>> CS is NOT all about programming, there are countless branches of Computer Science were programming has *nothing* to do
> Name one. I bet there's some programming involved in there.
Bet taken.
In our department (one of) the largest research groups are into type theory. Also hard computing science fields like, complexity theory, computability and (say quantum) algorithms require little or no programming.
> Sure, you need the theory as well. If you've got an algorithm that you think is always more clever than the currently accepted best - or that breaks something currently thought of as unbreakable, etc, you need to prove it mathematically. But a lot of people will think that you're probably pulling a fast one if you don't have actual data to back it up, so you probably should implement it.
I am a Ph.D student in algorithms, and my (and most of my research groups) view is the exactly opposite. Proofs are the good stuff, proofs are what you want, and by proving facts is how a/the science advances. Having benchmarks demonstrate an inability to prove/formulate claims, or indicate that you are actually studying something else than computing science. Also benchmarks have the old problem of being easy to tweak to indicate a more favourable result than might be expected.//T
Slashdot Mirror
>> CS is NOT all about programming, there are countless branches of Computer Science were programming has *nothing* to do
//T
> Name one. I bet there's some programming involved in there.
Bet taken.
In our department (one of) the largest research groups are into type theory. Also hard computing science fields like, complexity theory, computability and (say quantum) algorithms require little or no programming.
> Sure, you need the theory as well. If you've got an algorithm that you think is always more clever than the currently accepted best - or that breaks something currently thought of as unbreakable, etc, you need to prove it mathematically. But a lot of people will think that you're probably pulling a fast one if you don't have actual data to back it up, so you probably should implement it.
I am a Ph.D student in algorithms, and my (and most of my research groups) view is the exactly opposite. Proofs are the good stuff, proofs are what you want, and by proving facts is how a/the science advances. Having benchmarks demonstrate an inability to prove/formulate claims, or indicate that you are actually studying something else than computing science. Also benchmarks have the old problem of being easy to tweak to indicate a more favourable result than might be expected.