I don't think a completely accurate mirroring of the gameplay would be the way to go, rather it is the setting and action that I'm going for. Here's what I've thought so far (as in, while in the bus on my way to school):

There are some element I seem crucial to the successful emulation of AC with DitV:

The strong implementation of the Creed itself in-game
The possibility to gain the aid of towns people (as this'll most likely be a one on one game)
The know that this game is about missions
The fact that the player can't possibly fail (and this one is the tricky one)

Now, aside from these above elements (and there might be more I haven't thought of... or maybe less o_O) I think you can pretty much extend in any possible way. To me it seems (with what's been shown in the E3 demo) that these elements were given enough importance to maintain them. One key element they seem to be going for, and that wasn't part of the first game, is "lots of variety". I personally think this variety shouldn't be hard to get in a table top RPG form.

And yes, I agree with you. You need a world where talking is as important as fighting. But I have a different perspective here. In the game, there were lots of talking, but it was not part of the gameplay itself, it was linear and was only cinematic (for obvious reasons). But in an RPG this can be overcome. You no longer have to have a scripted chat between the protagonist and the NPCs, you can actually have interaction with them. Bear in mind Altair had issues with the creed, he thought himself above it and went in a constant discussion with the the master assassin about it. this can now be played trough the conflict resolution system.

Now, how do we maintain the elements we need to emulate Assassin's Creed?

The strong implementation of the creed, even if not entirely followed by the main character, mind you, gives the setting enough structure. the brotherhood itself gives the same support to the plot as the faith in the original setting.

The possibility to gain the aid of townspeople. And I do think this one's important, this aid, even in the electronic game, can often be he difference between success and failure. I do think, though, that one can get way more creative with it. Not just have the brothers of some 10 women in distress help you by blocking the way of your enemies or have scholars help you infiltrate highly guarded locations. Make the assistance more interesting by having interesting NPCs give it to the player. Make people who're being victims of the injustice generated by the target's actions give the player assistance to eliminate that who's interfering with their well being.

Know that the game is about missions. Just as in dogs, you go to a town to see what are it's sins and cleanse it from faulty individuals. Here it is no different, except, this one time, you already know who you need to get rid of from the beginning.

The player can't possibly fail. And this is the one that worries me. This game should be centered on the character, it missions and the discoveries he makes while advancing the plot. If the character were to die, the whole game (in my opinion) would crumble around it. This gives us a little of a problem. How do we tread this the appropriate way? In the video game, when you loose, you simply go back to the last checkpoint and re.try the mission. Do we do this same thing in an RPG? how many times is it too much times to try? when do we get sick of trying? conversely, what about being able to fail only investigation and townspeople aiding missions but, as you stated, not being able to fail the main assassination? This represents a problem of it's own. for then the challenge of the actual assassination is completely lost. How much does this affect the enjoyment of the game itself?