Mourn and Imaro's points are both valid.
It is perfectly fair to compare 4Ed to 3Ed, 3.5's initial Core, and 3.5's final version, depending upon what you're comparing.
If you're comparing initial release to initial release, comparing 4Ed to 3Ed (or 3.5 Core, given the changes that were made) makes sense. There, you're examining base states: what can you do right out of the starting gate?
If you're comparing overall options between 4Ed to 3.5 on the basis of lessons learned between editions (like comparing the transition from 2Ed to 3Ed Core) then its the latest incarnation of 3.5 you need to examine. Then, you're asking: Given all the refinement & developments in game design that happened in 3.X, what changes cropped up in the new release that do or don't make sense (IYHO). This matters because the latest version of a game is the version against which the newest version must compete against in the market.
IMHO, 4Ed is a lesser game on both criteria. That isn't to say its not fun or anything else- just that it doesn't have the initial flexibility or depth of the 3.X Core books, and there were certain definite late-stage improvements to 3.X that could have been translated into the Core of 4Ed.
Well, since it's all IYHO and IHMO, it is of course a subjective statement, so take mine with the same reservations:
There are a lot of core concepts that needed to be changed to achieve many aspects of the design goals for 4E, like game balance over all levels and classes, as well as against all monsters, extending the Sweet Spot, defining tiers of play, extensibility, simple character creation, easier monster use and encounter building, easier NPC and monster creation. I think a lot of the 3E core mechanics and assumptions - including some of its flexibilities did stand in the way of this.
That is unfortunate, and it might always be a limitation for 4E. But overall, I think the design goals and the results are worth the limitations. If you don't agree, well, 3E is still there. If you think there must be a "better" way, I just haven't seen it. I cannot claim to have seen the majority of d20 OGL Systems or anything, but the stuff I've seen so far did not convince me. I need to see actual "proof".
One big thing in 3E is the differences between spellcasting and weapon combat. These are two big subsystems that try to affect the same aspect: Combat. While both used some broadly similar concepts, there was a fundamental asymmetry in it that makes the act of balancing the system impossible. Bo9S basically is the attempt to have both parts use a similar subsystem, and thus resolving the asymmetry (but it aims only for the combat system). The same is done by the 4E power system.
Similar problems though existed with the skill system and the spell system. Again, skils and spells often try to solve the same problems (Knock vs Open Lock, Knowledge (History) with Legend Lore, Teleport vs Survival). And again, these systems can't be easily balanced. 4E tries a new approach, and splits combat spells and "skill" utility spells into different subsystems, and links them stronger to the subsystems for combat and skills with th combat powers and rituals.
A different example (on differences in flexibility) might be class creation:
It was very easy to create a class for 3E, because there were little guidelines for it, and you just had to create one to three core features. But it was hard to figure if the class was really balanced or how it could break the game.
4E requires more work, since it gives you stricter guidelines to achieve the balance it aims for, and you might spend a lot of work creating new class powers and features that fit the "theme" of the class. It was a lot easier in 3E - many classes have only two-to-three out-standing class features (Feats for Fighters, Barbarians with Rage and Damage Reudction, Rogue with Sneak Attack, Duelist with "Int to AC"), and most of the variation is done by scaling them. (Get extra dice, get more feats, Rage benefits improve further and so on).
But 4E demands you to create powers that cover 30 levels - not one for every level, but the levels you get powers (which is still the majority), you need more then just one - you need 3-4! That means you devote more "design & development" into each class, and that you also need more page count to cover its abilities. So, in the end, we end up with 8 instead of 12 classes.
There's only one thing I really want to comment on.
