Well, we'd have to know the exact definition of "useful" that the OP is using. If the OP means that, "useful" means that backtesting will improve the system, then we can't argue against it, because there are systems that can't be improved by backtesting. (Imagine a system that issues random buy/sell signals. Now imagine a system that does a lot of calculations, but in the end, it's signals are essentially random. Neither of these systems can be improved via backtesting.) If the OP will accept that "useful" can also mean that backtesting can tell you when a system may not be profitable, then I agree with you.
I don’t think that backtesting is useless. It helps us to check the accuracy of our strategies. It helps us to optimize and improve our strategies for future references. We can easily find flaws in it and try to change our strategies if it’s not working well for us. I personally do a lot of back testing and find it really useful.