Long post ahead.
I have never been a huge fan of the Legend of Zelda games, at least not since Link to the Past. There are plenty of reasons to like Zelda-- the cohesiveness of the design, the high production values, the archetypal storytelling-- but for some reason the games don't keep me engaged. And I think it's because at a high level they are completely deterministic.
In any Zelda title, you can open the pause menu and view a status screen that shows your progress towards collecting all the artifacts required to complete the game. From the outset, I know that I need to obtain this specific set of widgets to 'win,' and end the game. Subconsciously then, the game becomes not about exploring a world to find a succession of interesting places, and finally defeat Ganon and free the missing princess (the primary conundrums of the narrative,) but instead of fill in this chart on my pause screen, piece by piece, in order, until the game is over. It become mechanical-- about completing a series of incremental acts, all predetermined and laid out to me from square one, instead of organically progressing towards a larger final goal. And I naturally disconnect from that kind of experience. It feels too artificial, like I'm just going through the motions of pressing the right buttons at the pre-determined times, making no choices or impact of my own.
Now of course this is true of almost all games-- in any given video game, I can only perform a pre-determined set of actions, leading to a pre-determined set of outcomes. However, the element of predestination in one game can be much more or less explicit than in another. It comes down to both structure and presentation. Half of the issue is just how linear the progression of the overall game is; the other half is whether or not the player sees it coming.
So, you could think of the progression through a narrative game as a piece of twine, with a hung curtain hiding its endpoint from its beginning. As the game is played, the player pulls the twine towards himself, gradually revealing more and more of it from behind he curtain, until he reaches the endpoint, revealing the full length of the twine and completing the game.
With a Zelda game, to my eyes the curtain is largely transparent, evaporating the anticipation of finding out what comes next, and the illusion that my progress forward is determined by my own decisions. Instead of building a final composition within the game's framework, I'm simply painting by numbers.
But, Zelda aside, I think that the twine analogy can be a useful tool for plotting the structure of any narrative game. I am talking about games that have at least one predetermined path through from the start to at least one discreet ending. Each possible path, each parallel universe in this string theory, is an individual thread that makes up the piece of twine. The defining attribute of the twine when describing a specific game's structure is how the individual threads diverge, reconverge, and eventually terminate.
Let's take my pet example, the posterboy for linear single-path structure, F.E.A.R. It would be a single piece of twine; there are no branching paths or significant player choices that meaningfully impact the outcome of the overall progression. F.E.A.R. is a single, unified length of twine that the player continually pulls on from start to finish. However, it's still compelling because this level of authorial control allows the designers to take advantage of more cinematic techniques than with other structures, and the constant anticipation to see what lies next behind the curtain keeps the player engaged. Other games that might be a single length of twine are Gears of War, all of id's FPS's, and Prince of Persia: Sands of Time.
Now consider a game like Hitman. The game is composed of a set sequence of levels, but the player's progression through each individual level is not restricted to a single thread. Instead, available to the player are a great number of possible threads to follow to the required objectives, none of which might be obvious to the player upon first glance. Hereby, at the beginning of each level the length of twine diverges into perhaps a dozen individual threads, which then tangle together into a loose knot; the player must examine the gamespace to pick out a viable thread to follow. The longer the player spends with this knot, the more it is unraveled, until the player has fully untangled the threads from one another, allowing him to see all the possible paths from the point of divergence back to the point of reconvergence (completing his objectives and exiting the level, which must happen for the player to proceed.) The beauty of this approach is that the player only need untangle one single thread and follow it to progress to the next knot; on the other hand, a devoted player can spend hours upon hours on a single knot, untangling it further and further until they have separated and examined each of the individual threads therein. After traversing the sequence of individual knots, the twine reconverges one final time, presenting the player with a single unified outcome to the game overall. Other games that might share this structure are Thief: The Dark Project and The Warriors.
Now picture an open-world game, like Grand Theft Auto 3. Centrally, there is one primary thread. This thread leads the player through the core story missions-- the critical path-- that ground the game in a central sequence of events. The player can simply follow this single thread through to the end, experiencing the full progression of the core experience, and complete the game, reaching its endpoint. However, all along the length of twine, additional threads branch off and terminate on their own. These are the sidequests, the optional missions that the player may take on to acquire additional assets ingame, or simply to flesh out their own experience. These secondary threads may be passed up entirely or returned to after skipping them initially, but they never impact the central thread. Eventually, regardless of how many side-threads the player has followed, they reach the same conclusion upon finishing the central thread, and completing the game. There is only one overall endpoint. You could picture this piece of twine taking the shape of a pine tree-- the critical path is the trunk leading all the way to the top, while the secondary threads are the branches the split off and give the tree its distinctive shape. Other games that might share this structure are Morrowind and Oblivion, Bully, or Fallout.
Though I do love many of the above-mentioned games, something that disappoints me about them is that they all lead to a single conclusion, regardless of the broader decisions the player makes beforehand. Any meaningful impact that the player's choices seem to have on the overall arc of the game narrative is an illusion. This is why games that feature multiple conclusions, dependent upon meaningful choices made by the player throughout the game, are particularly interesting to me.
One title that might come to mind then would be Deus Ex. The player makes choices throughout the game, many seemingly meaningful-- whether this or that character lives or dies. However, when you examine the game's structure, you can see that it has most in common with the series of knots above. While I can approach the game's challenges with a number of different strategies based on the supplied affordances, my decisionmaking loses meaning when I reach regular, pre-determined narrative chokepoints. The most notable of these might be the fact that I must flee from my government agency and join the NSA; regardless of the choices I make up to that point, the game proceeds down a single vector before again splitting off into a number of separate threads. The threads reconverge periodically throughout the game, at points where the designers have made a central narrative decision for me. Likewise, the multiple endings are an illusion; all of the threads reconverge again immediately before the ending-- at the doors of Area 51-- before splitting cleanly into three distinct threads, each of which represents one ending I may choose at my own whim. Like in the games above, nothing I have done up to this point has any impact on the ending I receive. Instead, I'm told to choose either Door A, B, or C, and walk away with the ending I've picked. The other game I can think of that shares this structure is Vampire the Masquerade: Bloodlines, which is even more transparent as it hands you the possible endings as options in a dialogue tree.
S.T.A.L.K.E.R.: Shadow of Chernobyl shares something in common with the Deus Ex endgame structure, but distinguishes itself by providing an entirely open-world framework as opposed to DX's controlled narrative, and offering at least one ending which is based entirely on the player's own incremental decisions made throughout the body of the game; and depending on the player's approach, this may be the only ending available for that particular campaign.
S.T.A.L.K.E.R. is an open-world first-person shooter with moderate RPG elements. How the player navigates the gameworld is up to them, as is the range of actions they may take. All but the NPCs most essential to the critical path can be killed, factions can be joined or turned against, and the player can generally exploit the game with idealism, pragmatism, or profit in mind. The interesting part of the game's twine is that it does have a central thread that the player may follow to the end, as well as subquests that split off; but reaching the end of the central thread is not guaranteed, and which tangential threads you follow actually impacts what conclusion or conclusions you receive.
Basically, your destination at the end of the game (spoilers ahead) is the Sarcophagus that houses the carcass of the Chernobyl nuclear power generator. Inside the Sarcophagus is rumored to be a Monolith, that will grant the wish of anyone who addresses it. Throughout the game, a variety of metrics are tracked in the background, unbeknownst to the player--which factions he has aligned with, how much money he has hoarded, and so forth. Dependent upon his overall behavior throughout the game, he will only make one resultant wish when he addresses the Monolith at the endgame (for instance, "I want to be rich" if he's gathered a certain amount of money.) Hereby, the player's own incremental decisions and demonstrated disposition throughout the game have a meaningful impact on the outcome of the overall narrative thread, lending gravity to his actions and acknowledging his uniqueness as an individual actor.
However, the attentive player will follow the central narrative thread all the way to its endpoint and discover that there is a secret entrance to the true source of the Monolith's power, deep beneath the Chernobyl NPP. The player's personal Monolith ending is also available at this point, but the player is rewarded for completing the entirety of the central thread by being given the option to explore a different, more narrative-focused than player-focused, conclusion. The player may decide to either join with the C-Consciousness Project that powers the Monolith, or destroy it, and with it wipe the Zone around Chernobyl from the Earth. Again, this ending is not available to all players-- the Chernobyl NPP can be reached without having discovered the tools required to reveal the secret entrance. If the player investigates the central thread fully they have access to the game's "true" ending, but only if they have, through their own decisions and effort, acquired the tools to reveal it. Hereby, even the "stock" endings to the game are a personalized conclusion based on the individual player's actions earlier in the campaign, while the ability to choose between one encountering them them allows the player to maintain the power of decision up to the very end.
In S.T.A.L.K.E.R.'s case, the twine immediately diverges from its starting point. It splits into two threads-- the central thread, and a secondary tangential thread, which represents the player's side-actions, the ones that determine his Monolith ending. These two threads proceed parallel to one another. The tangential thread splits further and further as the player defines his secondary ending, eliminating possible tangential threads through his own actions. Likewise, at the point where the player has the option to pursue the secret entrance in the NPP, the central thread splits off into two-- one branch is a dead end; the player will not reach the C-Consciousness project, and only has the secondary ending available, while the other continues the central thread to its true conclusion. If the player does follow the central thread to its true endpoint, the central thread and secondary thread reconverge immediately before the finale, at the doors of the NPP. From there, the player may explore the NPP and decide which path to take. However, unlike Deus Ex or the like, not all endings possible at the campaign's start are available to the player one they reach this chokepoint, but instead only those he has made available to himself through his own actions over the entire course of the game. He may have three possible endings available, or only one, never two. The point of this is, as bolded above, that the possible conclusions are entirely dependent upon the meaningful incremental decisions made by the player, organically over the entire course of the campaign. In the end, everything I do as a player matters. Conversely, in a game where all possible threads reconverge before the twine's endpoint, and are all available for my unrestricted choosing, any decisions I've made up to that final narrative chokepoint become meaningless; all those earlier threads might as well have been one single length of twine.
I feel that player-dictated feedback is important, as it's what sets games apart from film or books. Even when it's simply choosing a series of binary forks in the path, that they actually lead somewhere unique to that player is essential to the experience. Gradually revealing a length of twine can be enjoyable, but is it any different, really, than the film experience? And if my individual actions don't impact my branching campaign's outcome, have I really accomplished anything unique? If we return to the tree analogy, as an invested player, I don't want my twine to come out looking like a pine, or a palm, but a full-grown oak.
Long post ahead.