The Annals of Improbable Research, also known as AIR, is many things. First, AIR is a science humor journal. Now, hearing that, you might be tempted to toss this book aside, because maybe:

a) you don’t like science and won’t understand the book; or

b) you love science and know that science is too important to let people laugh at it.

Either way, you might be right. But I doubt it.

The Best of Annals of Improbable Research  is not very random sampling of some of the juiciest bits that have appeared in the magazine.

Paperback: 224 pages

Publisher: W. H. Freeman; Reissue edition (Sep. 15, 1997)

Language: English             ISBN: 0716730944

Dimensions: 10.9 x 8.6 x 0.6 inches; Shipping weight: 1.2 pounds.

Computer models predict that, if global warming is really happening, temperature in the troposphere should rise along with those on the surface. Recorded surface temperatures are, indeed, rising. However, both data from weather balloons and observations made by satellites suggest that temperature in the troposphere have remained constant since the 1970s. Over the tropics they may even have dropped. This counter-intuitive result has caused sceptics to question how much warming, if any, is actually going on.

There are, of course, three possibilities. One is that the sceptics are right. A second is that the models are wrong. And the third is that there is something wrong with the data. Three papers published in this week’s issue of Science suggest that the third possibility is the correct one.

Heat and light, Economist, August 13th-19th 2005.

Not exactly: data are never wrong, data are simply the result of the measurement performed, statements accepted at a given value (face value). However, interpretation of the data (i.e., assignment of the data to a particular physical quantity in the computer model) could be wrong. As I have learned a long time ago, in the freshman physics lab, you may encounter a problem with the measurement setup and procedure, but that you can master; the real problem is to understand what you have actually measured.

Data by themselves are not information. Information has to be extracted from the data. Data by themselves may have no meaning, and only when interpreted by a set of epistemological criteria (some kind of data processing pattern) data may take on meaning and become information. For example, voltage readings of a thermocouple are interpreted as temperature while voltage readings of a pressure transducer are interpreted as pressure; yet, they are voltage readings, all the same.