Monday, July 17, 2006



Last post, I mentioned "keys".

When a player plays a game, they usually become fond of some aspects of the game, and strongly dislike other aspects. Liking to dislike a given aspect is also common.

Your favorite character, your favorite spell type, your favorite bad guy: these are all things that players tend to feel strongly about.

This is generally much, much clearer in tabletop adventures, because tabletop games grow and change in response to the players. So when the players really like something, the GM tends to make it more important.

For example, in my pseudo-Star-Wars game of long ago, certain people had preferences for certain NPCs. They would treat these NPCs radically differently - even though they were no more or less cool or trustworthy. Which NPC often varied from player to player.

I use these characters to pull players around. If you want a player to get emotional, you make his favorite character get emotional. If you want a player to get worried, you threaten his favorite character.

I quickly learned the power and fragility of this tactic, and it evolved into something I call "key control".

Key Creation

The basic idea of key control is that every moment a player plays a game, he is investing in the game. Key control is the art (or science) of pushing all that investment into one or two things, rather than letting it simply bleed out into the game world at large. Usually characters are the best option, but I've done it with space ships, kingdoms, concepts, prophecies, and even game mechanics. I suggest sticking to characters.

Done correctly, every hour of gameplay strengthens the emotional bond the player has to this key. Every ounce of investment a key has can be used to pull the player deeper into the game which, in turn, creates a more plentiful investment for the next hour.

How do you do it correctly?

It's simple: the key has to have strong, flawed, independent effects, and the key has to evolve in response to the player.

What are "strong" effects? Functionally, the key has to be at least 75% as strong as a player when it comes to affecting the environment. The key has to be useful (or threatening) enough to be noticed.

This doesn't have to be in the same arena as the player's power: if the player is a warrior, you could give him a seer as a key, or a thief. Neither is 75% as strong as a warrior in battle, but the GM arranges the universe such that their powers come in handy in other situations.

Generally, however, you will find that keys which act in the same domain as the player are going to be the most effective at drawing emotional investment. My NPC with the ability to bombard planets from orbit and blow up space ships was most popular with the player who liked shooting naval cannons, whereas my pillar of light NPC was most popular with the pillar of light PC.

Also, there is no maximum power cap: 75% is a minimum, but the key can be a zillion times stronger than the player. However, the further they get from the player's power level, the harder it gets to humanize them.

That's the second part. "Flawed". The keys have to have weaknesses either to be exploited or defended. This weakness can be anything at all. I've used weakness in combat, moral struggles, terminal illnesses, critical judgement failures... the only requirement is that the key's weakness be reliable. It has to be something that the player learns exists and then can learn to deal with. It gets the player involved, allowing him to act with effect.

Oh, the weakness can slowly change. But "slowly" is the operative word. That's going to come up again, later, but for now let's move on to "independence".

The key might be on the player's side (or not), but it has to be independent. Thinking and moving for itself. This keeps the player from taking it for granted - a key is not always 100% available, and often chips in (or against) when not called for, to show independence.

Don't fear making the key change sides, or refuse to do something. A balky key is fine (and a powerful tool). A significanly unreliable or irritating result from a key will make the player invest pure and unbridled hate, and not the good kind.

So don't make the key so independent that the player can't harnass them. The key needs to react to the player somewhat, and their actions should be logically sensical. Independent, not random.

That's the last part. The key has to invest in the player. There has to be feedback: the player does something, the key does something, and that changes the relationship between the two. The key is a changing entity, at least to the player's eyes, and the player has to believe he is shaping its progression.

Oh, and, durrrr... the key has to be around to be invested in. That doesn't mean it has to be with the player, but the player has to be exposed to things which are related to the key. A city the key burned down, if the player knows it was the key, is just as effective as having a key in the player's face, giggling.

Managing Keys

Okay, so that's the basic idea of a key: you create something at least 75% as strong as a player, but with weaknesses. It is an independent creature, but reacts to and remembers the player. It changes over time, slowly. (This is also a good way to insure it remains applicable in the scenario the player is playing.)

You're building a key!

If you try this on a tabletop, it will probably just click. You'll find the player nearly desperate to spend more time with the key, trying to enhance it and strengthen the investment feedback cycle.

You can allow the player to turn the key into a black hole and fall into it, but this usually results in a jaded endgame: the truth of the matter is that a simulation only goes so deep. Eventually, you run out of interesting things for the key to do, because as they get more invested, the feedback cycle demands ever-larger changes. (This also takes up more GM time, often not a good idea.)

What I usually do is keep the players at some distance. Although the keys react, they never get too close. Then you just use multiple keys.

This allows multiple players to invest in the same keys, and what the keys lack in individual power they make up for in dexterity - having three tokens allows you to do some moves that are impossible with only one. Like setting them against each other.

Players are naturally proprietary, so allowing multiple players to heavily invest in a character is asking for trouble. I find this is when it is good to use something that isn't a character, such as a concept or a nation.

This may not be the best way, and some GMs are comfortable running heavy romance plots (which is often what key obsessions try to turn into). It's up to you to moderate investment to a level both you and the players are comfortable with.

Key Application

Okay, so you've got your keys. What are you going to do with them? All this power... let's put it to use to torture and reward our players.

Keys have to be used delicately. Not because the emotional investment is delicate, but because the universe is delicate. If you do something too fast, too hard, the player will disconnect from the universe emotionally, and you'll get none of the benefit you have worked so hard to provide.

You can use a key for nearly anything: a key can pull people into a plot in a thousand different ways. Getting threatened and volunteering are the two most common, but be creative. A key can also be used to jump-start another key.

No matter what you are doing, there are two rules:

1) Careful with the change in control. Control is a drug. Giving the player more control over a key is a reward like stuffing a lab rat with cocaine. Taking control away kicks him in the teeth with withdrawal symptoms.

If you plan on reducing the amount of control a player has over a key, you have to broadcast it way in advance, or it'll be serious trouble. (Also, the control decrease should adapt to the player's actions: it's part of the key, and that's one of the first rules of creating a key.)

Also, allowing too much control will overdose the player, and the key will stop being a key. Remember: independent.

This means that if the player has much control over a character, then kidnapping that character will risk breaking the player's immersion. Much better to kidnap, say, the key's father, and then let the key poke the player into a rescue.

This has the doubly good side effect of keeping the key in effect. If you kidnap a key, you need to go to pains to make sure the player hears all about the key's suffering, preferably with appeals by the key directly to the player. You need the player to know that the key exists and is relying on him.

You may find that you fall into a "give control to create an investment spike, then leech control slowly away over the next few sessions, then give control" pattern. I don't see any reason this is bad - I've used it myself.

2) Keep the Key Present

Keys only pull when the player can feel them directly. Telling him his favorite NPC was kidnapped will only access a tiny portion of the investment. Having his favorite NPC beg him to help over the airwaves will do significantly better.

Don't be afraid to really, really stretch to keep the player feeling the NPC's presence. Use dreams, discarded luggage, footprints, favorite books, other NPCs gossip, pictures/video/audio - anything.

Just keep the key close. Every hour in which a player doesn't hear something directly from or about the key reduces its strength by half for purposes of leading the player.

3) Threatening a Key

One of the two most common methods of using a key is the threat. The other is the poking - a key says, "please, let's do this!" That's straight forward, but the threat is not.

Threatening a key is delicate but extremely effective. You have to follow the two first rules: no vanishing keys, no sudden changing control levels.

This means that you can poison a key, but it should be a poison that slowly weakens them over a long period of time, not one that instantly puts them under. You can knock them down in combat (always fun), but there should be several rounds of desperate struggle, first.

Never just "do something" to a key. Always lead up to it with a tiny version of that thing, or at least the idea that that thing is coming.

The exception is a highly independent key which goes off on his or her own. That's part of the key's unreliability, not something you as the GM are doing just to kick the player into motion. (Okay, it is, but it isn't perceived as such.)

Whew. That's most of it. One laaaaaast thing.

Keys in Computer Games

Keys in tabletops are easy: a GM can adapt a key to any given player or situation without much difficulty.

Keys in a computer game?

There's several problems with that.

First, AI is extremely poor. Poor enough that most players can't bring themselves to believe a character is worth investing in. You can get around this by making it absolutely clear that the characters are just plain idiots. People can love stupid people.

You can also get around it by improving the AI - always a toughie. Or you can give much of the control over to the player. Which should ring warning bells and leads us to:

The second problem is that most games give the players a huge amount of control over the NPCs. While this can work, it makes the players hypersensitive to changes in their control or the effectiveness of the character. Which means you have to be ten times as careful when you are using your keys.

The third problem is that adaptive characters are extremely difficult. The keys need to be a feedback loop, but a game character is hard pressed to smile on command, let alone evolve meaningfully in response to the player's action.

You can get around these issues by using non-characters as your keys - concepts and space ships are very common substitutions and I endorse them.

But people are probably the most powerful keys. Non-character keys are primarily useful for getting several players to all invest in the same thing.

So... it's a trade off. If you decide to use character keys in a video game, be very careful about it.


As a reward for reading this far, here's a cartoon. Textual Harassment liked the starlet, so here's another picture of her. (Although I think he is more fond of full-body shots...)

"And there's our starlet, looking sensual and fierce in her Vera Wang 'Arrow Through the Head'."

"Like a tiger, how lovely."

"Yes, and that arrowhead is made of a single, perfect ruby."

"That's what we've come to expect, of course."


Craig Perko said...

Oh, forgot to mention: reducing the power of a character is the same as reducing control over a character.

Textual Harassment said...

LOL everybody knows I'm a fool for the cartoon chicks.

On this keys thing, it seems like many of these tricks are just as applicable to linear narrative. The only new trick is adapting the key to the player's preference (which is a great idea. You don't want to try to pull the player around with a character he hates).

Even stuff like the player controlling the NPC, and the NPC having independence can be "faked" through scripted events. So I don't think this is limited to interactive stories.

Chill said...

I actually stumbled upon this idea while running my Mage campaign. It is still very fondly remember by all who participated and I think that my use of 'keys' (although I didn't call them that) was part of that. One of the funnest parts of the campaign was when the party essentially created a key, a magical shotgun called the Blockbuster. :D. There were more of course, and some I used better than others. In fact, I think the only thing I would do differently would to recognize when keys are not interesting the player enough.

Oh, and I assume that your PAC engine design was an attempt to apply this idea to games, yes?

Patrick Dugan said...

One difference I infer between a PAC engine and the Rocket Hearts engine proposed here earlier is that PAC passively reacts to the discourse based on a system of metrics, and RH just sort of throws out some content (which can get fairly emergent in the interactions) and with good design and a lot of balancing, lets the player find their favorite keys through play.

For Fianna the concept of paganism versus a homogenous christian religion, of Irish nationalism and the underlying rebellion, are a fairly strong set of keys that effect the same participation and give the context for the whole thing. A player can suit themselves to the religion thing, or politics or both. The tricky part comes with the characters being the keys, their
AI should be passable in the context of what the conceptual keys portent, and players should be able to find their favorite characters through play based on who they're playing and how. Rather than have a top-down GM orchestrate things based on the percieved favorites, I think putting out enough information on the system should let the player induce their own perogatives based on some wild causality. It'll be HARD to balance, but thats the cost of becoming a legend. :)

Or maybe I'm full of shit, Craig?

Craig Perko said...

Textual: Yes, you can do it with a linear narrative. The difficulty comes in predicting what kind of player you'll have, and making sure there's a suitable key for them.

But it's something that happens in many, say, movies. Star Wars has three big keys: Darth, Obi-Wan, and R2D2. Each of these appeals to a different type of audience member at different times. :D

Chill: Yeah, I'm not saying this is new, I'm just saying that I've never seen it clearly described.

PAC was an attempt to do this along with a somewhat related concept.

Patrick: PAC doesn't have to be a passive system. The difference between PAC and Rocket Hearts is that Rocket Hearts is easier to do because it has no global metrics.

You can make anything a key - religion and nationalism work. But that means they have to be independent, adaptive powers which the player believes he is shaping/changing his benefit from.

That's not easy with things like nationalism.

Hey, the word verify wants me to type "chaos". Neat.

Chill said...

This also explains why people get so attached to their soldiers in X-Com (or whatever squad based strategy game you prefer). They've almost unwittingly capture most of the qualities that you need to be a good key.

Craig Perko said...

Yes, their growth and the randomization of their hits was independent enough to serve that purpose. :)