That actually depends on what you want to protect. If you want to prevent someone from wholesale copying of your code then yes it makes little sense since other laws would come into effect to (hopefully) protect you. However, what if your code contains some unique routines/code and/or solutions to problems not generally in the public domain. In this latter case one would want to spend some time and effort to help prevent decompilation.

Anything can be cracked. I have had the oppertunity to "code" propietary algorythms (propietary means someone paid economists and mathmaticians to "plain talk" extreme calculus to me). Data was being pulled from all over the place, data names were cryptic and the "pricing" iterations were long and tedious. It doesn't make sense to reverse something like that - one would be better off inventing their own "short hand" from public domain sources. Complicated solutions have to be "owned" for them to applied commercialy and profitably - i think!:-)