These threads keep popping up again and again, yet the "argument" goes on because the participants are talking about two entirely different rationales for paper-trading.
No one suggests -- or at least I hope they don't -- that paper-trading simulates live trading. How can it? Anyone who thinks that PT in and of itself is going to prepare them for live trading is going to be sorely disappointed.
This is not to say that PTing a new platform has no value. One should obviously get used to whatever it is he's using before trading with it "for real". To do otherwise is a waste of money. However, this sort of PT should take no more than an hour, if that.
All of this, however, has nothing to do with using PT to test a strategy, one which has been backtested and which is ready for forward-testing (preferably with replay). If the strategy holds up in backtesting and forward-testing, then one is ready to PT it "live" in real time. If it doesn't, then it's back to the strategy in order to determine the fault.
As for the PT in real time, the point of it is not to simulate live trading but to determine if the strategy continues to maintain its integrity. If it doesn't, then it sure isn't going to improve any if one trades it for real. If it does, then one can go ahead and begin trading it for real using small lots. If it then fails, one can look to oneself for the fault in addition to the strategy (though if the testing has been done properly, the fault is most likely with oneself).
In a nutshell, if one employs PT properly, PT will improve his results. If he doesn't, it won't.
See The Edge and Trading Journals, below.