The system may be very effective at deterring many would-be criminals. Any system will be ineffective against some subset and so this statement would still apply.
Under any system, most crime will be committed by those whom that system does not effectively deter.
So, in order to improve the system and reduce crime, you need to understand those who are currently not effectively deterred.
More of the same is not going to work if you hope to improve outcomes.
In any case, the parent's point seemed pretty empty.
> Who's to say that the current deterred subset of the population won't begin to commit crimes if our style of punishment is changed?
This is what research is for. We may never know what is optimal in an absolute sense, but we can figure out some relative costs/benefits of different approaches to problems and choose our tradeoffs appropriately.
Crimes of passion (i.e. many violent crimes, especially) don't take the system and its deterrents into account.