At last week's Microsoft Professional Developers Conference in San
Francisco, keynote speaker Douglas Adams told this story of a scientific
study about dolphin intelligence:

"The researchers wanted the dolphin to jump out of the water and make a
certain noise, and then they would reward it with a fish. At first the
experiment seemed to be working. The dolphin would jump out of the water,
make a noise, and get the fish. Then the dolphin's performance started to
degrade. The dolphin would jump out of the water and make a different
noise, or apparently make no noise at all. So the data seemed random, and
the experiment didn't seem to be working.

"Then the researchers took a look at the sound recordings they had made. It
turned out that the dolphin was indeed making a sound each time it jumped
out of the water, but sometimes the sound was below the human hearing
threshold. And there was a pattern to the variations. It turned out that
while the scientists thought the experiment was about getting the dolphin
to jump out of the water, make a noise, and get a fish, the dolphin was
using the experiment to calibrate the range of human hearing."