Discworld takes a more practical view, there, a thought experiment is one that you can't do and which wouldn't work if you could. But the kind of thought experiment we have in mind is one that scientists carry out all the time, usually without realizing it; and you don't need to do it, because the whole point is that it wouldn't work. Many of the most important questions in science, and about our understanding of it, are not about how the universe actually is. They are about what would happen if the universe were different.

Someone asks 'why do zebras form herds?' You could answer this by an analysis of zebra sociology, psychology, and so on ... or you could ask a question of a very different kind: 'What would hap­pen if they didn't?' One fairly obvious answer to that is 'They'd be much more likely to get eaten by lions.' This immediately suggests that zebras form herds for self-protection, and now we've got some insight into what zebras actually do by contemplating, for a moment, the possibility that they might have done something else.

Another, more serious example is the question 'Is the solar sys­tem stable?', which means 'Could it change dramatically as a result of some tiny disturbance?' In 1887 King Oscar II of Sweden offered a prize of 2,500 crowns for the answer It took about a century for the world's mathematicians to come up with a definite answer: 'Maybe'. (It was a good answer, but they didn't get paid. The prize had already been awarded to someone who didn't get the answer and whose prizewinning article had a big mistake right at the most interesting part. But when he put it right, at his own expense, he invented Chaos Theory and paved the way for the 'maybe'. Sometimes, the best answer is a more interesting question.) The point here is that stability is not about what a system is actually doing: it is about how the system would change if you disturbed it. Stability, by definition, deals with 'what if?'.