In the last part, we looked at what went wrong with the PC version of Ghost Master, an innovative haunting game developed by now disbanded studio Sick Puppies, and designed by my team at International Hobo. In this final part, we look at what went right in the game design, and how the elements of the design came together.
Originally, I hadn't intended for this to be a two part piece, but popular demand (read: one comment by Corvus) convinced me that the Post Mortem would be more complete, and more balanced, if I also took the time to comment on what worked in the game, and the positive side of the game design process for this project. Since it was cathartic to write about why the game was a commercial failure, perhaps it will leave me feeling more positive overall if I close on more positive notes.
Before the Sick Puppies studio even existed, Gregg came to me with a concept for a haunting game and asked me to come up with an initial design. Gregg and I had previously worked together at Perfect Entertainment on the acclaimed and occasionally best selling Discworld adventure games, and we had developed a good working relationship. Gregg knew I had a gift for the rapid generation and development of design documentation, and that's what he needed at this point. The concept documents were what he used to secure the funding that enabled him to build Sick Puppies, I believe.
Gregg's initial concept had a strong vision, which is always an asset in a game. He wanted to build a haunting sim, with its inspiration being drawn from 'reality TV'. In essence, Gregg had watched how people enjoyed the 'goldfish bowl' entertainment of shows which place people in a confined environment and then basically jerk them around, and presumably thought it would be even more fun to do the same with virtual characters. Since it was going to be a virtual environment, haunting was a great way to go, as it brought with it the opportunity for artists to show off their graphical prowess with special effects.
The centre of the gameplay was clear right from the beginning: that scaring people is fun. This is why children love to hide and jump out on people. In fact, even to this day I occasionally jump out and scare my wife for no other reason than ilinx and paidia - spontaneous anarchic fun, if you will.
The initial design for Ghost Master was put together in a few days, and much of the design work I did then remained in the final game. The core concepts were as follows:
- The player controls a team of ghosts (Haunters) which they position to scare people (Mortals)
- Mortals have different degrees of Belief - those with high Belief are easy to scare, whilst those with low Belief require convincing of the supernatural to 'batter down their defences'
- Mortals have a certain degree of Willpower which determines the degree of Terror required to make them leave the haunting (Flee). The higher the Willpower, the harder to make them Flee.
- The amount of Terror generated in mortals determines the degree of resources the player has available. An invisible Fear Factor percentage (the average Terror across all mortals) is converted into Plasm, which is used to power the Haunter's powers.
That, in essence, is the core of the gameplay: scare Mortals to get Plasm to power Haunters to scare Mortals.
For the most part, all of this made it into the game, and the core gameplay mechanics work just about perfectly.
Haunters & Fetters
During the design phase, there were several issues that needed developing. Chief among these were the nature and types of ghosts available, and the powers that they would use. From a very early stage, Gregg and I were in agreement that we would aim to minimise the amount of micromanagement - creating a game that anyone could pick up and get some fun out of, but that supported more advanced play for the interested player.
What we wanted to do was create a situation in which the player's key decision is where to put the ghosts, and how much to power them up. As a result, much of the focus of the way the Haunters work rests in the nature of the individual Haunter families. For example, a Gremlin can be bound to any electrical item, and has powers that affect their operation (for example, powers to cause them to malfunction, or to electrocute any nearby Mortal). A Banshee, on the other hand, can be bound to any Thoroughfare (a path or hallway) and has powers based around Noise and the weather.
I developed a set of Haunter families which covered pretty much all the classic ghosts, and that would be more or less intuitive as to how they should be used. But, not leaving anything to chance, the game has a system such that when you are considering where to place (Bind) a haunter, the places they can be bound (Fetters) light up clearly. This makes placing the ghosts a snap - you don't need to remember anything about the ghosts to use them, but at the same time the more you play, the more you learn how to use the ghosts.
The transparency of this system is one of the trivial strengths of the game in my opinion... New players can experiment, expert players learn the best ways to use particular families of haunters. You learn the important things by what might be considered "osmotic play".
The Binding and Fetter systems are in fact integral to the core gameplay: much of the skill in playing the game is in determining the best arrangements of your Haunters, and there is much subtlety to be found. For instance, Mortals can move around the world quite rapidly - especially if you scare them so much that they start running! As a result, you can't be trying to react to their every move - you need to be smarter than that.
The best results result from working out places that Mortals have to pass through - front and back doors to houses are perfect places to bind Haunters, for instance. Hauntings strategies that excel are built upon a strong team dynamic - a Haunter outside to scare people back in doors, and a concentration of Haunters in larger rooms where you can pack in more Mortals for maximum effect.
There is also plenty of room for the player to form their own approach... Weaker Haunters such as Spooks, which are like the classic Scooby Doo ghosts, and Hordes, which are basically swarms of creepy crawlies and the like, can be bound just about anywhere - whereas stronger Haunters such as Spectres and Phantoms have very specific Fetter requirements - such as a Murder site. So you have a choice: build your strategy to funnel Mortals to your heavy-hitting ghosts, or use weaker ghosts and use maneuvrability to scare them away. The former strategy is the best in terms of completion time, the latter is the easiest approach.
The game design is rife with situations like this, in which the player has a choice in how to play. In almost all cases, the choice is between the easiest option - which is the lowest scoring - and the fastest but hardest option - which is the highest scoring.
Powers & Plasm
The next part of how the game works are the powers that the Haunters use. I was mindful of the need for balance in this part, but we also wanted a lot of variety. The powers were set up into classes, and those classes assigned to different Haunter familes. This was mostly an 'under the hood' issue - the player wasn't expected to know any of the complexity of this system, but each Haunter family has a feel, and the assignment of power classes was based upon player expectations (Banshees wail, Gremlins screw up machinery) and game balance.
We expressly wanted to avoid a situation common in many strategy games (and bear in mind, we did not see this as a strategy game) whereby the player has to learn a lot of custom abilities or unit types in order to be able to play the game. We needed a way for the player to face a simple decision in terms of how powers were used, but that in turn would produce a lot of diversity and play choices.
The way I structured this was based upon a series of power bands. All the powers in the game are divided into ten bands - numbered 1 to 10. Haunters have different power levels which reflect the highest band power they can use - only the big guns can go up to Band 10, and as already mentioned, these are balanced by having more restrictive fetter conditions. The player's in game resource, Plasm, is used to power Haunters up to different bands - so the decision the player faces is simply which band to put a ghost into. The Haunter itself then decides which power to use on its own (all ghosts are quasi-autonomous agents - a rare thing in games).
The key to this was simple rule based AI... In general, the Haunters use whichever powers they can in any situation - although there is more complexity to this, and the player can even indirectly manage their team by giving rule based orders - such as "only use such-and-such a power" or "don't use your powers unless such-and-such a thing happens". There's also a touch of fairy dust in the Haunter AI (which I designed) - each ghost has a certain unique identity, from the uncontrollably anarchic to the staunchly reliable, although they tend to get more house trained the more you use them.
The beauty of the power band system was, like the Fetter system, you can play the game with very little knowledge of what's going on under the hood - but once you start getting good at the game, you can start really making the powers work for you. And the powers are lots of fun to play with - the artists and programmers excelled at creating visuals for powers which, on paper, I had only intended to be functional most of the time. Gregg and the team made my dry list of perfectly balanced powers come alive, and I'm certain fans appreciate their attention to visual and aural detail more than my cost-benefit analysis of the powers which ensure that every band X power is balanced to every other band X power. (I had a meta-document which detailed the text book strength of a power at each band - although certain custom side effects had to be costed in on a case-by-case basis).
The power band system makes using Haunters simplicity itself. To reiterate the core mechanic: scare Mortals to gain Plasm to power Haunters to scare Mortals. So, a novice player simply throws their ghosts into the field, turns them up to the highest band they can and then lets the fear generated turn into more Plasm - which allows them to put more ghosts into the field, or ramp up the powers of the ones already out.
Now if another designer had worked on the game, they probably would have wanted to use a standard resource mechanic: Fear produced units of Plasm which are 'spent' when powers are used. But I didn't want it to work this way. I didn't want haunting to be an economic challenge - I wanted it to be a tactical-strategic challenge. This isn't shopkeeping, it's fearmongery! So Plasm isn't 'used' when a ghost uses a power. Instead, Plasm is like a wattage, and the ghosts simply share the available power supply... Use a big ghost at full power, or lots of small ghosts at full power, or a big ghost at half power and some smaller ghosts... Lots of choices, once again.
I personally think this mechanic worked perfectly. At any point in the haunting, you have a certain amount of Plasm, and you use this to power up a certain number of ghosts - but using the ghosts doesn't cost Plasm at all. If you discover that your current approach isn't working, no problem, just bench your haunters and try a different configuration - all you've lost is a small amount of time.
The player can lose Plasm, however, just in a round-about fashion. Because Plasm is proportional to how scared the Mortals in the haunting have become, if you don't keep them scared, your Plasm starts to fall. This creates the internal pressure that keeps the game pace working - you need to keep scaring the mortals to keep your Plasm supply up.
In practice, your Plasm supply is only truly vulnerable until you get the first few Mortals to flee - once a couple of the little wusses Flee, you have a reserve of Plasm to fall back upon and don't have to worry about running out. This is because a Fled mortal contributes to the overal Fear Factor by counting as a "50% scared mortal"... in effect, if you scare someone sufficiently to make them run away, a residue of their fear remains behind. This is part of the Plasm mechanics, which were the most mathematical of all the design problems I had to solve while working on the game.
The Mathematics of Fear
The original formula I devised for Plasm supply was that we would average the ratio of Terror to Willpower across all Mortals and call this Fear Factor. So, if there are two Mortals, and both are one quarter of the way to Fleeing (Terror = 1/4 Willpower), the Fear Factor is 25%. Then, the amount of Plasm was defined as 20 x Fear Factor - so from 0 to 2000. At 25% it would be 500. The costs of the Plasm Bands were then gradiated on this basis, using an exponential relationship, so that it gets dramatically more Plasm expensive to run the big gun powers.
There were a number of problems. Firstly, at the start of the haunting, mortals aren't afraid, so there's no Plasm, so you can't haunt. This was easy to solve. We assume that there's an unease in the air just from your very presence, and give a default Terror value to all the mortals (this is actually modified by Belief, so the more vulnerable Mortals begin more scared than the sceptics). This didn't wholly solve the problem, however, as when the game was up and running (more than a year after initial design) we had the problem that this Terror started to bleed away rapidly.
This was part and parcel of the way the Terror for Mortals work: keep scaring them with powers, and they remain scared. But if you ease up the pressure on them, leave them alone, and, worse, let them get together with other Mortals who aren't afraid, then they start to calm down. At the start of the haunting, since you have done nothing, they lost their Terror quite rapidly - not good. This was easily fixed by adding a "Ghostly Presence" rule - the initial Terror cannot reduce until the player uses their first power, so the player has all the time they need to scope out their surroundings and decide how to haunt.
The second problem was more thorny, but I thankfully spotted it even before the game was up and working: even the lowest Power Band requires a certain amount of Plasm, and it was possible for the Fear Factor to be so low the player could not use any of their ghosts. This was unacceptible - a local minima in the Plasm mechanics which would make the game completely broken. The easiest fix was to call this state a loss - but I didn't like that idea.
Instead, I broke out my copy of MathCAD and punched in the numbers for the Plasm equation I wanted. In essence, with high Fear Factor values, I wanted a value equivalent to 20 x Fear Factor (otherwise I'd have to regear all the Plasm mechanics I'd already done), but with low Fear Factor values, I wanted values that were high enough to power the lowest bands of Powers at all times. A little tinkering gave me the formula I wanted (which- in a wholly inappropriate fashion - is listed in the back of the Prima guide for the game). I'm really pleased with this piece of design, as I headed off a number of different problems before we even had the game working - the quintessence of that part of the designer's role in which they act as an anticipatory problem solver. The result: all of the power costs remain how they should be, but at no point does the Plasm supply fall so low that you cannot afford to get some Haunters out into the field and using powers. Perfect.
The result of these mechanics, coupled with the Fled effect described above, is that hauntings naturally have three basic phases:
- Exploration: the player has all the time they need to consider what to do, thanks to the "Ghostly Presence"
- First Strike: there is pressure on the player to scare fast and hard in the early haunting in order to get the most vulnerable Mortals to Flee, thus securing their Plasm supply for the rest of the haunting
- Elimination: finally, once the player has secured their Plasm supplies by picking off the more vulnerable haunters, the player has a certain amount of Plasm with which to tackle the more tricky Mortals.
This is roughly how it was always intended to function. Note that I purposely wanted to avoid the player being in a state of constant tension as many other games endeavour to do - I wanted the player to have a part of most hauntings that they could actively enjoy torturing mortals with their powers. What's the point of being the scarer if you don't get to enjoy it once in a while?
Annoying People & How to Dispose of Them
One element of the design I cannot take credit for is the Mortal AI. I originally envisioned the game coming together on a much smaller time scale, so I devised a very simple Affinity based AI system that would be adequate to the task of moving the Mortals around. Gregg had bigger ambitions, however, and as the project went on the programming team eventually had to build a complex set of AI mechanisms to run the Mortal behaviour. This subject is a whole topic in itself, and not something I can justly catalogue, except to say how glad I am that we had such great programmers on the team.
However, at the design side, the Mortals needed to be created in such a way as to ensure that some of them would be hard to get rid of, but that getting rid of them would still be a flexible process. To ensure this, a multi-pronged approach was used. As already mentioned, each Mortal is defined with their own values for Belief and Willpower, but they also have two personal Fears (things that scare them more than anything else), and a Madness stat.
Just as a low Belief score reduces the effect of Terror on a mortal (they simply refuse to believe that what is going on isn't a hoax), a high Insanity also reduces the effect of Terror - because mortals which are crazy just aren't scared as easily. This might seem counter intuitive - why would I want to cause Madness if it's going to protect a Mortal from Terror? However, there are two elements of the Madness mechanics which make it work in the context of the game.
Firstly, if you raise a Mortals Madness above their Willpower, they go permanently Insane which counts as Fled. And whereas Terror can fall with time, Madness is permanent. And as if that wasn't enough, Insane mortals go around acting all crazy and scaring other Mortals - which has the bonus effect of ensuring they don't calm down, because there's some fruit loop lunatic running around the haunted house with them!
The net result is that hard to scare Mortals can be disposed of in numerous different ways:
- Convince & Scare: use truly supernatural effects (telekinetic powers, bleeding walls etc.) to build Belief, and then terrify them.
- Uncover Their Fears: use Haunters with psychic powers to uncover the hidden Fears of a mortal, and then use ghosts which will tap into their weakness. For instance, a mortal with a Subconscious Fear of Blood suffers 3 x Terror from Blood effects - so such a Mortal can be made to Flee easily with the right choice of ghost.
- Drive Them Mad: use ghosts with Madness powers to drive them crazy... As it happens, most Mortals with high Belief are quite vulnerable to Madness effects, so this can be particular effective in certain cases.
The flexibilty of this set up works brilliantly, with one minor problem: Madness powers were intended to be an advanced option for expert players to use, but sadly Madness effects are collectively not as fast as Belief/Terror (for various reasons too complex to go into here). Since expert players want to set score records by getting fast completion times, Madness isn't a very effective option for advance play.
That said, the Madness route does work perfectly, and I have to say, it's still a hell of a lot of fun to play with. If we'd got a sequel, the slight problem with their use could easily have been fixed (by shortening the recharge times of insanity powers, for instance) and as it stood, it would take someone with my intimate knowledge of the mechanics to realise anything was wrong at all, so it would be churlish of me to complain.
A more serious problem is that the player is taught everything in one tutorial right at the start of the game. But there's too much to take in at once. You're just getting to grips with Haunters and Fetters, Plasm Bands and the whole notion of Belief, Terror and Willpower when the tutorial level ends. The game was too innovative to get the player up to speed in one single step, and myself and the tutorial programmer were more interested in trying to be inventive in how we presented the tutorial (which is dynamic) than in checking that we were teaching the player what they needed to know when they needed to know it.
We should have split the tutorial material into bite sized chunks, spread across the whole game (as planned for the sequel), so that, for instance, we could teach people that the weather powers don't create weather states they tinker with the very components of weather - temperature, precipitation and wind speed. Want a storm? Combine rain with multiple powers raising the wind speed. Drop the temperature to turn it into a blizzard. There's much to uncover in the weather subsystem alone. The game never quite teaches the player these facts, which are important to completing The Blair Wisp Project - although thankfully the game world is so dynamic, and the puzzles are always multi-threaded, so most players muddled through all the same.
Some small mention deserves to be made of the team selection mechanics. I always wanted the game to feel a bit like Mission: Impossible with ghosts at the start of each haunting, and so we have a selection screen which is like a set of dossiers on the Haunters you have recruited. This gave us all sorts of problems.
The player starts with a starting team, but then gains new Haunters by finding them in particular hauntings. These 'local ghosts' have not been laid to rest - they have a problem which causes them to be stuck at a particular Fetter. The player has to solve a (usually) simply problem to lay them to rest, after which they can join your team.
This mechanic caused problems because a few of the hauntings required specific powers and abilities to solve, and we couldn't guarantee at any point in the game which Haunters the player would have on their team. The solution we adopted was to give the player some extra Haunters at the start of each Act of the game to ensure that they would always have a minimum set of powers. Not exactly elegant, but it worked.
A related problem was that only a player with some experience would be able to choose which ghosts they wanted for a particular haunting, so we needed a 'Suggestion' button to give the player a default team. But this default team couldn't be fixed, because we didn't know at any point in the game which team the player would have - every player would have different ghosts.
The solution for this turned out to be specifying the recommended team in a set of 'streams' such that if the player had the best Haunter for the job, it would be recommended, otherwise there would be a string of replacement suggestions (up to four deep). With some considerable logical problem solving, we were able to come up with sets of ghosts for the Recommended Team rubrics which fulfilled the requirements of always giving the player at least some minimum chance of completing a given haunting.
It works quite nicely in the final analysis: the Recommended Team is always competent, but there is always plenty of potential for the player to improve upon this recommendation. Which, in point of fact, is half the fun of the team selection screen anyway.
The Trouble with Time Travel
Sadly, there was one problem with the Haunter teams that could not be solved by game design alone: when you replay earlier hauntings, you can't take in all the ghosts. The reason for this was technical: many of the ghosts require custom solutions on a level by level basis. The Dragoon, our Headless Horseman, for instance, required pathing work for each and every level he could appear in. Because of this, Gregg had to insist that there were limitations on which ghosts you could take back.
The solution I went with was that you can only take in those ghosts which you could realistically have at that point in the game. This wasn't perhaps the best choice. Not only was it torturously difficult for me to calculate those restrictions (because of the trivially branched structure of the game i.e. we don't know the order the player is completing available hauntings), but it's disappointing not to be able to take ghosts back into all the previous hauntings and let rip.
In retrospect, the better solution might have been to let the programming-complex haunters be banned from entering the Time Gate (the in-game justification for replaying hauntings is that you travel back in time), but give the player free reign with the ones that pose no problems. It wouldn't have made much difference.
Similarly, Gregg insisted that you couldn't gain unfair advantage from using the Time Gate to replay earlier hauntings, so the benefits you can get are capped such that the benefit you get in the game's reward resource (Gold Plasm) is limited: you can go back and improve your score, but you only get the difference between what you scored last time and your new, better score.
This was a complete waste of effort, frankly, as it was more complicated to implement but didn't give any advantage. The only thing the player can use Gold Plasm for is buying powers for your Haunters, and as it happens unless you're a really, really good player, you will struggle to get enough Gold Plasm to power up all your ghosts. There was just no reason to deny the player the freedom to replay old hauntings to get more Gold Plasm, but I wasn't confident enough that this would be the case in advance, otherwise I could have convinced Gregg. (He was always won over by intelligent argument - but he wisely needed to hear the arguments first).
The clearest evidence of the fact that this restriction was misguided is that within weeks of the game's release there was a trainer released by the enterprising modders giving players infinite Gold Plasm... a sure sign that we worked too hard to stop the player getting something they kind of needed. And if we hadn't restricted the player from getting Gold Plasm from replaying hauntings, it would actually have extended the play window (by rewarding the player for replaying any haunting they enjoyed), so in the end this hurt us. Although, in real terms, the amount it hurt us was trivial.
Ghost Toys, Ghost Games
On the whole, I am extremely pleased with the core mechanics I designed for Ghost Master which are interesting, unusual and completely original. There's no game quite like it, which turned out to be a blessing and a curse, but in terms of game design achievement, I'm very proud of the game mechanics as they are.
There's much more I could talk about, but on the whole, the above is about the most accesible account I can give of the key design issues addressed in the project without getting into seriously minute detail. Already, it is probably quite difficult to get to grips with this account without having had some experience of the game.
What the game design certainly seemed to fullfil is a happy balance between a toy and a game: players with low game literacy can, with a little application, sit down and play around with the game quite happily, getting simple joy and wonder from just watching the beautiful animations and artwork the development team created, and laughing at the people as they run around scared witless in response to what you do with your team of ghosts. This is coupled with a complex set of mechanics which do support advanced play - although because of the short length of the game, and the problems already mentioned about the Triple Pumpkins, very few players got to really appreciate this side of the mechanics.
To have created a game which plays as both a toy and a game is something I consider a great achievement, and I more regret not including a true sandbox mode than not solving the appeal of the game to the typical Type 1 Conqueror player (about whom I knew nothing at the time, since my audience model was still very basic). When I recieved emails from parents saying their young kids love playing with the ghosts, but wondering if there was way to disable the Plasm mechanic so they could just play with the ghosts, I realised for the first time that we hadn't actually gone as far as we could have done with the toy side of the game. As ever, if we'd got a sequel, we would certainly have fixed this.
I believe Ghost Master can be enormous fun to play; it has its share of problems but then what game does not? That there are players out there who somehow overcame the lack of marketing and found the game, and grew to fall in love with it, however briefly, is some compensation for the commercial failure of the game.
After all, I am not involved in marketing - I can't be held accountable for the market failure of a game which, in broad strokes at least, met all its design goals, and offered up a little something, new, different and entertaining in an industry which for the most part seldom ventures far from familiar territory.
To everyone who played and enjoyed Ghost Master, you have my thanks - you made the whole process worthwhile.