We say that climbers "conquer" mountains; I think tasks can be challenges/conflicts if they are a struggle to complete. And I think that, with the right degree of segregation and abstraction some engaging and tense conflict "mini-games" can be created for most conceivable challenges in D&D.
I think this could make a really blasting version of D&D, if done well, and I really do wish that this was where WotC were focussing their efforts.
This post described the cases where task-based (D&D/traditional RPG style) rules are useful really well. Where I disagree is that, for other cases, DM fiat is the best primary vehicle for in-play decision making - I don't think it is.
...
In summary, I recognise both styles of play @
Kamikaze Midget describes, but if I wanted to get to the one he claims is "non-rules focussed", I wouldn't start from D&D. I would still want rules - they would just be very different rules to those of "traditional" RPGs, and the role of "GM" would be very different, indeed.
...For me, the interesting division is why someone is totally OK with this in NPC and PC verbal interaction, but is not OK with this for NPC and PC combat.
Why is it that in one circumstance you say, "If you can make a convincing case that makes sense, you're good," and in the other circumstance you say, "You must roll dice and perform maths and play this mini-game?" What does each of them give you?
I mean, Amber Diceless is a fine game. So when do we decide that we want the rules?
I think the key thing about mini-games for resolution ties back into Kamikaze's point. My answer is that if you have that level of detail in the game, sometimes you want to use it and sometimes you don't. Most games answer this by having a somewhat involved system for the standard, but then DM advice to skip over a portion if it isn't important. (Though many systems aren't really very clear about this, or when it should be done.)
Burning Wheel kind of pushes this idea to the limit, thanks to its central focus on the characters' beliefs. If this thing you are trying to do is really important, you pull out the "artha" (various kinds of fate points), and most likely a relevant mini-game. If it is sort of important, you roll with it, and most likely have at least a partial failure, but gain some artha due to trying. If it isn't important at all, the DM will "Say Yes"--giving you want you wanted in the narration, but no advancement or artha gathering opportunity.
All that said, there is a
huge difference between what people at a given table often want versus what will work for a wider range, nearly all the time. I think about a 1/4 of the replies here thus far have basically been, "Why don't you just do simple thing X"--implicitly suggesting that, "this works well enough for me, and probably will for you too." Problem is, "just mix in a blender at high speed some roleplaying, a skill roll or two, and adhoc DMing to smooth out the rough edges," is often a satisfactory system for a substantial number of people. But it isn't satisfactory for everyone, and certainly reveals its unsatisfactory rough edges on occasion even to people who like it. And if that is the best a game designer can do, then frankly they might as well admit that they are focused on setting and drop the game designer moniker. Because people can find that design easily enough for themselves (and have, many times).
Now maybe from a marketing or mass appeal perspective, D&D being a gateway game, should settle for that kind of solution. Focus on setting and production values--and keeping the costs down otherwise, to move products at a low margin. If one makes that business case, then they don't need much in the way of designers. They just need someone to tweak skill lists and feats and powers, clean things up, and play around the edges, the way the 2E team did.
If designers are going to earn their keep, I think they can try for something a little more ambitious than that, though.