I’m a trifle perplexed by how perplexed Mark Liberman is by Things that are rarely better than they normally are. He quotes Matthew D. LaPlante of the Salt Lake Tribune in an article called Interpreters in high demand in Iraq as saying “It is rare, Hamblin knows, for these kinds of situations to end better than they normally do.”
Now, the reporter might have had in mind a distribution of outcomes for such situations in which it's literally true that a very small fraction of the results are above the mean (or some other interpretation of the "normal" outcome). However, I don't think this is what he meant.Why not? It’s the obvious meaning. At least, once you get away from statistical normal and back to lay normal, meaning something like expected or usual. There doesn’t seem to me anything Escherian about the sentence, even if it is a trifle awkward.
Look, let’s suppose that four out of five weekdays I eat two slices of pepperoni pizza at the local pizza joint. On Mondays, my schedule doesn’t allow me time to eat there (or the pizza joint is closed on Mondays, or some other reason that fits the supposition), so I usually just get a doughnut or a bag of chips and eat at my desk. Once in a great while, however, I will get a salad. Maybe one Monday out of five. One lunch out of twenty-five. Now, I think we can say that I normally eat pizza for lunch, but that I often eat a lunch less healthy than that. It is rare, however, for me to eat a healthier lunch than I normally do. Yes?
If there are a handful of outcomes that arise from a situation, and one occurs far more often than any of the others, that one is the normal outcome, and if it happens, the situation worked out normally. If one happens far less often than the others, that one is rare, and the situation rarely works out that way. If the rare outcome is the only one better than the normal one, we say that it is rare for these situations to end better than they usually do.
In the specific case, Mr. LaPlante is talking about a situation where a US patrol approaches an Iraqi civilian in a car that has been modified. I don’t know anything—anything at all—about the likely outcomes in such situations, but let us surmise that the usual outcome is an arrest, a frequent outcome is violence, and a rare outcome is “handshakes and smiles”, and Sgt. Hamblin knows this. Why, then, is the sentence confusing?
In fact, Mr. Liberman says that he thinks that Mr. LaPlante wanted to convey that such a situation usually ends badly, and that such a situation rarely ends well. Well, he does convey that. More than that, he conveys that there is a normal ending, and that the normal ending is pretty bad, and that frequently when the ending is not normal, it’s even worse. That may be false, but it is what it conveys, and I don’t see any reason to believe that Mr. LaPlante did not want it to convey exactly that. I think the perplexity arises because Mr. Liberman does not infer that the second part that I do, that is, that the second-most-common outcome is even worse than the usual outcome.
The note then goes on to mention the average muddle, which is a serious muddle that reveals people’s discomfort with statistics. It is true that people mistake average for normal or expected or usual, and are then confused in their expectations. In an area with a rainy season and a dry season, for instance, the average monthly rainfall is not typical. The average, oh, college freshfolk’s textbook budget may be typical, but I doubt it; there are a lot of frosh who buy nothing, and a lot who buy everything. If, say, gas prices go up in one region, the average price of gas will go up even if most drivers pay no more than they did. Averages alone tell you little about what is normal. But normal doesn’t tell you much about what is average, either. In situations where one outcome is very common, I would think it is normal for it to be rare that the outcome is better than normal.
chazak, chazak, v’nitchazek,
-Vardibidian.