I have always had a hands-on approach to this problem because I find that having the pieces in my hands can show me more effectively than all the fancy software can, how minor alterations to length, angle, weight and position can produce different results/reactions. What do I mean by hands-on? It means that I need active participation as opposed to the theoretical approach of computer software. Without a hands-on approach I don't get the feedback necessary to this kind of research. It is not always possible to test every potential alteration with the kind of software currently available for free and which will work on a home PC. In my experience you have to input each variation of angle, length, weight or position and run the test but the results are not always informative and a hands-on test will suggest other possibilities not recognisable in a software run. You cannot imagine every possible variation and just input it - without the pieces in your hand and arranged and rearranged on the work bench you simply will miss opportunities that occur to you as you manipulate them.
Only those who routinely use hands-on building practice will understand my point of view and I suspect that those who favour the CAD/CAM approach wil make the counter argumenty equally effectively, nevertheless that is how I work and although it takes much longer than using computer aided design it will, in my opinion, win out in the end.