747 Junkyard argument
The tornado argument made by creationists and proponents of intelligent design states that attributing the development of life to natural forces such as evolution through natural selection is like expecting a tornado moving through a junkyard to result in a fully functional Boeing 747. The argument was originally made by British astronomer Fred Hoyle. This analogy depends on a fundamental misunderstanding of the "randomness" involved in the development of life, as well as a blurring of the separate issues of how life arose from non-life and how subsequent life developed from earlier living things (the jet is clearly supposed to suggest the complexity of current living organisms).
ID advocates sometimes present calculations showing the impossibly low odds of a given protein spontaneously self-assembling from a batch of amino acids. William Dembski uses this approach in his paper, "Specification: The Pattern That Signifies Intelligence" (383k PDF). Such calculations are irrelevant because they ignore important features of proposed evolutionary mechanisms — the very features that get around such seeming impossibilities, in fact.
The tornado argument depends on the common fallacy of equating "natural" explanations of life with "randomness". Only a small part of evolutionary theory is actually based on randomness. Genetic mutations and natural genetic variation present in populations are, to a large extent, random; and the kinds of selective pressures encountered by individuals (predation, food supply fluctuations, etc.) are to some extent random in nature. However, the differential benefit of one characteristic over another in dealing with these environmental pressures (that is, the "fitness" part of "survival of the fittest") is not random. Some adaptations are clearly beneficial to the organism and some are clearly not. This means that Darwin's proposed driving force behind evolution, natural selection, is anything but random.
In addition, evolution doesn't work quickly by way of massive, uncontrolled forces, as tornadoes do. Evolution theory suggests that small changes, accumulated over extremely long periods of time, result in the current diversity of life.
Most importantly, the tornado analogy lacks the two main elements that make evolution work: reproduction (which enables "descent with modification") and selection (which enables increasing complexity). The lack of these aspects reinforces the improbability of anything useful coming out of the process.
The fact that the argument posits the creation of a working airplane reveals another misconception: that evolution has as its goal the creation of complex living organisms. Evolution has no final goal or purpose; it is merely a consequence of variation among individuals coupled with environmental pressures.
Finally, the kind of calculations made by Dembski are based on (or perhaps intentionally rely on) a fundamental misunderstanding of what probabilities should actually be considered. The odds of a particular group of amino acids assembling into a particular protein may indeed be small, but the kinds of amino acids and proteins that current life is based on are not the only ones possible. Indeed, even the mixture of atoms that life on Earth is primarily based on is not the only possibility (see Wikipedia:Alternative biochemistry). And at the other extreme, the current range of living things we see around us are not the only possible life forms that could have evolved.
To illustrate the previous point with another analogy, consider the probability that Dembski's own parents would create a child exactly like Dembski. The odds are astronomical. But, of course, they did. On the other hand, consider the probability that Dembski's parents could create any child. Those are much better odds.
Ultimate Boeing 747 Gambit
Richard Dawkins: The Ultimate Boeing 747