So the playtest is out, I've had time to digest the material, and I even played a short session with some non-gamer friends. Below are my commentary on various elements (note that this is based on the May 24th Playtest Packet).
The Core and Skills
The core system is really solid. Its strengths are simplicity, mechanical consistency, and (mostly) transparent math. I'm also a big fan of the flexibility. Essentially, all d20 rolls are either attribute checks or variants thereof. Even saving throws fall under their purview; there is a saving throw for each attribute, depending on the specific effect (for example, if something is best resisted or avoided by the use of intellect, you'd make an Int save). Since everything boils down to attribute checks, improvising actions is more intuitive. Because of clunkier math, in previous editions it often wasn't mechanically advantageous to "resort to" attribute checks, but in Next it actually works. Skills are an add-on to this system rather than a separate subsystem unto themselves. The open-endedness of skills also caters to improvisation really well, and rewards creativity with mechanical advantages. One of the big keys here is NOT tying a skill with a specific attribute. Finally, one of the major design goals is to support virtually any style of play via modularity. You could run a 1st Ed style game, a 4E style game, etc. but (presumably) without the quirks and bugs of those editions.
It's not inherently better or worse, but it is a specific tool in a DM's toolbox. One of 4E's big weaknesses is that it's extremely difficult to run combats gridless. While tactical depth is a huge strength of the system, from the scale of the adventure it's inflexible in that the system pushes DMs toward big set piece battles almost exclusively. If Next is able to replicate 4E's tactical depth (or near enough) for big set piece battles but allow for simple gridless skirmishes using the same rules then it'll be a huge win.
The idea here is an extension of situational modifiers like combat advantage. Instead of adding a +2 or a +4 to your roll, you get to roll 2 D20s and take the best result (if you have advantage) or the worst result (if you have disadvantage). While it's admittedly more difficult to simulate degrees of advantage using this system it does offer the tactile pleasure of rolling more dice, requires less math/bookkeeping, and is easier to apply after-the-fact. Besides, it's not like modifiers are completely gone (for example, half cover still provides +2 to AC and Dex saves, whereas three quarters cover provides +5 to AC and Dex saves).
One of the big weaknesses of this playtest in my opinion is that Opportunity Attacks (OAs) are not in the core rules. Combined with the fact that you can split up your movement (move 10 feet, attack, move 20 feet), and it's really easy to ignore the beefy melee guys and bash on the squishies. Which realistically doesn't make sense. A guy with a sword (who knows how to use it effectively) isn't going to let you traipse on by to beat up his friend. A common complaint regarding OAs is that they slow down combat. Quick combat is nice and all, but there needs to be something in place to give enemies a reason for fighting the Fighter. Personally I don't think that OAs suck up too much time (indeed, if a lot of extra attacks are rolled they could end a combat quicker), but there are also a few alternative systems. One is that enemies who walk past a melee guy attack with disadvantage. Another is that to get past you need to roll a Dex contest (opposed Dex checks). Finally, something that Mike Mearls (lead designer) proposed in a live chat is that if you want to disengage you can give up your action (attack) to do so, and if you opt not to do that only then will your opponent get a free swing (in other words, it's standard OA rules but they can be avoided for the price of an action).
It's Vancian, but so far not so bad. The spells are more balanced than in previous incarnations, with many of the deadliest and/or most debilitating effects having HP thresholds (i.e. sleep cuts movement in half, but only creatures with 10 HP or less need to make a Wis save or fall unconscious). Casters also have at-will spells so they aren't stuck shooting crossbows or throwing daggers when they run out of spells. The number of spells is also reigned in (as is spell scaling). This will not only maintain better balance, but reduce bookkeeping. Consider me cautiously optimistic, but fearful of eventual power creep.
Healing and Resting
You can heal yourself non-magically, rolling your HD to regain health each day. I don't think it provides enough self-healing (at least not compared with healing surges), so dependency on magical healing and/or magic items is likely to be high. I'll need to see how it plays out over the long term before I pass judgement on it, but so far I'm not liking it.
The designers admit that they've spent the least time working on monsters, and it definitely shows. It's a far cry from 4E's concise, functional stat blocks. Monsters lack variety (there's not much difference between a kobold and goblin except AC, HP, and attack values), and those that do get some neat toys (casters) use the PC spell list, which is STUPID. I want everything I need to run a monster contained in its stat block; I do NOT want to have to cross reference several spells on different pages of a player's book. Finally, in the sample adventure the condensed monster stats omit the monster's attribute scores/modifiers. Given how important attributes are in the core rules this decision completely lacks any semblance of logic.
Caves of Chaos
This updated version of a classic adventure module is what we're given in the first playtest. I won't go into details, but I really like the sandbox nature of the adventure. Even the layout of the map resists railroading. Also making a comeback is the notion that not everything you encounter is wise to fight. Specific instructions on having monsters call for reinforcements is provided, so players who make foolish decisions could end up way over their heads really quickly (whereas clever players will benefit from making wise decisions).