Community

LEVEL (as in, range of levels in the game)
1-30.

The ten level split was the best way I’ve seen yet to differentiate the scale of narrative at each tier.  I’d keep the same split as in 4e, but Epic characters need to be simpler to run.  We never played this tier because it was too brain breaking to track all your own powers, item powers, theme powers, etc etc.  Let players pick a few core abilities, then improve those as they go.  Don't make us repick all our powers every 10 levels.  Let us alter or improve them as we go though.

POWERS (4e power mechanic)
GOOD
I would say two things about this though. 

Firstly, I have never had more fun playing D&D than once powers were introduced.  I would say they should stay.  There has to be a balance though between too many and not enough.  Essentials strongly didn’t work for me because I felt pigeon-holed into one vanilla version or the other.  Yet, power selection options were so numerous, that for all the games of 4e we’ve played, we’ve never played Epic level.  It was too hard to keep track of all the options. 

Secondly, how the powers are structured is important.  It was too annoying to track round-to-round buffs, so all powers that offered short duration effects were not selected in favor of the ones that either lasted until the end of the encounter, or else had (save ends) qualifiers.  Picayune +1 for 1 round powers sat in the same selection group as more useful, longer lasting ones.  So I think you were on to something, but it needs cleaner, smarter execution.

VANCIAN SYSTEM (3.x caster mechanic) (also 0e-1e-2e)
BAD
Very few games do this any more, and with good reason.  Look at what this is intended to simulate and do it better.  In this system, a caster learns to harness his power more efficiently and more powerfully with more experience.  Maybe offer a way to expand on core powers/spells, not unlike how psions have always worked.  I’ve always thought that simulated magical ability progression very well.  Point tracking is a chore, but I liked the 4e version where you could augment powers with a manageable pool. 

CLASS ROLES (defender, controller, striker, leader)
GOOD
This worked out okay, and let people spec for maximum impact in the game. It was a little too definitive sometimes, but what I found was that a different class could simulate the way I wanted to play better than the one I started out thinking I wanted to use.  I don’t think I was ever a better player in-game either.  I found I preferred a warlord that grants other attacks and buffs with all his powers.  It was quite popular with the group too.

MAGIC ITEMS HAS TO BE MECHANICALLY (you want magic items to be almost necessary? a big part of the game?)
COHERENT
Magic items are a must in D&D I think.  I was highly disappointed in their treatment in 4e though.  The same thing happened with items as with power selection.  Any that didn’t have either useful static bonuses, encounter powers or end of encounter powers were never considered.  Here again, balance was important, and droves of items had abilities that we found to be completely useless, placed in the same “level” as items that rocked all night.  There was almost no sense of what made an item level 4 vs level 5, and forget about designing/making your own.  For my money, the best item is the one you can make for yourself.  3e had the penultimate way of doing that, detail wise, but not structurally.  PLEASE put more work into this aspect of the game.  It’s great to run around with items, but no one will ever buy boots of water walking when a 40g potion is way more cost effective to cross rivers with.  I love special materials and ways to augment existing items, as well as ‘heirloom’ type items that would unlock new powers as you level up.

ABILITIES SCORES (str, con, dex, int, wis, cha)
ESSENTIAL
This is so iconic I think it wouldn’t be D&D anymore without these.  When I'm designing/redesigning other systems, I use this as a basis for comparison.  This has proven itself out, in my book.

A LOT OF CLASSES (a dozen or more maybe)
GOOD, KINDA
A ton of classes isn’t as important as being able to build the character you want.  You need to be able to customize your class to the Nth degree.  You don’t need 50 classes if you have ten that can be tweaked the way you want them.  That’s why I liked the Builds variety of 4e.

A LOT OF RACES (a dozen or more maybe)
GOOD, KINDA
Again, this is less important than having the ability to customize the path your race follows.  We have never had a way to advance your race and your class at the same time.  You always had to choose one or the other.  You should improve as an elf and as a fighter as you progress, not just the one. 

ABILITY SCORES ADJUSTMENT FOR RACES (you want a particular race to be better with some classes and bad with others?)
YES
Again, I think it’s the options you allow for people.  Maybe I want the elf that sees in the dark to be advancing that aspect of his abilities to compliment my roguey ways.  Maybe I want to focus on INT to enhance my caster aspects.  Let us pick from a slew of possibilities or specialize in one aspect or another, and let those advance as we level.

FEATS (AND OTHER OPTIONS) THAT GIVES YOU STATIC +X BONUSES
GOOD
+X is okay, but +X and a new ability unlock is better.  So, if I get a +1 to hit with a 2H sword, it's nicer if I also get "on crit target must also save vs prone".

HIT POINTS
GOOD
Can’t take HP out of D&D, that’s also too iconic.  I prefer the other mechanic of Vitality/Wound points, but D&D without hit points, that’s just crazytalk.

HEALING SURGES (part of 4e heal system)
GOOD
Best edition yet for getting through a fight.  The time saved not slamming down an ocean of potions is very appealing.  Gold went to better things. 

DEFENSES
AS AC
I prefer the iconic AC here as well.  DR is a better idea, a la the way you did d20 Star Wars, and I’d love to see an easy formula to convert to this to try it out, but I think you’d be silly to not have it as the default.

SKILL CHALLENGES (part of 4e system)
Excellent concepts, needs better implementation.  We ended up using a third party variant because itw a smore functional.  First time we had an edition where your skills well and truly mattered. Our group became quite creative with applications of skills.
 
SKILLS HAVE TO BE
IMPORTANT
Skills should compliment your abilities like areas of expertise.  The simplistic listing helped me streamline other RPGs I play.  The skills describes things so broadly, they can be interpreted in a creative way that doesn't intrude with excessive can/can'ts.

NORMAL CHANCE TO HIT (against an average monster of your level)
50%
50% is realistic and functional.  Can't see why would you do otherwise, really. 
 
ALIGNMENT IS
FLUFF, KINDA

I’ve never found this useful outside the Good/Evil keywords that enable a lot of thematic abilities like Smite Evil.  I much prefer a system of descriptors like Palladium used to have.  That better informs me about what my character is or isn’t likely to do.  I would leave it out in favor of a quality or trait idea, such as Pious or Principled or something that comes with the class or is an option to make into a key factor of the character.  It matters to divine characters so that they continue to match the alignment to the will of the gods.  No one else need worry about it.  My formerly evil rogue can turn over a new leaf, switch to good alignment, and there are no rewards or consequences.  A paladin that drops his alignment loses all powers though.  Unless the player makes this a part of the story though, this is most often the result of a DMs decision that is based on a disagreement with the player about what the will of the deity is—a disagreement that the DM will always win.  Alignment should matter to everyone or no one.  Define for the player what the deity commands that they *must do versus *never do.  Give the player a reward for making this part of his/her choices.  Sort of like earning or losing ‘faction’ with a deity, and each level of it opening up new choices.  If my warrior takes every opportunity to venerate Kord, but has no divine class levels, then I’m waiting around, hoping that the DM makes it worth my time.  If there was a structure in place, one that provided even more benefits for divine characters, there is more incentive for players to consider aligning choices with deific concerns.

SAVE OR DIE (finger of death)
BAD

Which is to say this is only bad where resurrection is not facilitated well.  When our characters die in 4e we often re-rolled rather than resurrecting.  The gold cost, ritual time and etc, etc. meant no one ever did this in most of our parties.  It was easier in our stories to let the DM scribble up a way to add in a new person, rather than haul a dead PC’s carcass with us until we had the loot to rez them, and not let the player play.  We also had no way to “pre-res” someone at times when we did have resources.  Maybe we missed out on something there, but we never had any way to play that way.  Resurrection abilities are an under-considered, under-developed part of the game, I think.  Might have just been our group, but ainstant-death scenarios always feel like the DM being lame, cheating almost.  Most games these days are premised on having a fighting chance to succeed.  We had the most fun when we had to suddenly throw all our efforts behind stopping a party member from being petrified, swept away by a river, or from falling off a massive cliff.  Save vs Death would have never allowed those extreme moments of drama.  Those are some of the moments our group remember the most vividly.

If your game allows relatively easy rezzing, then ok, auto kill PCs off one bad die roll.  I think that’s a mistake as a mechanic, myself.  As a way to have a real sense of risk in an encounter, I would say ok, but only as a point of story, never as a standard monster feature.  You could leave it as an option, but don’t standardize this, please. 

Thanks for asking.