Knowledge 133 The Problem One suggestion put forward by Daniel Dennett is the creative process is a two-part task — generate ideas, then critically assess them. I can, in principle, make a program write out every possible book less than 500,000 words long. Provided I don't store the results this will not collapse the Universe. This just leaves the problem of writing another program to read all the output and ring a bell each time it finds some interesting truth. This second program might be called an appreciation program. Let’s examine this approach. I can write out a very simple program to do this — provided I cheat and ignore the complexity of the term ‘something interesting. In plain English: Count up from one until I get an interesting fact, write it down and stop. Loop i++ until i == (Something Interesting), Print i This generates two problems. We need to make a program that can tell if something is interesting and it will need to be fast because it is going to be handed a huge amount of junk. Clearly I have a process running in my brain that can determine if something is interesting, but it is quite slow. It takes me an appreciable time to open a book, leaf through the pages and declare it either junk or interesting. Leo Tolstoy had a process in his brain that allowed him to create something interesting but I want to prove he did not do this by generating random junk and sifting through it. Let’s look at the mathematics. We know simply counting sequentially through every number would take too much time, but why not generate random numbers and run our critical eye over them? Surely this would give a faster result. Let us try with a short poem. How hard would it be to come across something as simple as a four-line poem using this technique? This poem, by the late Spike Milligan, is only 23 words long, including the title, and I have a powerful computer. Wouldn't it be possible to generate it using a computer? Unfortunately, no. We humans don't have a good he