Hitman: Blood Money. I'm going to go out on a limb and say it's the best game I've played in... six months? A year? I'm a longtime Hitman fan, ever since playing the pre-release demo for the first game of the series in late 2000. My friends and I absolutely fell in love with Hitman: Codename 47. Due to the nonlinearity and massive replayability of each of the levels, we spent over six months playing the game on a near daily basis, re-exploring and pushing the boundaries of Lee Hong's compound and the hotel in Traditions of the Trade over and over again. When we discovered the slow-motion, freecam, and giveall cheats, the game's lifespan was extended another six months. Codename 47's engine brought a bunch of revolutionary touches together for the first time-- features like ragdolls physics (which were meaningful to the gameplay through the body-dragging mechanic) dynamically shifting cloth banners and foliage that reacted to the player's movement, glass that dynamically shattered and fell, persistent bullet holes on both world surfaces as well as characters, and more. Together with a unique spin on stealth gameplay-- more like infiltration, really-- and freeform levels you could replay individually from the title menu at any time, Codename 47 was truly unique, something I'd never seen before.

But I digress. Hitman 2 and Contracts lost a lot of the strengths of the first game, making the play more restrictive, turning the enemy AI into a frustrating mess, and skimping on the graphical touches. For years I was on a disappointment yo-yo, get psyched for Hitman 2 before being horribly disappointed, getting my hopes up for Contracts only to lose faith in the series again. Finally, during the lead-up to Blood Money, I started to hear a lot of exciting info on the Games forum. I couldn't help but get hyped up once more for another Hitman game, reluctant as I was. And to my surpise, and great relief, this was the one they got right. It brought back the magic of the first game, and added so much more. Incredibly, it was worth waiting 6 years for.

The good stuff in the game is the breadth of missions and the amazing environments that IO has built; what makes the game great is the way that IO opened up the play, giving 47 more abilities, and therefore the player more leeway in accomplishing his goals. 47 can now throw items, which opens up a whole range of possible offensive and passive actions, such as tossing a mine over a wall and setting it off, throwing your equipment into a guard's field of view to draw his attention, or simply throwing a knife at a dude to take him out. 47 also has new ways to get up close and personal, using melee attacks and human shield grabs. The ability to push NPCs around, shoving them over railings and down stairs to set up accidents, is another hilariously useful addition. There are more ways to do any given thing, and more Plan B's available when things get hairy. Essentially, the designers implemented a more robust set of affordances to the game, in the form of new actions that 47 can perform, and new types of items for him and the NPCs to interact with. Instead of the limited inputs of "shoot or strangle" as in the earlier games, 47 can affect the world more subtly by throwing objects around it, and the AI reacts more robustly by taking these objects into account and reacting to them. Similarly, 47 can now affect NPCs more directly but less overtly, by pushing them or grabbing them as human shields, to move them around the level. To use another Word, the player is given more agency in how he wants to approach each mission by the gameplay's range of affordances, leading to a more nuanced and satisfying experience. Put the rich gameplay and lush environments together with a tightly-constructed but unintrusive tale of political intrigue, and you've got a game that grabs you and just won't let go. It's outstanding. Simply outstanding.

One thing I've seen people say often about the Hitman series is that they're not action or stealth games, so much as puzzle games. This springs from the fact that all of the NPCs behave in an identical, clockwork fashion each time the player starts up any given level. If Guard A walks through Door X at the :25 mark on your first run, he'll walk through Door X at the :25 mark on your second run and third and thousandth. Barring player disruption, the actions of every character in every map and the patterns they create will be exactly the same every time. This encourages one approach of devising "perfect" playthroughs by observing the level, reading the predictable patterns, and then finding a way to time your own actions right so as to slip between the gears of the clockwork and accomplish your mission. Someone described the experience as similar to being Bill Murray in Groundhog Day, but to much different ends. It's an apt comparison, and it's one of the weakest points of a game that I still love regardless.

What I'd like to see in the Hitman framework is something a little less predictable. Currently, NPCs have a series of nodes that they path to in a single, consistent pattern. Say that a man has four nodes: his sofa, the bar in his kitchen, the swimming pool, and the shower. He sits in the sofa, gets up to go to the bar, takes a quick swim, showers off, and sits back down on his sofa. He repeats this forever. I propose a slight alteration to this scheme. Each NPC still has a limited total number of nodes that they path to, but it's not in a static sequence. Instead, each node relates to each other node through a variety of properties that determine the likelihood of pathing to any given node when he leaves his prior node. So, in the above example, if the man starts out at his sofa, he might have an equal likelihood of going to the bar or the swimming pool; his likelihood of going to the shower from the sofa might be extremely small, or nonexistent. Since the chance of going to the bar or pool is equal, the destination is chosen at random. From the time he leaves the sofa, both the sofa node and swimming pool node start to build priority; the rate of priority gain can be set for each node. When the NPC reaches any given node, its priority is reset to zero. So, from the beginning of the level both the bar node and swimming pool node begin building priority; when he leaves the sofa, its priority begins to rise; when he arrives at the bar, its priority is reset to zero. When he leaves the bar, the swimming pool, which has been gaining priority longest, is first choice for the NPC.

Now, to keep this from turning into a more complicated system of achieving the same results, some randomizing factors come into play. One: for an NPC that has three or more nodes, the NPC randomly approaches one of the two nodes with highest priority. So in this case, he might go to the swimming pool, or he might return to the sofa. Additionally, some NPCs share nodes. So if someone is already on the sofa, he will be forced to go to the swimming pool. When an NPC has half a dozen or more nodes, and he shares these with multiple, other NPCs, you can see how his patterns would become much less predictable.

The point is, this guy gets up and goes to the bar. Then does he go back to the sofa, or to the swimming pool? Let's say he hits the pool. He swims for a bit, then he's hardwired to hit the shower after the pool. Then what? He either returns to the bar or the sofa. Then what? I find this much more interesting and exciting than knowing that he will move from sofa to bar to pool to shower to sofa in a neverending loop.

This is a simple example, but my goal is to maintain some predicatability--the range of possible locations and actions for any given NPC is limited-- but make the NPC behavior in the game dynamic in nature, instead of static. The gameplay then becomes a matter of observing NPCs, determining all their possible actions, and then adapting your approach to fit what they might do. Predicting their location and your own options becomes a matter of reasoning and improvisation instead of memorization. It also ups the tension, because you can't be absolutely sure who might be coming around the corner at any given time. It changes each map from a static, predictable, and therefore gratingly artificial and eventually boring clockwork, to a box of low-key controlled chaos. Dynamism. Tension. Improvisation. Fluidity.

Another jarring aspect of the gameplay in Blood Money--and this goes for many stealth games as well-- is the absolute conviction with which NPCs stick to their predetermined routines, even following extreme disturbances. This is most apparent in the levels that take place in a smaller, more residential setting, such as A New Life or Til Death Do Us Part. You can set off a mine in the middle of the dancefloor or snipe the groom at the wedding, and after an initial freakout, all the NPCs will go back to their normal routines. You can burn the wife to death in A New Life and within 5 minutes the husband will be back to nonchalantly watching TV in the front room. It's completely unrealistic and weird. If half the party guest at my wedding got blown up by a mysterious bomb, do you think I'd be up at the altar 15 minutes later?

What I propose are "standard" and "panic" positions. Standard position is the set of nodes and behaviors that NPCs on the map initially use. Panic position is engaged after a major disturbance-- when an explosion happens, when gunfire is detected, when an important NPC is killed. Panic position engages a secondary set of nodes and behaviors more consistent with a state of great duress, which continues for the rest of the mission. So in A New Life, if the husband was sniped through the front window, the wife would gain new nodes: crying by her husband's body, hiding in her room, and taking swigs from the bar; the guards on the map would fan out to cover the area more vigilantly; the hired help would leave the area. On Til Death Do Us Part, after a major disturbance such as an explosion, the father of the bride might hole up in his room with his gun in the ready position, while the groom might go aggro and begin searching the grounds for the perpetrator. The bride would hide in her room with a number of her bridesmaids. The guards would go on more active search patrols. Again, this would make each mission more dynamic, as the entire place has two possible states, determined by the player's actions; along with the more dynamic system of NPC desire nodes, all but the highest level predictability is eliminated, giving the player a different experience each time they enter a level, and removing the somewhat robotic feel of the missions. Working towards dynamism and a system of strictly controlled chaos-- knowing what could happen, but never being SURE-- would push an already hugely enjoyable game to a whole new level.

Let's see what IO does with the next in the series :-)

No comments: