The new D&D Next playtest packet came out today, and it's as dense with changes as any packet we've yet seen. Only races have gone mostly unchanged, I think. I am surprised by a lot of these changes, and the early summary is that I hope most of these changes are experiments rather than expected final forms.
The first change one comes to is the change to skill training. All characters now have skill dice, the size of which is dictated by level, and get to roll that die when using a trained skill. This starts at 1d4 at first level, and scales up to 1d12 at 20th. This is better for "no guarantee of success," and worse for differentiation. I'm happy to see that they've rescaled DCs so that first-level characters can't accomplish the "nearly impossible." I don't think differentiation is that big of a deal, but it does mean that players no longer make choices as they gain levels to get better at one skill rather than another. Skill tricks are a really nice addition, though: when using a skill, the player may either add the skill die to the d20 roll or activate one of her skill tricks. Regrettably, these mechanics are restricted to rogues.
The one (relatively major) implicit problem with skill tricks is that anything a rogue has to pay a skill trick slot and expend a skill die to use, another character cannot be allowed to do on a skill check. Introducing skill tricks thus provides an upper bound to what a character can do with a skill check. I have wanted open-ended, loosely-defined skills in the game from the start, and skill tricks are quite the opposite.
The XP table has changed, and now goes to level 20. The numbers are slightly larger, and based on what I've seen in running Aurikesh, it would probably take three sessions to go from first level to second rather than two. (This assumes a sometimes breakneck pace for a party of 5-7.) I find it strange that PCs stop earning feats after 9th level, though ability scores continue to increase every four levels. Based on today's L&L column by Mearls, I suspect that they're leaving space open for benefits from prestige classes.
All of the classes are now mapped out to 20th level, and most classes (everyone but wizards) now gain martial damage dice and a martial damage bonus, though fighters, monks, and rogues get the most of both: 6d6 martial damage dice and +20 martial damage bonus. The fighter class gives the strong impression that they wanted a few more variations of ways to give the fighter more damage output, since the fighter stops getting new Maneuvers at 10th level and starts getting Combat Surges, which grant an additional attack and increase their damage from martial damage dice still further. Most of the interesting Maneuvers from the previous packet are gone. On the whole I would say that the fighter stays effective, but has lost any flavor it possessed. It does generate some obnoxiously large damage totals, though. 1d8+6d6+20+5 (Strength bonus) + whatever magic the weapon may have (let's say another 2d6 for a flame tongue) yields a beastly 57.5 average damage per round, assuming the fighter didn't hold anything but the skill die back for a Parry.
Martial damage dice are part of a broader ability called Combat Expertise, which is confusingly worded but ultimately explains that the PC's bonus to weapon attacks applies only when the character is proficient in the weapon; further, the character can only apply martial damage dice and martial damage bonus with weapons she is proficient in wielding. On top of losing every kind of benefit, using a weapon with which you are not proficient results in disadvantage that cannot be negated with advantage. I am surprised to discover that all characters see some reduction in weapon attack bonus from this - it looks like they're finally going back to working on math in the Bestiary.
The cleric now has several more options, selected through deity. Channel Divinity has diversified into a slew of options, so the people who couldn't imagine trickster gods granting the ability to turn the undead. Many deities grant two different Channel Divinity options. I don't think these will be all that impressive at high levels, but they're reasonably solid at low-to-mid levels (I'll define that here as levels 1-12 or so.) Many of them are damage bonuses for clerics that should behave like fighters, but in a more cleric-y way. Choice of deity is also choice of armor proficiencies, and as I suggested here, a lot of deity options boil down to making the character more like one of the other classes. (Ironically, none of them make a cleric more like a monk.)
The cleric gets high-level spells, of course, but for spell levels 6-9, they only ever get one slot per level, rather than the two they've gotten for every lower level (though they now get a third first-level spell). Clerics also do not gain domain spells of levels 6-9. That's fine, I guess, except that it means levels 12, 14, 16, and 19 have nary a new class feature. It does fit in interestingly with the fact that high-level save-or-die-like effects are not expended on a successful save, comparable to the Reliable keyword in 4e.
Monks no longer have to be Lawful, but are called out as "usually lawful." Apparently the D&D designers don't know that players always want to be the exception? (For players out there who play to type rather than always, always against type, I obviously don't mean you.) Anyway, moving on: monks break the mold of other classes, and get something new at almost every level. Many of these are immunities, which saddens me. At 20th level, they become immune to poor ability score rolls, and all of their choices of ability score increases are retroactively irrelevant. In addition to having a progression of maneuvers (two at 1st level, two more thereafter), monks get ki abilities that are more explicitly supernatural. Let me spell this out for you:
Path of Mercy = Water Benders
Path of the Phoenix = Fire Benders
Path of the Four Storms = Air Benders
Path of Stone's Endurance = Earth Benders
These are all reasonably stylish, but these abilities are a huge change to the style of the class as a whole. On its own merits, I think it's probably fine. I certainly had elementally-aligned, discipline-based warriors in one of my own settings. I think I'd be happier if elemental monks were not the default flavor - possibly a prestige class.
Rogues are the big improvement (and a massive overhaul), though I am still irritated that the rogue class is the only apparent way to gain proficiency in thieves' tools - the Guild Thief background certainly doesn't seem to do it. As I mentioned above, rogues get skill tricks, which are pretty much rogue-only feats that interact with skill dice. They also get one rogue talent, a combat boost of one kind or another. Assassinate and Sneak Attack are interesting in that they require the rogue to have advantage, and then give up that benefit to accuracy in exchange for quite substantial extra damage. They're both devastating, though it's a bit difficult to measure them against one another. Curiously, the rogue's scaling ability of 11th-20th level, Ace in the Hole, means they no longer care about accuracy, at least a certain number of times per day.
Rogue schemes offer a bonus feature, which might be anything from a feat to some proficiencies to some extra skill tricks. I have no idea why D&D assassins use shields in 4e and D&D Next. There are a few other inexplicable things here, like the fact that only rogues can have Great Fortitude, Iron Will, or Lightning Reflexes (which is currently called Uncanny Dodge). Anyway, I have my quibbles, and I would like to see the rogue moved up to d8s for Hit Dice, but all in all the rogue is now probably the most interesting class in the game.
Wizards have changed in only a few fine points. Their spell preparation has changed, giving them a bit more flexibility. I think the change is pretty reasonable, and if they know there's only one spell of a particular level they ever want to cast, they can prepare only that one and give themselves more options at other spell levels. I expect the total number of spells that wizards get to prepare will increase slightly in final printing. There are also some tweaks to the three arcane traditions, powering all three up to different degrees, and I like where they are at this point. The Scholarly Wizard won't appeal to everyone, but its ability to cast rituals directly from their grimoires rather than having to prepare the spell might do some interesting things in the long term. It certainly makes them superior utility casters. Wizards get more spells per day than clerics, but like clerics they get only one slot per day of spell levels 6 and above. I have fewer complaints about the current design of the wizard than of any other class. I am surprised to see that spell DC scaling has been removed from both wizards and clerics.
The How to Play document has a lot of small changes. Critical hits are slightly adjusted, though it remains unclear how variable damage such as martial damage dice should be resolved. Falling damage isn't much of a threat unless it's more than (10 ft * character level) - they'd do better to go back to d10s. There are rules for casting in armor, but they're currently no less punitive than "wizards can't cast in armor," and since they stop some clerics from casting in armor, they're actually more punitive than the previous rules. Once there are ways to increase one's proficiency in armor, they'll be fine. Two-Weapon Fighting now makes sense, more or less. The character adds all stat bonuses to the main hand, and makes a separate attack with the off-hand that can gain no bonuses whatsoever. It's a -2 to hit for very slightly more reliability - at least, that's what you get without the Two-Weapon Fighter specialty.
Specialties and Feats are now written in such a way that most players will purchase feats a la carte rather than as part of a Specialty. I dislike this very strongly; I thought the Specialty approach was in every way more interesting, as it came together to communicate story in a way that a la carte feat purchase has proven not to do. I certainly intend to require Specialty-style purchase, since I'm more than happy to write new Specialties for my players. The aforementioned Two-Weapon Fighter specialty offers the ability to dual-wield non-light weapons, which is only really good if that's what you were buying into this specialty for in the first place. (I take from this that the specialties are designed on the assumption that you'll cherry-pick your way around anything you don't especially need.) It eventually goes up to Two-Weapon Strike, which sacrifices a variety of maneuver options for advantage in attacking with one weapon. Rogues with this ability can always Sneak Attack, unless something imposes disadvantage.
A lot of the general areas of specialty that were missing in the last playtest packet are back for this one, such as the Defender. Funny enough, the Defender specialty gets something pretty close to a marking mechanic at 9th level. The fighter maneuvers that are intended to cover the same territory are useful damage mitigation, but don't encourage the enemy to focus attacks on the fighter in preference to the fighter's allies. I'm glad that there's some kind of marking mechanic, even if it's not as "insistent" as a 4e mark, and even if it's not available till 9th level. Also, super weird: 5-foot-steps, familiar to 3.x players, now cost one of the four feat slots a character ever receives.
From a broader perspective, the important change is that very few feats, and no specialties, are focused on non-combat prowess. Personally, I would be happier with the three non-combat-focused feats if they were cut down to just the one that grants more trained skills. Given that there are ways to buy maneuvers without being a fighter, I assume there will eventually be a feat to buy one or two of a very short list of skill tricks.
Spells have been hugely expanded, since there's now a need to support play above 10th level. There are a number of other changes; rituals no longer have a price tag, but include a list of suggested components. Some spells still have a material component cost, such as stoneskin, true seeing, and true resurrection. I think there has to be a more naturalistic system that could be worked out here, but I'm not up in arms about it. Many attack spells also have a scaling mechanic again, dealing more damage if cast with a higher-level spell slot. Scorching ray is still auto-hit and fixed damage, but less damage now. 0th-level spells are once again called cantrips, and now have their own scaling mechanic by caster level. Lance of faith, for example, scales up to 8d6 damage to a single target.
Which really points to how the D&D Next has stopped attack-bonus scaling, and has gone kind of hog-wild on damage scaling. I would love to see damage scaling reined in a bit for all classes, especially since monsters aren't dealing such inflated damage totals, and just adding up 8d6 (much less the damage expression I listed for a 20th level fighter, above) becomes a significant investment of time. I'm still inclined toward ending level scaling at or near 5th level in my own game, as we switch over to this set of rules.
The bestiary is, at least, hugely improved. Even as player attack bonuses have edged off slightly, monster attack bonuses have jumped, though I think monster ACs and hit point totals still need some help if a monster is going to keep PCs' attention for more than about one round. Doubling all hit point totals should about do it. I know it became fashionable in 4e to chop monster hit points by about a third, though my game never showed me the slightest need for such a move; what I saw was that monsters getting to use three abilities from their list was unusual, and most of those weren't going to hit anyway. Obviously I can do the "heavy lifting" and juice up monster stats on my own if that isn't the baseline.
In summary, I think they made some steps in a good direction, and several more steps in a regrettable one. I think everything here will change again six or eight more times before it's a done deal, though Mearls's public commentary gives the impression that they're close to what they expect will be a final form on a lot of things. If nothing else, reading the playtest documents for D&D Next gives me a lot of food for thought on what I would do differently and why, if it were up to me. Just now, I'm pondering how one might create a classless, point-based variant of D&D Next, using Total XP as the trigger for gaining more hit points and possibly satisfying some prerequisites.