Let's call a string of characters that can be typed in an hour or less a "typable" string. In principle, all typable strings could be generated, and a team of intelligent programmers could throw out all the strings which cannot be interpreted as a conversation in which at least one party (say the second contributor) is making sense. The remaining strings (call them the sensible strings) could be stored in an hypothetical computer (say, with marks separating the contributions of the separate parties), which works as follows. The judge types in something. Then the machine locates a string that starts with the judge's remark, spitting back its next element. The judge then types something else. The machine finds a string that begins with the judge's first contribution, followed by the machine's, followed by the judge's next contribution (the string will be there since all sensible strings are there), and then the machine spits back its fourth element, and so on. (We can eliminate the simplifying assumption that the judge speaks first by recording pairs of strings; this would also allow the judge and the machine to talk at the same time.) Of course, such a machine is only logically possible, not physically possible. The number of strings is too vast to exist, and even if they could exist, they could never be accessed by any sort of a machine in anything like real time. But since we are considering a proposed definition of intelligence that is supposed to capture the concept of intelligence, conceptual possibility will do the job. If the concept of intelligence is supposed to be exhausted by the ability to pass the Turing Test, then even a universe in which the laws of physics are very different from ours should contain exactly as many unintelligent Turing test passers as married bachelors, namely zero.