While I'm still waiting for Amazon to deliver my copy of Player's Handbook 2, the very first preview of PHB3 has just come out on D&D Insider (for subscribers only). The article (labeled as 'playtest', since it's an early version of the rules) shows a very interesting concept to be presented in that book: a new multiclass system to make hybrid characters. Unlike current multiclass rules, this allows a nearly even split of features and powers between both classes from the first level. When this preview was announced, before I knew how it would be implemented, I was convinced that such a system was doomed to be utterly broken. I was wrong.
Enter Hybrid Multiclassing. The current system falls short when it comes to allowing characters to be as good in one class as in the other, but hybrid classes do just that. For each class, you get half the at-wills, more or less half the other powers, and roughly half the features. This could have turned out disastrous, had the designers chosen the easy route of just handing out a selection of class features. Instead, they have come up with some very clever solutions that handle most problems, at the cost of a slightly increased complexity.
It works as follows: each class has a hybrid class entry, much like a regular class description, but with reduced statistics and features. To make a hybrid character, you pick two classes and add the benefits from each hybrid entry. A character made this way should be close in power to a single class one. In addition, you can pick your powers from either of your classes, but you must have an at-will, encounter or daily power from both classes before you take a second power of the same type for a single class.
The key here is that, rather than having regular class features, hybrid characters get hybrid features, which have reduced power or restrictions in their use. Although feature selection varies, you'll always have, at least, the signature abilities for your role: defenders have marks, strikers deal extra damage, leaders heal, and controllers cry. It seems they are still experimenting with feature restrictions, though: some have decreased frequency (like healing word, limited to 1/encounter), others see their effect reduced (Divine Challenge just deals Cha modifier damage), and the rest only work when using powers of the appropiate class (like Combat Challenge or Sneak Attack). This last group is my favourite, and I hope it sees more use.
I also like how proficiencies are managed. A hybrid character is proficient with all weapons and implements available to any of its classes, but only gets armor proficiencies possessed by both. That is, a Fighter/Wizard hybrid gets martial weapons and arcane implements, but can't wear armor heavier than cloth. This ensures that characters will have the tools for each class' attacks and prevents hybrid combinations with excessive defenses, although it may make too many builds have below average armor class.
Although the whole premise is, in my opinion, very solid, we have been shown an early iteration, and it's not without issues. There is, in particular, one point where the system could fall apart - the method for acquiring additional class features. Right now, there is a feat called Hybrid Talent that you can take (just once) to gain any single class feature from one of your classes, as long as you don't already have a hybrid version of that feature.
This is wrong at many levels, and I expect it will get changed before the final version. Features vary too much from one class to other for such a feat to work properly: among other potential problems, for some classes with few relevant features it would allow hybrid characters that had virtually all the benefits from their first class, plus the hybrid feature from the second. This would be true for Wizards (who would only miss the lackluster Spellbook) or Rangers (who would get everything but the often-unused Prime Shot). In addition, there is a lot of potential abuse with strong features like Battlerager Vigor.
Overall, we have a very promising system that could add a lot of variety to the game, and induce many players to rebuild their characters from scratch. This could definitely become a major selling point for Player's Handbook 3, if they get it right. Unfortunately, PHB3 is almost a year away, and the version presented isn't fully functional: only the eight PHB classes, plus the Swordmage, are supported, and the vital Character Builder won't be compatible for now. Still, it shouldn't be difficult to adapt the rules for new classes, and there are already fan-made suggestions of hybrid versions for the PHB2 classes.