ZombieRoboNinja said:
A sneak attack is a pinpoint blow that takes advantage of your enemy's tactical weakness to hit them where it counts. Why should this be easy to do with a 50-pound, 5-foot-long hunk of iron?
Because you can hit an awful lot of vital points at once?
Rogue 1:"Didja see how I shanked that guy in the spleen with my dagger?"
Rogue 2:"Didja see how I shanked that guy in the spleen, liver, kidneys, lung, heart, and eyeballs with my greatsword?"
I'm going to 3e grognard hell for this, but I think SA with smaller weapons only is more thematically correct and breaks verisimilitude LESS for me. As long as a rogue can wield a greatsword via a feat or 'fighter training', not a problem. I have no problem with Tuffy McStabsyou, Guild Enforcer, shanking an unsuspecting guard with his shortsword (SA damage), then, when the other guard sees him (Tuffy lacks Combat Advantage now), drawing his Sword Of Mighty Hacking (+1 Greatsword) and going at the guard mano-a-mano, using some of his rogue tricks to confound his foe.
Build flexibility comes from many places. It's not a problem to me if a 'pure rogue' can't meet every possible archetype of every character ever called a 'rogue', provided you can build an *effective* character via feat choices or multiclassing. It's like complaining that, for example, a wizard who takes 20 levels of wizard but spends feats on martial weapon prof, armor prof, and weapon focus is an utterly useless 'swordmage', when you could just take a level or two of fighter and then take a 'swordmage' PrC.
I am more concerned by the promise of 'role efficiency' and multiclassing. If you can multiclass across roles (Striker/Defender), it seems you'll HAVE to sacrifice some level of role ability in favor of flexibility, and that goes against the stated design goal of "No one will ever be bad, or even just mediocre, at their role, EVER". OTOH, if you can be 90% as good as a 'pure' Striker or Defender while multiclassing, it makes multiclassing way too attractive. 3e ultimately solved most multiclassing problems by eliminating frontloading or creating PrCs which were optimzed for 'class a+class b', even to the point of making a Paladin/rogue PrC.
