But. 'the best choice' is not a thing! I mean, not in isolation!
*nods* I was taking a little time to think up my own response to the above, but I think you've hit what I was going to end up at.
I have some sympathy for the idea of defining creativity so that it excludes pure optimisation exercises (given some options and a precisely defined utility function, pick the highest-utility one of the options). If nothing else, you have to define it so that it excludes something or else there's no point having the word in the language at all :-) But in many real-world cases, the utility function is not well defined; for programming in particular it's extremely common that five years down the line something turns out to have been part of what you would have liked to be optimising for in the first place (reusability in some particular way, maintainability in the face of an unexpected constraint), and also a lot of utility functions have messy human concepts like 'ease of use' or aesthetics somewhere in them.
And that flexibility of the utility function is what distinguishes a purely computational optimisation exercise from something that has a predictive, intuitive, imaginative, creative dimension at the point where you apply interpretation to the unclearly specified requirements and decide what actually measurable proxy for those requirements you'll choose to optimise for.
(no subject)
Date: 2013-12-25 11:34 am (UTC)*nods* I was taking a little time to think up my own response to the above, but I think you've hit what I was going to end up at.
I have some sympathy for the idea of defining creativity so that it excludes pure optimisation exercises (given some options and a precisely defined utility function, pick the highest-utility one of the options). If nothing else, you have to define it so that it excludes something or else there's no point having the word in the language at all :-) But in many real-world cases, the utility function is not well defined; for programming in particular it's extremely common that five years down the line something turns out to have been part of what you would have liked to be optimising for in the first place (reusability in some particular way, maintainability in the face of an unexpected constraint), and also a lot of utility functions have messy human concepts like 'ease of use' or aesthetics somewhere in them.
And that flexibility of the utility function is what distinguishes a purely computational optimisation exercise from something that has a predictive, intuitive, imaginative, creative dimension at the point where you apply interpretation to the unclearly specified requirements and decide what actually measurable proxy for those requirements you'll choose to optimise for.