Recommend
3 
 Thumb up
 Hide
15 Posts

BoardGameGeek» Forums » Board Game Design » Board Game Design

Subject: Automated foes who interact rss

Your Tags: Add tags
Popular Tags: [View All]
Jonathan "Gorno" Fashena
United States
Westchester
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
This is redo/repackage of my (terribly underappreciated! ) General Gaming thread on the apparent lack of interacting foes in co-op games, but more forward-looking: http://www.boardgamegeek.com/thread/401738

And related to this Tower Defense thread:
http://www.boardgamegeek.com/thread/383113

I'm musing about a co-operative game along the lines of Return of the Heroes, Shadows over Camelot, or Arkham Horror where the players must overcome various foes to win. It's also applicable to a solo game (one-player co-op ) or even a competitive game where there are independent factions the players don't control. As in the much simpler case of tower defense, different baddies have different automated behaviors, rules (either definite or partially randomized) that determine their actions and reactions to various events.

The players use their actions to try to manipulate these behavior patterns for the team's benefit, in particular, by exploiting them to play the different enemies against each other instead of the heroes, who could never hope to defeat them all directly. For example, the players in a fantasy game might exploit "Rat Men retreat to their lair when driven off" by driving them in the path of "Wargs (who) chase their opponents any distance," thereby clearing the path of the Wargs without a fight. A literary example from the Lord of the Rings is when (if I remember correctly) Merry and Pipin set the orcs of Sauron and Saruman to fighting, facilitating their escape.

Even more interesting, these simple behavior patterns, manipulated by the players, could interact to produce complex, inobvious, and surprising "emergent behavior" that the players would have to speculate about and discover for themselves, giving them a complicated virtual opponent to "outsmart" (or be "outsmarted" by, if their plans backfire!) which it seems to me is ultimately why games hook gamers, psychologically speaking. If there were multiple types of enemies, each with a few simple behaviors, interesting multi-faction interactions and cascades could be imagined and precipitated by the players.

As a counter-example, in Arkham Horror, say, you seem to have to seek out each threat, then defeat it: my impression is that they don't react except to defend themselves, and never interact (between what I'm calling "factions"). One might instead imagine the Cult of Nyarlothoptep being tricked into sinking the schemes of the Deep Ones, or the Vikings in SOC into picking off the Picts!

Gorno
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
James Casey
United Kingdom
Caversham
Berkshire
flag msg tools
Where'd he go?
Avatar
mbmbmbmbmb
Its an interesting idea, though I will predict it will be very difficult to pull off successfully. Lots of computer games have tried to harness this "emergent behaviour", but it rarely ends up being more fun than a more "scripted" approach. With a board game everything needs to be done by the players rather than a computer, so it needs to be simpler.

The main pitfalls I can foresee:
- that the player ends up sitting there while the game "plays itself"
- that the game becomes overly complicated to produce the required behaviour
- turns become overly long as you have to calculate what the "AI" is doing

I would simplify it as much as you can - pare down the idea to one specific example, with just two or three factions, make a prototype, and see where it goes. Maybe use an existing game (Descent?) as a core so you test the viability of your idea first.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Christian Marcussen
Denmark
Odense C
flag msg tools
designer
Avatar
mbmbmbmbmb
Not much to add except this is one of the areas that I have long wanted to work on. It's a kind goal of mine to make a game like Doom or Descent, but without the need for a "game master"... So all in all - I wish you the best. It is no doubt and interesting and challenging goal to set one self!
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Phil
Germany
Bremen
Bremen
flag msg tools
badge
Avatar
mbmbmbmbmb
Reminds me a bit of Escape from Atlantis (if this is the right one). While every player is on its own each turn every player may control one monster which can be used to
a) move it away from your units
b) move it towards and perhaps devouring other players' units
c) move it towards other monsters to devour them

Very simple but fun and perhaps could be improved. But when those monsters fight too much with themselves (not in this game but a new one) I fear that my own units/character would have less importance and I would be better of playing the GM of any dungeon crawl.

EDIT: One possibility:

Every player controls two sides: His army and one monster army.
His points in the end is something like "defeated other monster armies multiplied by survived own army". So you have to rely on the help of other players to keep your army alive but can use your monster army to defeat the other players army and their monsters.


Or even something like moving your characters through a city in which Godzilla and King Kong are fighting over the white bitch. So the player wins who gets her to safety (every player has his own safehouse) while the monsters are drawn towards her. So with moving her you can control the monsters but when they meet they start to struggle which might change the layout of the map due to collapsing buildings.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Nigel Buckle
United Kingdom
Thornton Heath
Croydon
flag msg tools
designer
NSKN are publishing my next game :)
badge
Omega Centauri Published in 2014
Avatar
mbmbmbmbmb
This is hard to pull off. One possibility is to have the players take on 2 roles in the game, playing both the good guys and the bad guys.

Middle-earth did this with the hazard/resource part of the game. You play resources on your turn hazards in your opponents turn.

Byzantium had you play both sides of the conflict

And The Peloponnesian War, 431-404 BC if you played solo - you switched sides once you started defeating the non-player.

So you could have a game where you co-op play one side and individually play the other (through cards or secret actions etc), then you have human brains playing both sides.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Wright
United Kingdom
Stockton-on-tees
Teesside
flag msg tools
badge
Avatar
mbmbmbmbmb
I have looked at programming opponents for a dungeon game I am creating to remove the need for a GM or to direct the behaviour of the GM in operating the monsters.
For monsters or an opponent to be different the GM / or the system has to have rules that dictate this behaviour.

The model I used was to split each monster into a number of steps representing unhurt, hurt wounded etc. Thus depending on the wound / moral level they would act accordingly.

Examples being
- Pack of dogs - will charge when outnumber foe. Will retreat when wounded
- Troll - will always charge and not retreat.

Using verbs such as charge, retreat, stand etc will give behaviour.
You can also name the target, such as attacks closest unarmed or attacks the boss etc.

However the problem is that as the behaviour becomes more prescribed it becomes easier to predict. I am unconcinved either way whether this is good or not.Maybe you want players to get toasted on the first encounter and when they understand the behaviour they can more easily overcome the foe, using their predictable behaviour as a weakness.

The final problem with programmed behaviour is that the players can see what it is going to do even if they have not encountered it before.

The way around this is I cam eup with was to use a behaviour deck, 52 cards that listed actions and targets on a table. The behaviour was drawn and the monster card would reference a behaviour, thus although the players saw this behaviour they would not initially know if it was typical or not of their behaviour.

2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Nigel Buckle
United Kingdom
Thornton Heath
Croydon
flag msg tools
designer
NSKN are publishing my next game :)
badge
Omega Centauri Published in 2014
Avatar
mbmbmbmbmb
I like the idea of a behaviour deck - it's a bit like the monster cards in DungeonQuest where you had to decide to run, wait or fight.

If you had each monster with a behaviour deck (so an agressive monster will have more attack closest character type cards) it could work - but players will learn the decks, so you could adjust by having a subset each game, making them a bit less predictable.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jonathan "Gorno" Fashena
United States
Westchester
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
hellium wrote:
I have looked at programming opponents for a dungeon game I am creating to remove the need for a GM or to direct the behaviour of the GM in operating the monsters.
For monsters or an opponent to be different the GM / or the system has to have rules that dictate this behaviour.

The model I used was to split each monster into a number of steps representing unhurt, hurt wounded etc. Thus depending on the wound / moral level they would act accordingly.

Examples being
- Pack of dogs - will charge when outnumber foe. Will retreat when wounded
- Troll - will always charge and not retreat.

Using verbs such as charge, retreat, stand etc will give behaviour.
You can also name the target, such as attacks closest unarmed or attacks the boss etc.
Just what I was thinking! Moved to an open landscape, instead of confined to rooms in a dungeon, interactions become (more) possible, as well as unexpected chains of events.

(I'm not actually working on this, merely musing about the concept.)

Gorno
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris J Davis
United Kingdom
London
flag msg tools
designer
badge
Overtext pending moderation...
Avatar
mbmbmbmbmb
Maybe this effect could be achieved by a deck of cards. There are a few similar methods I can see (let's use examples with robots, ninjas and pirates):

1) Each card relates only to one particular faction and describes in detail the actions members of that faction perform when it is drawn. So we could have a card that says "Ninjas activate. Any ninjas within 4 spaces of a pirate move to the closest pirate and attack. All other ninjas move 6 spaces towards the nearest robot."

2) Each card lists in slightly less detail the actions every faction performs. So we could have a card that says "All pirates move 2 spaces north. Then all ninjas move 2 spaces towards the nearest pirate. Then all robots attack the strongest ninja or pirate within range."

3) Each card lists seperate action for each faction, but only particular factions are activated based on some other factor in the game. You could simply have the factions on rotation, so that with the first card drawn the ninja part is used, with the second card drawn the pirate part is used, and with the third card drawn the robot part is used. Or if the game involved some kind of role selection mechanic then a secondary function of the role cards could be to indicate which factions activate and the order they activate in.

(I like no.3 the most, by the way).

Depending on what instructions you put on the cards, you could have a simple form of AI. For example, robot cards would include more instructions that moved them towards attacking from a safe distance, ninjas are more likely to move towards pirates, pirates more likely to move towards robots, etc...

Just an idea...
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jonathan "Gorno" Fashena
United States
Westchester
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
waferthinninja wrote:
Its an interesting idea, though I will predict it will be very difficult to pull off successfully. Lots of computer games have tried to harness this "emergent behaviour", but it rarely ends up being more fun than a more "scripted" approach. With a board game everything needs to be done by the players rather than a computer, so it needs to be simpler.

The main pitfalls I can foresee:
- that the player ends up sitting there while the game "plays itself"
- that the game becomes overly complicated to produce the required behaviour
- turns become overly long as you have to calculate what the "AI" is doing

I would simplify it as much as you can - pare down the idea to one specific example, with just two or three factions, make a prototype, and see where it goes. Maybe use an existing game (Descent?) as a core so you test the viability of your idea first.
All great points/suggestions. Another pair of potential pitfalls is that (did someone already say this?), players would start out clueless (and always lose) until they had a number of plays under their belts, then "solve" the game, which is no fun either. A random starting arrangement/mix of opponents would "mix things up" to fight this. One could also have different starting setups/scenerios with difficulty levels to keep giving the players new challenges.

Still another problem is that one doesn't always have the same players, so newer players wouldn't know what they were doing and would just get dragged along with the team. Yet another pitfall (which must be common to all co-ops) is "who decides?" I don't know how this is dealt with in existing co-ops, but perhaps one could have occasional steps where only the current player sees new info (say, draws a card) and must decide how to react without revealing it to the others.

It would be nifty to see some of these elements as variants to established co-ops.

Gorno
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jonathan "Gorno" Fashena
United States
Westchester
New York
flag msg tools
badge
Avatar
mbmbmbmbmb
hellium wrote:
The way around this is I cam eup with was to use a behaviour deck, 52 cards that listed actions and targets on a table. The behaviour was drawn and the monster card would reference a behaviour, thus although the players saw this behaviour they would not initially know if it was typical or not of their behaviour.
Puts me in mind of "blind biasing:" http://www.boardgamegeek.com/thread/362454

For multiple factions, that's a lot of cards. The simplest (but bland) random approach would be a table, with a few different results at the end to differentiate each monster from a generic base-line. Another idea would be die biasing, for example, having an extra die that is thrown with a standard set and might override/modify them... one could even have the set of dice consist of a die colored or identified with each different type of monster (like a bunch of different Dragon Dice), so you always roll the same batch, but the current monster's die is the master die, and modifies (or dictates how to interpret) the others in a way that varies with the monster...

Gorno
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
castiglione
United States
Sunnyvale
California
flag msg tools
badge
Avatar
I've toyed around with this when I was trying to create a solo dungeoncrawl that had the feel of an "old-school" D&D game where you could set one group of monsters against another (and the monsters were tough enough so that if you attempted to defeat them through direct combat, you'd probably die).

Basically, I envisioned groups of monsters moving randomly throughout a dungeon with each group of monsters having monsters that they were enemies with (always attacked), allies with (always helped) and were neutral towards (pretty much ignored unless they were attacked or their allies were attacked) and with the relationships being "interlocking" and non-symmetrical.

However, I got disenchanted with the whole idea for a bunch of reasons. The top reason was that monsters moved either randomly or until they got within line-of-sight of an enemy (which they would then rush towards). That wasn't very appealing to me. The only "trickery" on the player's part would be to get the monsters to chase them and then run towards a group of enemy monsters. I tried to think of another way to "motivate" monsters to move but couldn't think of one.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Wright
United Kingdom
Stockton-on-tees
Teesside
flag msg tools
badge
Avatar
mbmbmbmbmb
johngorno wrote:
hellium wrote:
The way around this is I cam eup with was to use a behaviour deck, 52 cards that listed actions and targets on a table. The behaviour was drawn and the monster card would reference a behaviour, thus although the players saw this behaviour they would not initially know if it was typical or not of their behaviour.
Puts me in mind of "blind biasing:" http://www.boardgamegeek.com/thread/362454

For multiple factions, that's a lot of cards. The simplest (but bland) random approach would be a table, with a few different results at the end to differentiate each monster from a generic base-line. Another idea would be die biasing, for example, having an extra die that is thrown with a standard set and might override/modify them... one could even have the set of dice consist of a die colored or identified with each different type of monster (like a bunch of different Dragon Dice), so you always roll the same batch, but the current monster's die is the master die, and modifies (or dictates how to interpret) the others in a way that varies with the monster...

Gorno


I like the dice idea providing a slant on the basic behaviour, I wonder what it could be in terms of outcomes.
In my design at one stage I had a scale of behaviour with regards to the combat area. From Flee through retreat, stand, advance, charge and bezerk. Thus the die could alter that up or down, not necessarily now but when the behaviour changes (say from events or wounds etc) then the bias dice comes in. For example orcs with their chatic randomness could go from stand to bezerk charge in one step. Instead of building up, I thought that as the wound status of the creature changed the revised behaviour would take over. So if you wind a troll up and wound it it may well go on a rampage rather than advance in a hulk like fashion.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jim Cote
United States
Maine
flag msg tools
badge
Avatar
mbmbmbmbmb
I've had ideas where players each contribute secretly to the "behavior of the game", for example, by playing cards face down to various piles, which get resolved at various times. No one knows who put what in each pile, or necessarily how many. Now that I type this, it kinda reminds me of the Future/Current Events in Through the Ages. However, instead of simple events, the cards could be things like: influence in various board elements, votes for office, votes for changes of rules, movement/attacks of enemy units, etc.

Fearsome Floors has some of this in a much smaller way.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
James Hutchings
Australia
Unspecified
Unspecified
flag msg tools
mbmbmbmbmb
castiglione wrote:
However, I got disenchanted with the whole idea for a bunch of reasons. The top reason was that monsters moved either randomly or until they got within line-of-sight of an enemy (which they would then rush towards). That wasn't very appealing to me. The only "trickery" on the player's part would be to get the monsters to chase them and then run towards a group of enemy monsters. I tried to think of another way to "motivate" monsters to move but couldn't think of one.


As well as allies and enemies, creatures could also have motivations. For example they could have greater or lesser interest in gold, magic items, taking prisoners etc. They could also have greater or lesser reliability (how easy is it to get them to change sides).
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.