I had originally intended this to be a comment on the original post, but it ended up being entirely too long. At any rate, this is a direct response to Mr. Warren Spector's recent post on Telltale.
It's a bit daunting to comment on an article by a designer whose games and design experience I hold so much respect for, and I fear that everything I'm going to say has probably been better stated elsewhere already, but I would like to cautiously venture into this discussion to add what little I can to this topic.
I feel that the difference between the experience found in The Walking Dead, et al. and that in "games" is the difference between playing and gaming. I'm very much reluctant to try to define common words in specific ways again (I've already done an odious amount of that in previous posts), but I believe useful distinctions are important for clarity of analysis, so here goes.
The small and/or conventional ways Mr. Spector mentioned (which weapon to use, direction to go, resource to expend)—these are all choices meant to game the system; they are specifically an interaction with the rules of winning.
In contrast, the "magic" of the player forgetting the player character is about choices derived from personally inhabiting the moral space of the game—an interaction with the rules of being (and trying to be a specific type of person definitely imposes its own rules).
Mostly, we play roles or characters (be it the one of the fiction of the game or ourselves imposed upon that fiction), we don't game them. That is to say, we don't try to bend our characters to fit winning conditions if we are playing a truly realized person/ourselves. But this distinction is far from strict; when we stick to rules and behaviors dictated by our character, we try to "game" the win/loss conditions to allow us to be that character, and the things we allow or disallow ourselves to do can cave in to the demands of winning.
So, when we min/max optimize and create a character with no intelligence or charisma, we tend to forget or disregard the impact these attributes would have on an individual--as far as the game allows us to (a prioritization of gaming over playing). Oppositely, if we intentionally handicap ourselves when taking morally upright choices or mechanically inferior weapons/classes, we challenge ourselves to see how far we can adhere to a set of pre-existing, personal values ("I am a virtuous person", or, "I am a skilled gamer"—in the second example, we can see that play doesn't have to involve narrative).
To put it in a more succinct way, for the purposes of this discussion, when the choices a player makes are defined by rules/values existing only inside the game, that's gaming. When the choices a player makes are defined by rules/values created or held by the player, that's playing. (As might be apparent, it's pretty much impossible to have a pure state of either).
I believe this distinction is useful in that it makes it easier to think about games that make "play" a game of itself, and also about how to better design such games. This, of course, has always been an important goal of traditional RPGs, where RPGs have mostly relied on pitting narrative choices against mechanical choices to pursue this goal.
But in terms of games whose mechanics are purely narrative, the puzzle or obstacle becomes not about how to win an encounter, but which choices best allow a player to fulfill the character she is playing. These types of games are "won" when the player is able to say, yes, I've maintained the character I want to be and still managed to get the results that I wanted. In the same way, they are "lost" (or rather, poorly designed) when either the player is presented only with options that go against her character, or the results of adhering to her character are always invalidation or loss (because this is so personal, at that point the game becomes unplayable).
Some design thoughts: Mr. Spector espouses the TTG games as paragons of games that operate mostly through play, but perhaps a better example can be found in Life Is Strange. Firstly, Life Is Strange shows that these games don't have to force uninformed and arbitrary choices to maintain tension and stake, because the stake is already the player herself. Moreover, because the choices are informed, they feel much more the player's own, which makes inhabiting and playing that game much more "magical", to borrow Mr. Spector's phrasing.
Secondly, Life Is Strange recognizes that optimization is important for play as well—we want to be as close to the person we are trying to recreate as possible, and Life Is Strange essentially builds in a limited "respec" of the player's identity-choices as one of its central mechanics (it's a time rewind feature). And, like any other well-designed respec feature, this function allows the player to explore the mechanics of the game (in this case, the personality mechanics of the NPCs the player deals with) without necessarily giving away the "correct" answer. (Plus, it lets the game do some neat things with memory and logic puzzles.)
Lastly, I would actually disagree with Mr. Spector that a blank slate provides the best vehicle for allowing the player to be herself in these games, and employing the opposite strategy is what Life Is Strange excels at. Life Is Strange has a strongly defined player character while going out of its way to make the player character as relatable as possible. The impact of this on the player experience is that, because it establishes a powerful narrative logic, and because the player is more likely to adopt the player character's character as her own, it mitigates the damage that might otherwise be done when the player is presented with none of her own choices. The choices mostly all seem logical and natural, reducing the number of "I wouldn't do any of these, and I'd solve the situation better by doing X" situations.
Importantly, a robustly defined protagonist also provides context and logic for each decision—it creates "gameplay" by more naturally and observably "telegraphing" social behavioral cues to inform player choices, which in turn produces more predictable and satisfying outcomes for the player when the player reads those cues correctly. And, while specific branches may be the same, the manner in which the player picks up or interprets these cues changes from player to player. The final outcome may not change, but the social or emotional intelligence and logic employed behind each player's decisions differ, which means the precision of each decision is not universally the same, and the internal problem solving undertaken in fact varies quite significantly.
Of course, the player can't produce surprising outcomes per se, but she certainly can produce surprising reasons for picking those outcomes, which I believe is rather close to a "player-driven solution to a problem". I would argue, then, that it is simply that TTG games have provided very little room for the player to actually develop individual theories that drive play choices by so strictly limiting what the player can know, or more or less eliminating the entire feedback/learning loop altogether.