As someone who spends a fair amount of time with software developed by researchers (mostly biochemistry visualization software), that for all the "research into what has been done", I think there is a huge amount of effort being duplicated all over the place. In my experience this is due to the relatively short timelines of grants, the fact that you have to spend most of your grant money demonstrating that what you created has some kind of scientific impact vs actually implementing the thing, and, as others have pointed out, the fact that a lot of the core work is done by diamond-in-the-rough superstar grad students who will move on after 2-4 years.