This is one of the great difficulties 5e will face. This element of play has gotten slimmer with every edition...never totally gone, but the mechanical and slick we get, the less play like this factors in.
I partially agree, but I'd like to say that one of the reasons I prefer 3e to 1e (and I've played a lot of 1e) is that - especially as a DM - 3e gives me a lot more tools for adjudicating off script and creative play. One of the real problems in 1e is that with no skill system, no unified saving throw mechanic, and with no combat manuever system, it was often hard to know how to treat any given proposition. How would you figure out the chance of success and the effects of success or failure?
This was a such a ubiquitous problem that it was generally ignored for fear of wrecking the game. A proposition even so simple as, "I climb the wall.", raised a minor panic. 'Climbing' per se - like everything else - was treated like a siloed ability. It was obvious how you resolved it in the case of the thief, but how do you resolve it for the fighter? Did the fighter automatically fail? Did he climb as well as a thief of 1/2 his level? Did you handle it on a case by case basis, essentially attaching the fighter's chance of climbing to every wall and then figuring out what the theives modified chance was hopefully in a way that didn't make the thief feel even more useless than he was.
The great thing about having a complex set of rules is, even if you don't normally use all of them, if one of the players pulls a stunt, you've got something to adjudicate it with. You can always think something like, "That's not in the rules, but its reasonable. OK, if he passes a jump skill check, then I'll grant him a +3 circumstance bonus on attacks and a virtual power attack feat for the duration of the round." It was a lot harder to approach the problem in 1e, and in practice you got one of two equally disfunctional responses: "No." or "Yes." "No" was bad because since 1e defined almost nothing explicity, it limited you to a very narrow set of repetitive actions. "Yes" was bad because it tended to make the game irrelevant, and the real game was whittling, browbeating, intimidating, confusing, conjoling, bribing, flattering, tricking, and anticipating the DM so as to always get told "Yes". I saw many 1e players who didn't make true propositions. Instead of saying what they wanted to do, they said what they wanted to happen. Then they'd argue with the DM that what they wanted to happen should happen, and if the DM said, "No, something else is going to happen.", they'd argue that they'd never do _that_ and that they should be allowed a different action. The real game became "Mother, may I?" mixed with, "Can we get a pool, Dad? Can we get a pool, Dad?"
I really feel that 3e meets both groups halfway (or nearly so if you do a little tweaking). The players that have to feel in control are given a clear relationship between what they purpose and what happens. When they propose something, I can outline my conditions: "You'll have to make a DC X skill check, and you'll get the following result." This let's them get a realistic sense of the rick they are taking and the reward they will recieve without them needing the unrealistic ability to know ahead of time how things will turn out. Players that on the other hand don't want to be hidebound and who enjoy the free form gaming that pen, paper, and imagination allows, can propose wild and crazy things without me having to say "No, if it's not a power on your character sheet, you can't do it."
If I was a 5e designer, that would be the tangental turn I'd want to make on the direction D&D (and its clones) have been going. The system needs to be rules heavy BECAUSE it needs to be freeform. Whatever rules you introduce have to add to that goal, or they aren't good rules for the system. The reason I stuck with 3.0 core rules and tweaked it myself, rather than expanding into the supplements or (*shudder*) 3.5, is that it seemed to me that the core 3.0 rules held that promise, but that the rules added to the game detracted from it. You started seeing lots of feats that didn't make you merely better at something, but which siloed a perfectly ordinary stunt that a kid on a play ground could attempt - like throwing someone you'd grappled - off as an ability that required a special feat. You started seeing lots of Prestige Classes of a generic variaty that seemed to imply that you couldn't be this thing unless you took this rigid, narrow, and inflexible class. And you didn't see a lot of addressing of what I saw as 3.0's core problems.