In RMU, applying shield DB or parry DB against an attack isn't an action.
The beta2 version says it is. On p. 102 (104 of the .pdf) this is spelled out explicitly:
"Using a shield is treated as an instantaneous action (0 Action Points for the first foe, 1 AP each foe thereafter)."
Sorry, you're correct for beta2, but that's something that's changing. Previously noted here: http://www.ironcrown.com/ICEforums/index.php?topic=17150.msg209467#msg209467 I should have mentioned it was a change though.
Ok, that's good; sorry I forgot about that discussion we had (it was a year ago!).
The rule itself seems fine, but why not codify it to make it more clear? That is to say, you seem to have a category of 'free actions' which are actions that always only cost 0 AP; you're just not saying it explicitly, and rather than having it as a general rule, you are making specific exceptions. So why not do what DnD did, and codify this into a general rule -- i.e. a category of 'free actions' that always cost 0 AP? Then in the action cost chart in Arms Law, you just list this and any other free actions as a 'free action', and have a note clarifying that free actions cost 0 AP. This system will be instantly recognizable to DnD players. DnD lists several actions that are free actions (including drop an item and speaking), and you might want these to be free in Rolemaster too:
http://www.dandwiki.com/wiki/SRD:Free_Actions
You've already got the basics in place of this system now, it is just that you are doing it piecemeal with individual special rules; so why not just codify it?
0 AP Perception: you can't act on the information anyway, it just disrupts the flow of play, better not to allow it.
You many not be able to act on the information, but that information will determine whether or not you are surprised or flatfooted, whether the attacker can use Ambush, etc. That information could mean the difference between life and death. Note that my Magician sees the Orc; she is not surprised by it. She just can't act because it is not her initiative. In a case like that, you have to give her a reasonable chance to defend herself, because you're trying to model real-time combat in a turn-based system. You have to give her a fair chance to react on an opponent's turn, or else you will have players gaming the system by just waiting until one player declares an action, then charging in and attacking them when they can't defend themselves.
In fact, in Dungeons and Dragons, one could act on this information by saying, 'Look out! There's an orc!' This is because talking is a free action. Can you not do this in RMU?
0 AP Quickdraw: you can't declare parry unless you've spent at least 1 AP on an attack action, so it doesn't make any difference whether you are able to instantly pull your weapon as an interrupt or not.
I think that extra rule is going to be problematic, because characters won't have a fair chance to react to changing circumstances. In my example above, if the Orc had been using a sword rather than bow and my magician had her staff out, then this rule would mean she could not attempt to parry the orc's attack at all? That seems unfair to her: she was not surprised by the Orc at all: she saw him coming down the well lit hall. But the system is treating her as if she were surprised.
It would be better I think to say that declaring shield use and declaring parry is simply a 'Free Action' that costs 0 AP. They you can specify that the parryer must also declare an attack action.
0 AP Drop Item: there are some edge cases where it might be useful to drop things as an interrupt, but it's not at all clear that it's good for the game that you can do so. If do you want to have something ready to drop as an interrupt ("if you take one step forward..."), you can use a conditional action.
Again, DnD went through this problem extensively, and RMU already says that dropping an item is a 0 AP action. Why not just say dropping an item is a 'free action', like DnD does?
0 AP Deflections/Bladeturn when surprised: it certainly makes a difference and I said so http://www.ironcrown.com/ICEforums/index.php?topic=17914.msg216717#msg216717 but I also said that when someone catches you completely by surprise, it makes sense that you shouldn't have time to get the spell off. If you have at least 1 AP of action before it happens, that's sufficient to cast your spell; if you don't even have 1 AP to respond, you are essentially surprised (or at least unprepared).
But she's not surprised, and it is arguable whether or not she is unprepared. However, the game is treating her worse. The surprise rules on p. 86 note that if a character is not expecting combat, but is aware of the attacker's presence, then the only penalty she gets is a -20 to initiative. Here, though, my character is actually not only aware of the attacker's presence but expecting combat; yet instead of a -20 to initiative, she is prevented altogether from parrying or casting a spell until the orc's attack is resolved.
Do you see how harsh that is? It is an effect of not allowing characters a reasonable chance to react to real time movement in a turn-based environment. This is why DnD has categories of 'reactions' and 'free actions', so that characters are not screwed by the turn-based rules. If you don't allow reactions, then players will figure out pretty quickly how to game the system by waiting till the last minute to attack, and then attacking enemies who are made defenseless by the system's lack of reaction abilities.
And, again, it might make sense to handle these differently if you are not going phase by phase.
Perhaps I was mistaken, but doesn't RMU offer the option of going to a 4 AP action count? You seem to be saying that this will be a problem only if you use the 4 AP count; why isn't that a problem for you then? I would expect that if there was a problem with any option RMU offers, then that is a problem that needs to be addressed.