It does depend a lot on the sub-field of CS. For systems work you spend a lot of time implementing ideas and running experiments. In a lot of cases it isn't n

It does depend a lot on the sub-field of CS. For systems work you spend a lot of time implementing ideas and running experiments. In a lot of cases it isn't necessarily that difficult to find an interesting idea that's probably viable. The real work is spending a lot of time writing code and work out all the annoying details to get to the point where you have a decent proof of concept.