That’s how they got Jimmy Dean Chocolate Chip Pancakes & Sausage on a Stick
Put aside the practical difficulties of generating text. You can easily program a computer to generate next week’s winning Powerball numbers. You have five numbers from 1-55 and one bonus number from 1-42. A computer will quickly randomly generate the 146,107,962 possible combinations. Have your algorithm weed out the losing numbers. Then buy a ticket with the winning number.
If you can spot the flaw with this plan, you’ll understand the problem with the OP.
it is not just used in art. When verifying a new microprocessor, you have to generate and simulate lots of instruction sequences. However when the designers do this they tend to create only the sequences that make sense to them, and explore possibilities that they’ve thought of. Most companies now use random instruction generators that create random but legal code sequences. These do things that the designers would never think of, and are very useful in finding all kinds of weird and subtle bugs.
Because of this, you could just as easily generate the potential words on the fly instead of looking them up. Assuming that the library consists of only strings of valid words, you would start with “See” and then walk through the word list throwing away words that don’t validly go after See, and then do the same for all two word combinations you generated, etc. Lots of space, but less than the full library.
A very similar technique is used to explore paths in integrated circuits. Instead of enumerating all the paths, and then throwing out the ones that aren’t useful, you start at a gate and grow the path one wire at a time.
But you’re aware that information theory says you haven’t stored the same amount of information in a smaller space, right? You’ve thrown away information. In fact, you’ve necessarily failed to satisfy the OP’s goal by eliminating future works which aren’t yet recognized as valid sequences of words.
George Gamow also addressed this problem. You can read it in the Google preview of his book “One, Two, Three…Infinity” starting at the last paragraph of page 11. link
Briefly, with a printed line 65 characters long and 50 different characters (including punctuation and special characters), there are 50^65 or roughly 10^110 possible lines. If each atomic particle in the known universe was a printer, and they started printing at the Big Bang at the rate of “atomic vibrations” (which he gives as 10^15 lines per second), by now they would have printed roughly 3*10^106 lines. That’s about .03 percent of all possible lines.
That’s true. I deliberately didn’t use letter sequences to allow for typos. However, if you have a filter you’d use for the search, you could also use it here for message generation - though you’d have to look at all possible n_1st words. Still better than the library.
Do you mean a room full of strategically constrained monkeys? Perhaps something like Rosetta@home? It’s a SETI-style distributed computing project that (as I understand it) sorts through possible protein structures looking for ones that might be useful.
All seriousness aside, it seems to me that the future is being created by a bunch of monkeys typing on keyboards.
An earlier version of the concept appears in Book III of Gulliver’s Travels.