Negotiation mechanics in games.

Recommended Videos

TheMadDoctorsCat

New member
Apr 2, 2008
1,163
0
0
So anybody noticed that human-to-human interaction in videogames tends to involve automatic weaponry of some description? I have.

I'm looking for a gameplay mechanic that allows for negotiation between a human player and more than one computer-controlled NPC. There are a number of situations that could be covered by this, but let's take a simple one:

I'm designing a game where there are two objectives: to make it to the end of the game and win the biggest prize, and to accumulate as many smaller prizes as possible (these act as the game's "score".) A single human player competes against computerised NPCs who are trying to achieve the same objectives and who play by the same "rules". The game has multiple rounds, and at the end of each one, somebody - either the human player or an NPC - is eliminated. (The elimination mechanic isn't important here.)

-------------------------------------------------------------------------------

So here's a simple example "round".

The rules are this:

- There are five players competing for four prizes in this round.

- Those prizes are (let's say): £5000, a car, a holiday, and a free pass to the next round of the game.

- The players must unanimously agree on which of them gets each prize - which obviously means that, since there's only four prizes, one of them goes away with nothing. (There are other mechanics to compensate for this - for example, the possibility of early or late-game "deals" between players who agree to help each other win in exchange for help not being eliminated.)

- If the players don't agree, nobody gets anything.

The easy part is planning the NPCs' behavior. Each of them has a distinct "personality" - for example, some are more likely to be cooperative and will look to make deals, whereas others are ultra-competitive and will only accept a positive outcome in each individual situation. Some may have alliances with others, and want to give them good stuff; others may dislike them and want them to take nothing. The NPCs can easily behave in definable and predictable ways.

-------------------------------------------------------------------------------

The part I'd like suggestions with is the actual negotiating mechanic. I've considered the following:

1) Turn-based: This would probably be the easiest but also the least elegant solution. The player would spend a lot of time just reading text from randomly-generated NPCs, which is not exactly my idea of a fun game.

2) Real-time dialogue: More fun but less practical. It's very hard to program in a system of dialogue where the options might change according to what others have said, especially if what they're saying is updating in real-time. For example, if NPC #1 proposes a solution, does the human player then need to have an EXTRA dialogue option added on the fly for "I agree / disagree with NPC #1"?

3) The "Populous" approach: taking direct control away from the human player and having them "guide" an NPC character instead. The character's actions represent the player's decisions - for example, have a panel that allows you to match a specific player with a specific prize, and watch your character argue for this position based on his or her specific character traits.

This is definitely the most practical way to go, and the player constantly has the option of changing their mind on the basis of what's being "said". But I'm not sure I want to take this much control AWAY from the human player, and have them just act as a "guide".

-------------------------------------------------------------------------------

So what do you guys think? I'm definitely leaning towards option #3, but I still think there may be a better way to go. Does anybody have any ideas to improve it?
 

dyre

New member
Mar 30, 2011
2,178
0
0
Honestly, this sounds more like a logical puzzle than a simulation of human negotiation, which often involves more subtle manipulation (telling half truths, getting the other guy to like you personally, "spinning" your offer to seem better than it is, etc). Maybe I'm misreading it, but it seems like when you break it down your game is essentially one of those logical question like:

Bob will only accept a car or a holiday. Bob will only allow Mary to win the £5000 prize.
Mary will accept any reward as long as it's within one tier of your reward.
Jon will not allow Bob to win a car or £5000, but he will accept any reward.

etc

Human negotiation is hard to simulate because it's really complex. The best I've seen it done is in Deus Ex Human Revolution in which you're given some data about the character and you have to manipulate him with slightly different approaches to convince him to do something, and even in that case the player really only chooses the tone of the argument, not the argument itself.
 

gargantual

New member
Jul 15, 2013
416
0
0
TheMadDoctorsCat said:
So anybody noticed that human-to-human interaction in videogames tends to involve automatic weaponry of some description? I have.

I'm looking for a gameplay mechanic that allows for negotiation between a human player and more than one computer-controlled NPC. There are a number of situations that could be covered by this, but let's take a simple one:

I'm designing a game where there are two objectives: to make it to the end of the game and win the biggest prize, and to accumulate as many smaller prizes as possible (these act as the game's "score".) A single human player competes against computerised NPCs who are trying to achieve the same objectives and who play by the same "rules". The game has multiple rounds, and at the end of each one, somebody - either the human player or an NPC - is eliminated. (The elimination mechanic isn't important here.)

-------------------------------------------------------------------------------

So here's a simple example "round".

The rules are this:

- There are five players competing for four prizes in this round.

- Those prizes are (let's say): £5000, a car, a holiday, and a free pass to the next round of the game.

- The players must unanimously agree on which of them gets each prize - which obviously means that, since there's only four prizes, one of them goes away with nothing. (There are other mechanics to compensate for this - for example, the possibility of early or late-game "deals" between players who agree to help each other win in exchange for help not being eliminated.)

- If the players don't agree, nobody gets anything.

The easy part is planning the NPCs' behavior. Each of them has a distinct "personality" - for example, some are more likely to be cooperative and will look to make deals, whereas others are ultra-competitive and will only accept a positive outcome in each individual situation. Some may have alliances with others, and want to give them good stuff; others may dislike them and want them to take nothing. The NPCs can easily behave in definable and predictable ways.

-------------------------------------------------------------------------------

The part I'd like suggestions with is the actual negotiating mechanic. I've considered the following:

1) Turn-based: This would probably be the easiest but also the least elegant solution. The player would spend a lot of time just reading text from randomly-generated NPCs, which is not exactly my idea of a fun game.

2) Real-time dialogue: More fun but less practical. It's very hard to program in a system of dialogue where the options might change according to what others have said, especially if what they're saying is updating in real-time. For example, if NPC #1 proposes a solution, does the human player then need to have an EXTRA dialogue option added on the fly for "I agree / disagree with NPC #1"?

3) The "Populous" approach: taking direct control away from the human player and having them "guide" an NPC character instead. The character's actions represent the player's decisions - for example, have a panel that allows you to match a specific player with a specific prize, and watch your character argue for this position based on his or her specific character traits.

This is definitely the most practical way to go, and the player constantly has the option of changing their mind on the basis of what's being "said". But I'm not sure I want to take this much control AWAY from the human player, and have them just act as a "guide".

-------------------------------------------------------------------------------

So what do you guys think? I'm definitely leaning towards option #3, but I still think there may be a better way to go. Does anybody have any ideas to improve it?
Y'know the PAX extra credits guys on youtube once randomly suggested an argument game that had all the system mechanics and move variations of a fighting game. I could see where it makes a lot of sense. Though fighters are less limited, as ANY physical attack can be dodged, reversed, countered, blocked or parried, and some arguments have indisputable, unchallenged absolute answers we're ultimately wrong or right, and that can leave less wiggle room.

Your game would work if the objective is not really to be right but only to sway the other party, like those Ace Attorney games. Its like the quote from the movie 'Training Day'

"It's not what you know. It's what you can prove."

Now the real problem comes in measuring what an npc's suspension of disbelief, intelligence, non verbal communication signs etc are. RPGs can easily do this with back end stats that measure the difficulty of reaching each person, but not everyone succeeds. (EX: Elder Scrolls: Oblivion) it's not easy to represent all those subtle aspects of human interaction in spatial simulation, where all those stats are only factored into the game's coding, and you have to find a way to represent as natural human action. Plus there's figuring out where to map the controls for certain player responses or actions.

Sure games do it, like Mass Effect series and L.A. Noire, but only when the conversation has a narrow focus. I.E. (here's why you shouldn't kill this person) (marriage request) (is this person lying?) (who's right or wrong on a subject). It couldn't be aimless preemptive, relatable talk like the way we do in the real world. There would always have to be a very direct point or consequence to a gameplay conversation under the surface for it to work.
 

TheMadDoctorsCat

New member
Apr 2, 2008
1,163
0
0
gargantual said:
Y'know the PAX extra credits guys on youtube once randomly suggested an argument game that had all the system mechanics and move variations of a fighting game. I could see where it makes a lot of sense. Though fighters are less limited, as ANY physical attack can be dodged, reversed, countered, blocked or parried, and some arguments have indisputable, unchallenged absolute answers we're ultimately wrong or right, and that can leave less wiggle room.

Your game would work if the objective is not really to be right but only to sway the other party, like those Ace Attorney games. Its like the quote from the movie 'Training Day'

"It's not what you know. It's what you can prove."

Now the real problem comes in measuring what an npc's suspension of disbelief, intelligence, non verbal communication signs etc are. RPGs can easily do this with back end stats that measure the difficulty of reaching each person, but not everyone succeeds. (EX: Elder Scrolls: Oblivion) it's not easy to represent all those subtle aspects of human interaction in spatial simulation, where all those stats are only factored into the game's coding, and you have to find a way to represent as natural human action. Plus there's figuring out where to map the controls for certain player responses or actions.

Sure games do it, like Mass Effect series and L.A. Noire, but only when the conversation has a narrow focus. I.E. (here's why you shouldn't kill this person) (marriage request) (is this person lying?) (who's right or wrong on a subject). It couldn't be aimless preemptive, relatable talk like the way we do in the real world. There would always have to be a very direct point or consequence to a gameplay conversation under the surface for it to work.
That's a good answer.

First off, this isn't an RPG. The "negotiation" mechanic would be used for several very specific and focussed situations, like the one I described above. Hence the "focus" part. Each character has some very clearly defined goals, as well as personality characteristics that affect how they're likely to try and obtain those goals. For example, "cooperative vs competitive" might be done on a sliding scale, or even as a binary choice.

I've been programming AIs for a long time now, a couple of decades in fact, so this aspect of it doesn't faze me. It's a complicated system, mathematically speaking (a lot of different variables to account for when predicting how a certain NPC "acts"), but it's the kind of thing I excel at. What I don't excel at is the user-interface stuff. That's what I'm more interested in.

Most games, like the ones you mentioned, have conversations between only two people - a player and an NPC - at a time. There's not a "negotiation" aspect so much as there is making a simple choice or reaching a specific goal. Even when there's a more complex conversation, it tends to follow strict linear paths that the player can affect only in very binary ways - choice A or choice B. That's not what I'm after here.

I'm looking to find a way to simulate the player being a single voice in a crowd. With skill and art, he can manouver his way into getting the desired outcome, but there should BE a skill to it. Hence my reluctance to use a "player-controlled character" to do the dialogue instead of the player choosing dialogue options himself - it takes away from the "skill" factor. You're more likely to succeed if you make the right appeals to the right NPCs at the right time.

Again - I will stress this point - my problem isn't programming the behavior of the NPCs. They're in a very tightly-focussed situation, with set goals that they're aiming for, and a finite number of possible choices they can make - that's easy enough to program. It's the USER-INTERFACE that I'm finding it difficult to master. That's what I'm after here.
 

gargantual

New member
Jul 15, 2013
416
0
0
TheMadDoctorsCat said:
gargantual said:
Y'know the PAX extra credits guys on youtube once randomly suggested an argument game that had all the system mechanics and move variations of a fighting game. I could see where it makes a lot of sense. Though fighters are less limited, as ANY physical attack can be dodged, reversed, countered, blocked or parried, and some arguments have indisputable, unchallenged absolute answers we're ultimately wrong or right, and that can leave less wiggle room.

Your game would work if the objective is not really to be right but only to sway the other party, like those Ace Attorney games. Its like the quote from the movie 'Training Day'

"It's not what you know. It's what you can prove."

Now the real problem comes in measuring what an npc's suspension of disbelief, intelligence, non verbal communication signs etc are. RPGs can easily do this with back end stats that measure the difficulty of reaching each person, but not everyone succeeds. (EX: Elder Scrolls: Oblivion) it's not easy to represent all those subtle aspects of human interaction in spatial simulation, where all those stats are only factored into the game's coding, and you have to find a way to represent as natural human action. Plus there's figuring out where to map the controls for certain player responses or actions.

Sure games do it, like Mass Effect series and L.A. Noire, but only when the conversation has a narrow focus. I.E. (here's why you shouldn't kill this person) (marriage request) (is this person lying?) (who's right or wrong on a subject). It couldn't be aimless preemptive, relatable talk like the way we do in the real world. There would always have to be a very direct point or consequence to a gameplay conversation under the surface for it to work.
That's a good answer.

First off, this isn't an RPG. The "negotiation" mechanic would be used for several very specific and focussed situations, like the one I described above. Hence the "focus" part. Each character has some very clearly defined goals, as well as personality characteristics that affect how they're likely to try and obtain those goals. For example, "cooperative vs competitive" might be done on a sliding scale, or even as a binary choice.

I've been programming AIs for a long time now, a couple of decades in fact, so this aspect of it doesn't faze me. It's a complicated system, mathematically speaking (a lot of different variables to account for when predicting how a certain NPC "acts"), but it's the kind of thing I excel at. What I don't excel at is the user-interface stuff. That's what I'm more interested in.

Most games, like the ones you mentioned, have conversations between only two people - a player and an NPC - at a time. There's not a "negotiation" aspect so much as there is making a simple choice or reaching a specific goal. Even when there's a more complex conversation, it tends to follow strict linear paths that the player can affect only in very binary ways - choice A or choice B. That's not what I'm after here.

I'm looking to find a way to simulate the player being a single voice in a crowd. With skill and art, he can manouver his way into getting the desired outcome, but there should BE a skill to it. Hence my reluctance to use a "player-controlled character" to do the dialogue instead of the player choosing dialogue options himself - it takes away from the "skill" factor. You're more likely to succeed if you make the right appeals to the right NPCs at the right time.

Again - I will stress this point - my problem isn't programming the behavior of the NPCs. They're in a very tightly-focussed situation, with set goals that they're aiming for, and a finite number of possible choices they can make - that's easy enough to program. It's the USER-INTERFACE that I'm finding it difficult to master. That's what I'm after here.
My bad I see what you mean. As a simple consumer with probably weaker math skills, and a more subconscious grasp of logic. I confess you've got me stumped. But I get it. You want the most natural real-time UI you can get while avoiding mechanics (I.E. charts and the dreaded dialogue wheel) that would drag down the gameplay's fluidity. I think thats why I used the term spatial simulation. The fancy word for traditional real time gaming or character control.

It seems like you're charting in fresh waters. Its easy to create a system of responses when the game prompts you, but tricker to map out a user interface for the player to "initiate" different types of unique interaction against quickly changing variables. Look at how long that last sentence was just describing the problem.

There are so many rules. The game gets more fleshed out where the players boundaries and limits are. After all our beloved games are about maintaining the 'illusion of choice'. Option 3: does sound a bit 'Final Fantasy XIII'ish to me where you're managing more than engaging. Option 2: definitely sounds difficult but the most interesting. If I had to analogize it with cartoon animation it's like the trade off between fluid frame-by-frame and static scooby-doo. The former is way more freeing and awesome to witness but tons more work.

Id say it depends on your time and resources or what you're doggedly willing to strive for. Worse case scenario if you have to go with option one or three. The pacing of the interaction with other NPCs could also help the system from feeling too lax. The only example I can think of is Cart-Life. Limited windows of time to make a decision also can make me as a player more tense and engaged.

I wish you the best of luck.