| Interesting puzzle: one can leave out one letter, then another, and measure at what rate does understandability decrease for people (until eventually it's complete gibberish). Also, given a large enough audience, one could try different combinations of letter-removal orders and distribute those randomly, and measure the same... the question being: * Which letters are the most vital to understanding written English? or * Which combinations of letters? Are the letters the same in all the combinations of a certain rank, or do they differ? (where certain letter combinations themselves make each other more necessary) -- etc. Alternatively: the same experiment, but where all words submitted by a user must match to a dictionary of real words. To that one may even add a grammar checker! (Which sets of sub-alphabets lend themselves best to proper English? Which allow for the shortest subsets? -- etc. etc.) This would be utterly useless and I have no idea how it could be done. The thing is, the hardest part of solving this puzzle, if one were to try, is not designing the experiment and writing the code, but figuring out how the hell to convince a massive audience large enough to collect the significant data to subject themselves to this experiment in the first place (or trying to design an AI that could be subjected to it -- and evolve the AIs themselves until they excel at deducing unintelligible text -- Would the results be the same for people and AIs?)... I like this puzzle because it's so utterly stupid yet it would probably take an incredibly gifted person who has off-the-charts OCD to actually solve. If I were a multi-millionaire I would hire a crack-team of engineers and mathematicians to work non stop on this particular problem; just for the hell of it. |