Surely it’s the issue in software development. Everything else is a case or subproblem of “how do I use a computer to solve this imperfectly defined problem?”
I mean, it’s borderline tautological, isn’t it? Software development is the practice of translating natural-language requests into executables. That’s not just a recurring question – it’s definitional.
This is why I’m nervous when strongly antisocial people say they want to go into software. The point of this field is to figure out what people need (or want, or mean, or like). Of course you can do this creatively rather than responsively, and with more rather than less respect for your customers. But you don’t escape human tastes and concerns when your job is to explain them to a microprocessor.
Well, there's the second issue of doing it such that the machine can produce results efficiently. Which tends rather towards the mathy computer-science side of things - which is where I think the less social "I just want to write code" people would be much happier.
Efficiency is implicit in most natual-language queries. Managing efficiency tradeoffs is a good example of something fuzzy yet important that you have to be able to figure out from natural-language queries and human interaction.
But yeah, if you really just want to work with numbers, you’ll probably be happier in CS academia.
I mean, it’s borderline tautological, isn’t it? Software development is the practice of translating natural-language requests into executables. That’s not just a recurring question – it’s definitional.
This is why I’m nervous when strongly antisocial people say they want to go into software. The point of this field is to figure out what people need (or want, or mean, or like). Of course you can do this creatively rather than responsively, and with more rather than less respect for your customers. But you don’t escape human tastes and concerns when your job is to explain them to a microprocessor.