly, methodologies are labeled according to discipline (“simulation,” “optimization,” “decision analysis”), which depend on the structure of the tool applied to the problem. This is good. We use a lot of labels. Some may be out of control as they are math or technical terms that are widely used. Others we generate ourself, like paper/report names or keywords. What OR labels should be improved?
On a related note, you can use bad naming conventions to get kids (or adults!) to eat their vegetables.
Related post: how to pick a good baby name.