Depends on the context, trying something just because it's new can be useful. You may diverge from the known methods/apparatus and eventually establish something better (in some dimension) than what's common. You may find you gradually converge on the known methods from a different angle; this can lead to understanding - like now I know why they do it like that!
Context is whether it's appropriate to do such experimenting and exploring.
Most new patents are for incremental improvements. Suggesting most people start from what's known to a high degree.
If you've got the resources though, why not try something ...