We’ve had a fair bit of discussion about the facing issue in the upcoming RMu, so if it helps, here’s my take:
There are two basic types of initiative systems: simultaneous (RM2 core) and sequential (RMu/D&D).
In simultaneous systems like RM2 core, some or all actions are essentially simultaneous. This is true of the spell and missile and more or less the movement phase as well: they are resolved simultaneously. This means that for example two Magicians casting Firebolt at each other can each kill each other in the same phase, because spells are resolved together, and all take effect at the same time. What this means for movement is essentially that any movement conflicts are adjudicated by the GM, and the GM can rule that Character A should be able to turn to face Character B before Character B can get to the rear of Character A.
The downsides of this system are significant, however. First, this simultaneous system requires an extra step to combat – the declaration phase – where characters state their actions. That slows down combat, and can cause metagaming problems by revealing what characters are intending to do to the players. It also requires players to try to predict what other characters are going to do or where they are going to be in the future, which is sometimes very hard to do, and leads to players trying to cancel actions. Secondly, the core rules never gave much guidance as to how easy or hard it should be for one character to get behind another, essentially leaving it up to the GM to work it out. Finally, the core rules did not treat melee combat the same way: instead of being simultaneous, melee combat was sequential, with each character resolving a full attack and implementing the results before the next got to act. This created still further temporal problems, especially because elsewhere the rules defined melee combat as a flurry of blows. The initiative rules though were treating melee combat as a set of single blows, entirely made and resolved by one character before the next could act. Melee was sequential, but everything else was simultaneous.
So, D&D, Pathfinder, the RM2 Companions (some of the new initiative systems), and now (imho) RMu offer a different type of system: a sequential system. In this system, each character takes an action – spell, missile, movement, or melee – and resolves it before the next character can act. This has the great advantage of treating all actions the same (rather than having different rules for melee vs. everything else), and it also speeds up play by eliminating the need for a declarations phase.
As you can predict, though, what you now need to add to a sequential system if you want it to work best is some additional opportunity for characters to react to other characters’ actions. Because now there is no GM to adjudicate and say, ‘No, you can’t just run 50’ around the Orc and to his back without him turning to face you; he can see you coming.’ D&D doesn’t need this in its core rules because it doesn’t have facing rules in its core; but it does add them in the DMG (p. 252) as optional rules, and these additional rules do indeed specify that creatures can change facing any time they move, or as a reaction when any other creature moves. Similarly, RMu has added a free change of facing so long as the character is not moving at a run (3x) or faster.
I much prefer the DnD/Pathfinder/RMu system, as it speeds up combat and clarifies the rules by treating all actions the same. My only beef with the new RMu system is that I would like to see clearer rules for movement conflicts than ‘Roll opposed skill checks’. I would also like to see zones of control actually be controlling in RMu, because these too prevent facing abuse: movement into and through an opponent’s zone of control right now in the core rules is not punished at all by the sort of ‘opportunity attacks’ that DnD and Pathfinder use, which means RMu’s zones of control are not actually controlling at all. I would also like to see more robust rules for opportunity attacks, which are hamstrung in RMu because they require the attacker to have already spent AP on an attack, as well as for the attacker to have declared an attack against the mover specifically (and spent at least 1 ap on it). Neither of these is the case in DnD or Pathfinder.
I posted my suggested rules for facing and opportunity attacks here, if they might be of use to you:
http://www.ironcrown.com/ICEforums/index.php?topic=19183.0
TLDR: I would first decide if you are using a simultaneous or a sequential system. It sounds like you want to move away from the RM2 core, which is a mostly simultaneous system, towards a sequential system. If so, I would indeed offer a free change of facing. I would also consider beefing up opportunity attacks – i.e. offering them free – when opponents move carelessly through another character’s zone of control, because zones of control and opportunity attacks are what make players have to think twice about abusing facing rules.