The recent hands-on impressions of Deus Ex: Human Revolution all emphasise the fact that the game stays true to the core tenet of the original: leaving the player free to choose between a wide variety of approaches to complete any given objective. As such, Deus Ex became well-known for engendering a deep sense of agency, but this applied more to its gameplay mechanics than to its story. The broad strokes of its plot were set in stone, with only the minutiae left up to the player. Naturally, a fully dynamic story is something that’s nigh-on impossible to attain in a game that isn't something like Minecraft. It’s less about what the player wants to do and more what the developers allow him to do. Those two can be the same if the developers did their job with a modicum of foresight, but it remains a compromise at best.
Deus Ex: Invisible War sought to foster that same level of agency, and in the first few levels, it seems to succeed. The player walks around in cities and interacts with NPCs who all give him conflicting goals on both a microlevel (collecting evidence on a corrupt senator or helping him out for cash) and a macrolevel (aligning with one of the game’s central factions against the others). But after a recent playthrough, it occurred to me that there is a very specific way in which the game ultimately fails to engage the player.
It first struck me when watching this post-mortem on Deus Ex: IW by its creative director, Harvey Smith. He specifically mentions the consolidation (console-idation? I kill me!) of the swimming skill and aqualung augmention in the first game into a single biomod in the second. While his example is technically a theoretical exercise (there is no swimming in Deus Ex: IW), it applies to many of the design decisions made for Deus Ex: IW and provides an interesting vantage point on what I believe to be one of its largest flaws. As Smith puts it, even though this consolidation process makes no difference on a mechanical level, it does curb a player's fantasy that he is exercising his own authority in choosing how to develop his character and how those choices are reflected in the game world. In my opinion, Deus Ex: IW breaks this fantasy in a myriad of seemingly innocent ways.
The key is Smith's use of the word "fantasy" and how it relates to player agency. The relation can in fact be as simple as calling player agency a fantasy. Agency only becomes an impediment once that fantasy is broken (1). As long as the player has the impression that his actions and choices matter, he will lend them credence and meaning (regardless of how many avenues there actually are). The same principle applies to level design, which can make a world seem more alive and expansive than it really is. Crafting agency then goes beyond offering visible dialogue choices or gameplay approaches. The former influences the story while the latter influences gameplay (any crossovers notwithstanding), and Deus Ex: IW focuses on the wrong end of the equation.
The game attempts to offer the player full agency in its storyline by allowing him to align himself with every faction in the game. This is most significant when considering the Knights Templar, who are consistently portrayed as irredeemably evil and easy to hate. They are the only faction which the player has to fight until its leader, Saman, attempts to sway him to his cause close to the end of the game. This is sorely inconsistent with everything that preceded, even (and especially) if the player actually decides to join them. Contrarywise, Deus Ex offered no option to stay with UNATCO, but in light of its overarching narrative, it couldn't, nor did it have to (2). Of the "flaws" that Deus Ex had, this was one its sequel did not need to fix.
Gameplay-wise, Deus Ex and Deus Ex: IW are by and large similar beasts. Whenever the player meets an obstacle, he consumes a certain number of his resources to get past it, be they lockpicks or explosives for doors, multitools or nearby computer terminals for cameras, and lethal or non-lethal weapons for guards. Deus Ex: IW drastically reduces the number of available resources to streamline the interface and ease up the learning curve (for instance, doors are now indestructable and multitools also function as lockpicks), but it's less about this reduction and more about how all the different approaches are implemented in gameplay so as to maintain (the illusion of) agency. The best illustration is how Deus Ex: IW handles keycodes.
In both games, keycodes are numerical and either handed out by characters or gleaned from datacubes. The difference is that Deus Ex makes the player physically type the code into a keypad, while Deus Ex: IW automates this process (using a keypad opens the corresponding door as if it were a simple toggle). The outcome is the same from a mechanical standpoint, because only one keycode is correct. But the agency effected by the player is diminished because he wasn't allowed to enter that keycode himself. This also eliminates all possiblity of guesswork from Deus Ex: IW, since the successful operation of a keypad depends on having completed the interaction which yields the correct keycode, rather than the keycode itself.
Ostensibly trivial design decisions such as this, when added up, contribute to the player losing his feeling of agency, even if it is just that: a feeling. Player agency's most devoted advocate Clint Hocking has posted a piece on its (d)evolution, which neatly dovetails with the above assertions.
(1) This is not the case when a game is specifically designed around multiple choices (in story and/or gameplay). An example in Deus Ex is the fate of Paul Denton. JC's actions decide whether he lives or dies, but this is never made explicit. Many players attested that they didn't realise they could save Paul until it was pointed out to them. Only playing the game both ways provides the full story. Outright gimmickry in other media (such as Run Lola Run's structure) would be needed to compete with the elegant multiple narratives that separate playthroughs in games can provide (a point gleaned verbatim from this article).
(2) Ironically, unused audio files suggest it was initially planned by the developers. This conversation has JC disbelieving Paul's evidence of UNATCO's corruption, but it's likely that a subsequent event would have made it untenable for JC to keep refuting Paul (perhaps the raid by UNATCO troopers that follows that scene anyway).