Isn't it just semantics? Why wouldn't a preventive action be considered "improvement"? And vice versa - almost any "pure" improvement can be presented as some sort of preventive action. Can't it?
The theory is, improvement is related to a change of status quo - everything is working correctly, but you want to achieve another level of performance. Say, the output of a production device is performing as expected, but you want to have it output more. Then you improve your process to get a better performance. But the important thing is - there´s no problem here.
In the case of preventive action, things seems to be working correctly but it´s detected that, given some circumstances, things can go wrong. Then you act to prevent the problem from happening.
So, theoretically OFIs should be really suggestions for improvements. It really should not have anything to to with preventive or corrective actions.
however, much of I´m used to see is the NC threat where the auditor say something like - look, this is somewhat good, but if you do not get it better, it will probably be an NC on the next visit.
This has nothing to do with improvement.
This is what pisses me off, because in my opinion it´s one of the aspects that help people think of a quality system as just paperwork

