Monthly Archives: February 2008


In other news, I’ve been watching and judging a TV show called jPod on CBC. It’s also named after a Douglas Coupland book I read two years ago, called JPod. The book is good too. The show is – get this – about a bunch of game developers! It involves swinger parties, middle aged housewives with marijuana grow-ops, gay innuendo with Chinese mafia kingpins, and online trysts with twin Wal-Mart greeters. The first episode or two are a bit rough, but it really picks up after that. Seriously, guys, I love this show. It’s fresh, it’s quirky, it’s relevant, and it’s hilarious. Do whatever you need to do to watch it. There’s no reason this show shouldn’t be a huge in-joke in the games industry.

Back from GDC

I think it was around the time that I staggered out of the Portal postmortem, exhausted, sick, after surviving on a meager ration of burgers and pizza for a week, that I realized that GDC was the best thing ever


The conference wore me out. Every minute of every day presented another unmissable opportunity. I got to meet famous designers, or old friends I had never actually seen before. There was always another presentation or roundtable. Every night there were several parties full of industry folks, which were genuinely fun. Unfortunately for my health, I consistently decided to spend my time taking opportunities instead of eating properly or sleeping. It was totally worth it.

There are two main parts to the GDC: The presentations, and the networking. I’ll start with the presentations. Namely, three of my favourites:

Conflict Resolution Without Combat was a roundtable discussion focusing on creating games wherein a player can resolve conflicts using methods besides blowing people apart with large guns. I have a strong interest in this subject. Most games seem meaningless to us because they deal with subject matter – war and ruthless violence – which is so narrow and outside a normal person’s experience. The people at the roundtable brought up a variety of examples of games which have done conflict resolution in the past, possible methods of moving forward, and how people in real life resolve conflicts without murdering each other. It was great to see how much interest there was in this area.

The Portal postmortem, subtitled Integrating Writing and Design was interesting, and not just because it was hilarious. I came away with a variety of ideas and principles to explore relating to achieving tighter integration between story and gameplay, designing carefully mediated experiences which don’t leave players stuck, and achieving a high happiness-to-production-cost ratio using creative design methods. It was also interesting to see how the creators of Portal achieved so much by really putting a personal touch into the game.

Finally, Chris Hecker’s Structure vs Style was a very interesting look at attempting to apply similar method to solving what look like very different problems. Chris wants to find a way to apply the philosophy that drives triangle-based 3D rendering to artificial intelligence design. The ideas he presented demand a lot more thought and touch on almost every part of the design process.

gdc3.jpg gdc4.jpg

The Inscrutability of Hitman: Blood Money

I’ve been playing a bunch of IO Interactive‘s Hitman: Blood Money. I think the design is a good subject for a design review. It’s a unique design, it excels so much in some areas, and falls pretty hard in others.

For me, Blood Money is a great game. It is the most cinematic game I know. The environments are unique but familiar, beautiful but believable. Every gunfight, sniper shot, or fiber wiring feels like something out a hard-boiled action movie.

But for all it’s good parts, the game has deep-seated flaws in its design. I’m going to focus on the faults. Don’t think I don’t like the game – they’re just more interesting to analyze.

Inscrutable Mechanics

From a Blood Money FAQ:

“So, in my first version of this, I wrote that I couldn’t find a way to accident Vinnie. Some people wrote in to say he could be thrown down the kitchen stairs to his death, but after multiple tries this does NOT work for me, so it’s not going in the guide. I tried throwing him over the balcony, of course, as I’m guessing most people did, only to discover its a non-throwable balcony. I even tried the open window by the squeakie toy, but no dice.”

-A Blood Money FAQ by ‘Guybrush Threepwood’

This paragraph from a Blood Money FAQ is a perfect example of one of the main problems in Blood Money: The game mechanics tend to be invisible, non-obvious, or very ambiguous.

We can think of games as systems to be learned. People derive pleasure from learning new, interesting things. In the case of games, many games give gratification by providing an interesting system – the mechanics of the game – to be decoded and then manipulated.

The issue with this in Blood Money is that the system is almost impossible to learn because the game rules are so inconsistent and non-obvious. There are many completely unpredictable gameplay mechanics which appear in one level and then vanish forever. Other gameplay mechanics appear repeatedly, but in non-obvious places. You will often need to figure out some completely novel game mechanic, which does not derive from what you learned previously, to move forwards.

The other problem with this situation is that it turns off newer players because they end up doing a long slog of trial and error to learn the missions before they can really enjoy the game.

Poorly Defined ‘Domain of Simulation’

Blood Money takes place in a realistic real-world environment. The developers seem to have attempted to allow a very wide variety of real-world actions to be performed by 47. The problem with this is that they’ve pushed so far into the real world space that they’ve overextended themselves.

Game players know that not everything they could do in real life will be possible in the game. Most games focus on a specific type of action from real life and simulate only that. For example, after playing DOOM for a few minutes, few players would expect to be able to talk to the zombie and ask him to be friendly, or to be able to use a radio to call for help, or eat nearby food, or take off your armor. DOOM focuses on a very narrow range of actions – running, shooting, pushing buttons. There is no confusion as to what possible real-life action the game will allow you to perform. You shoot, you run, you push buttons. That’s it. That’s DOOM‘s ‘domain of simulation’.

Blood Money‘s domain of simulation, however, is impossible to pin down. We can never be sure what real-world actions the game will allow 47 to do. Dialogue, for example. 47 can say certain things to certain people, but not anything to other people. Thus the dialogue game mechanic essentially needs to be re-learned for each level because it has completely different and non-obvious effects in each level.

This problem is so endemic that it is wholly possible to play the entire game without learning many of the mechanics. Most players probably finish this game without learning half of what it has to offer. You can only get all the goodness out of this game if you read a FAQ.

Examples of the domain of simulation problem:

Costume Rules: Often it is impossible to tell where you are allowed to go in a certain outfit. The only feedback comes when someone sees you in the wrong place and either shoos you out or pulls a weapon and starts firing. Also, you can carry certain weapons in certain areas, but guards will open fire if you take them into other areas. There is no indication as to why, or which weapons are allowed in which areas.

Containers: In many cases, you can hide a weapon in another object to get through scanners and searchers. But there is no logic to which items you can hide things in. Sometimes you can hide a pistol in a cake, but other food items cannot be used. Some containers can be picked up and used to hide items, others cannot. If you see a cake or some container-like object on a table, there is no way to tell whether it is a static decoration, or useful as a container.

Object Dropping: Some ceiling-mounted objects can be dropped on targets to kill them via the use of a bomb; others cannot. Sometimes it is a large light fixture. Sometimes it is a chandelier. Sometimes (oddly enough) it is a piano suspended on a rope. That last one I only learned by reading a FAQ.

Food, Drinks, Poison, Sedative: “The kitchen SEEMS like its full of useful stuff, but it ain’t. You can’t poison any of the food, and nobody ever comes in here.”

-‘Guybrush Threepwood’

Some drinks can be usefully poisoned, filled with sedative or aphrodesiac, or picked up and moved around. Others cannot. Sometimes people will freak out if they see you messing with certain drinks. Other times, they don’t care. Some characters drink certain drinks sometimes, others don’t. There is no way to tell who will drink what or when. You cannot tell who certain waiters are going to deliver certain drinks to.

The same goes for food. Some food can be poisoned and placed in certain places, where it will then be eaten by certain characters. But this is inconsistent. There is no way to tell where you can put which food, or who will eat it.

Consistent Panty-Sniffing: Some panties can be covered with ether to put nearby panty-sniffing FBI agents to sleep. Most, unfortunately, cannot, and you can never tell who is going to sniff what panties or when.

I also want points for legitimately getting the phrase “consistent panty-sniffing” into a game design article.

Pyro Rigging: Some flame devices (barbecues or pyrotechnics) can be made to explode by rigging them with lighter fluid. Other times, you need to open the pressure on a nearby valve to cause an explosion. Other times, you need to alter the timing on the pyrotechnics display to get a performer killed. Nobody tells you any of these things. You are left to wander around until you find a menu option like ‘rig barbecue’ pop up on the context menu.

Fuse Boxes: Some fuseboxes will summon a guard when broken. Sometimes the guard will come after a delay. Others require some other condition to be met. Most seem to be connected to television sets, but there is nothing about the fuse boxes to indicate this.

Climbing and Leaping: Sometimes you can leap between balconies, sometimes you cannot. Sometimes you can shimmy along a ledge on the outside of a wall, sometimes you cannot. Sometimes you can climb a ledge, sometimes you cannot. Sometimes you can scoot through a window, sometimes you cannot. Some vertical surfaces can be climbed, some cannot.

Who is Strippable: It’s impossible to tell whose clothes it is possible or useful to take. Example: the photographer in the porn mansion level is dressed like most normal guys, whose clothes you cannot take. The fact that his clothes are almost essential to finishing this level is invisible.

Fluffy and Bloody: You don’t know whether you will be able to move the dog’s carcass until you try (you can’t).

Hidden Information

Not all of the inscrutability of the game is derived directly from the domain of simulation problem. There is lots of other information which is necessary to play properly, but can only be learned via trial and error.

Searching Rules: In many areas, guards will search you for weapons before you can enter a certain area. One problem here is that it is often impossible to tell whether this is going to happen or not. You need to approach the guard and find out whether he will search you by simple trial and error.

Guard Sensitivity: Some guards will open fire on you if you look at them funny. Others will ignore almost anything. At certain distances, guards will ignore you holding a gun or dragging a body. If you get closer they open fire. No indication as to how far you need to be to be unseen.

Ratings Sensitivity: Some events will lower your final ratings and others will not. The game does not make it clear exactly who counts as witnesses in this game, or how quiet you need to be to get the top ‘Silent Assassin’ ranking.

CCTV Fields of View: It’s impossible to tell exactly where you will be recorded on CCTV until you have already been recorded.

Body Finding Probability: In many areas, there is no way to tell whether a body is safe from being found or not. The common and convenient body-hiding boxes help, but it still often possible to permanently hide a body without using one. You can never quite tell whether the body is safe, though, until you leave it for a long time, or it is found and you fail the mission because of it.

Detonator Conspicuousness: Sometimes, holding or pressing the bomb detonator will freak people out. Other times, they don’t seem to mind. No indication as to why.

Sniper Location Secrecy: Sniping a target from certain locations will get you found out. Other locations seem to be safe.

Ceiling Lethality: Some glass ceilings can be shot out, raining glass down and killing victims below. Others cannot, even in the same level.

The Solutions

These problems are too deep for a quick fix. Solving them would require a deep redesign of most aspects of the game mechanics. I believe it would be possible to execute Blood Money without ambiguous mechanics, but almost every aspect of the game would need to be changed somewhat.I’m wary of being too gung-ho about pushing changes to this game. It would be very easy to ruin Blood Money‘s cinematic quality by making it too obvious and ‘gamey’. I will try to suggest changes that will make the game more approachable and transparent, without sacrificing that awesome cinematic quality that makes Blood Money great.

Tight, Consistent Mechanics

The game would do better if it focused on a limited, consistent set of gameworld interactions. This doesn’t mean the game needs to be simple. Simple actions can be combined into complex strategies.

Were the mechanics 100% consistent, the missions become meaningful puzzles because a player who understands the limited list of available actions should be able to mentally work out a solution to any problem. Trial and error should never be necessary, and no part of the game should be left unnoticed.

Note that many of these possible changes would require a systematic redesign of the missions. These are the possible beginnings of an iterative process.


Food: All cups should be poisonable, and NPCs should always drink out of cups if it is at all reasonable. No bottles should be poisonable, and NPCs should never drink out of bottles. Even better: avoid using bottles altogether.

Unpackaged food should always be poisonable, and should always be eaten by any reasonable nearby NPC. Packaged food should never be poisonable or interacted with in any way. Even better: avoid using packaged food altogether.

Containers: All briefcases should be useable as containers, and treated the same way by NPCs. No other containers are allowed. Food should not be useable as a container ever.

Dialogue: 47 should have no optional dialogue. If dialogue is absolutely necessary,47 should be approached non-optionally by an NPC. The man is an assassin, not a con artist.

Body Hiding: This actually wasn’t too bad in the original game. The body boxes are a good idea. However, all railings, without exception, should allow bodies to be pushed over them. Also, all bodies, without exception, should be found if 47 leaves them in any vaguely public area. When a body is in a ‘body hiding’ zone where it will never be found, 47 should say something to himself about this and drop the body automatically. You should thereafter not be able to pick up the body.

CCTV: All CCTV cameras should have conspicuous blinking lights. They should all look the same, and they should all make the same noise. In addition, they should all project a faint holographic-like field of view in front of them a short distance to give an idea of where they can see.

Fuse Boxes: Fuse boxes should declare exactly what will happen when they are broken. Diana should also inform you of important fuse boxes and their function when you go near them.

Unique Mechanics Highlights: Any unique mechanic or interaction in a level should be made easier by a visual highlight on the object being interacted with, to show that you can do something with it. See: Bioshock or Call of Duty 4.

Digital Tension Meter: The tension meter should be changed from a mushy analog meter to a digital number of ‘tension marks’ out of 4 or 5. This is similar to GTA’s ‘wanted stars’. Different values give distinctively different behaviours to the guards. The fact that the tension value is digital means that players can learn the behaviour or guards at each level and be able to predict them more coherently. Also, when the tension ratchets up, it feels more significant when it is on a digital scale instead of an odd swinging colored bar. Transitions between tension levels should be caused by well-defined stimuli.

Hints: The game should give as many indirect hints as to the level’s mechanics as possible in posters and NPC dialog. If the mark is going to go to the strip bar, have his friend talk a lot about how much he likes going to the strip bar, or the fact that he is expected there in five minutes.

Diana in Mission: Diana says everything about the mission in the briefing. The problem is that it is far too much information to be retainable in one bunch like that. Instead, she should mete out information throughout the mission as necessary, calling 47 on his cell phone, or speaking through an earpiece. For example, in the casino mission, send 47 in with only very basic information. Then, when the scientist arrives, have Diana call to tell him so, and tell him what the scientist is going to do, and that he is carrying the diamonds. Tell the room number and the number of guards. Same with the sheik and neo-nazi.

Any time there is a non-standard mechanic, Diana should inform you of it clearly and tell you what you can do.

Better Feedback

The game should give the player more information as to what to expect should they perform a certain action. Here are some examples of experimental mechanics which the game designers could try out as starting points in this direction. I don’t necessarily think all these would definitely work well, but they would be good starting points for the iterative process:Searcher Indicators: NPCs who intend to search you for weapons should always be paired with conspicuous searching equipment like handheld scanners, and walk-through metal detectors. Scanners and metal detectors should blink in a distinctive and totally consistent color and pattern, and make a consistent subtle ambient noise. This equipment should be standardized throughout the game so that it can be easily recognized once learned.

Conspicuousness Warning Icons: Icons which appear beside options in your context-sensitive menu, indicating how conspicuous they are. A happy face indicates that you can do the action and nobody will mind. A worried face indicates that you will arouse suspicion. A red, angry face indicates that you will be regarded as immediately hostile. These icons could also appear beside items in your inventory listing. Dental floss: happy face. Tricked out double pistols with laser sights, scopes, and extended clips: very red and angry face.

Conspicuousness Monitor: A HUD indicator which indicates what reaction your current activities would cause were you seen. Happy face indicates that you aren’t doing anything suspicious. Worried face indicates that guards will become agitated and shoo you out of the area, or civilians may yell in surprise. Angry red face indicates that guards will shoot you if they see you. These should obviously be the same symbols as the conspicousness warning icons in the menu and inventory.

Map Allowability Overlay: Divide the map visually into different sectors according to where you are allowed to be in your current outfit, and indicate the areas with the conspicuousness icons and colors – the same icons and colors as the menu, inventory, and HUD monitor.

Higher difficulty settings could remove these aids, thus satisfying players who want to play the game raw.

The Benefits: Consistency Allows Dynamics

Consistent-izing this game would allow the design to add a lot of randomless to the level events while remaining playable. This would create massive replayability. It also becomes much more lifelike, because now the game is about reacting dynamically to events, not perfectly executing a sequence of pre-determined actions. Furthermore, it won’t turn off new players nearly as much with its incredible opacity and brutal learning curve.If some of these changes work in some form, Blood Money could go from being great but unapproachable, to great and easy to pick up – and very profitable.