Archived

This topic is now archived and is closed to further replies.

Diodor

Strategy&Law / AI is Game Design

Recommended Posts

This is about single player strategy games. I play Total Annihilations these days, and, as I go forth attacking and destroying everything on the map that isn''t painted in the same color as my commander, it makes me wonder, is it the way it has to be? Must there only be as few factions as to allow the player to win through raw force? Must the only goal be total domination? Must the only law be that there is no law? If the player''s strength is limited, so that he cannot ever destroy everything on the map (I''m thinking something along the lines of a maximum limit of 1% of the worlds total force - or less), is it possible to create a game where the player still has the same chances to win, to dominate, to be king of the hill etc.? The solution I think is proper design of the AI. Usually the AI in a strategy game is not thought of as a part of the game design process - the AI only comes after the game is designed - and its purpose is to play as well as possible - to win. But what if the AI isn''t written to behave as if it were a player, but is instead thought of as a means to entertain the player, as a part of the game design? What if there are many factions controlled by the AI, but many of them are cowardly, and unwilling to fight? What if the AI controlled factions obey certain laws - laws of allegiance and alliance, allowing or forbidding violence - laws which are in no way different from the other game design laws? Obviously a game is played differently depending on what the AI does. Can a whole new game be created based on interactions between AI players?

Share this post


Link to post
Share on other sites
It is certainly possible through diplomacy (team up with other AI). The old (but good) board game Diplomacy has you controlling a European country (1 of 7) in 1900 Europe. You need to capture the whole board but only start with three pieces (same as the other 6 players). Only by teaming up can you defeat enemies before betraying your allies and taking over the map.

Dan Marchant
Obscure Productions
Game Development & Design consultant

Share this post


Link to post
Share on other sites
I like games that involve some diplomacy with the coalition.

Another thing to consider in domination games is that a large empire can have civil wars erupt within. Some internal diplomacy is important.

Share this post


Link to post
Share on other sites
quote:
Must there only be as few factions as to allow the player to win through raw force? Must the only goal be total domination? Must the only law be that there is no law?


play final fantasy tactics for this part of the post

each battle has to follow one or many law decide by divine instance according to land or day and daytime, like no sword allowed or no healing allowing or most random law like no use of attack which begin with the letter A or else the sinner will be punish in some way according to the gravity of the action

player then has to play in more subtle way, and there is mssion which is not a kill them all

why not generalized this idea to common game? (not only strategy)
would be fun (if punition is a malus that can be double edge, like the poison item in bomberman 1 which let the player infect other player by contact)

>>>>>>>>>>>>>>>
be good
be evil
but do it WELL
>>>>>>>>>>>>>>>

Share this post


Link to post
Share on other sites
Just thinking along Civ lines, this could get very frustrating, and would have to be handled veeery caaarefully.

I think the setting ties in as well. In the thread on losing I posted a concept that works on the concept of limited-scope battles feeding into a larger conflict. In that particular case, it works because the player is fed differently than in the typicaly total-war scenario, so the AI doesn''t have much to do with it.

In terms of multiple-faction combat, I think it might be productive to take a WWI/II approach, wherein the player is a single nation in a coalition. Of course, if you include diplomacy such coalitions can develop naturally. I find, however, that in games such as the Civ line it''s just combat abstracted, since in the end diplomacy tends to be another means to total domination.

ld

Share this post


Link to post
Share on other sites
The one thing I''d like to see in a game with the option to ally with AIs is an "Allied Victory" option - the biggest problem with diplomatic options in games like Risk or Diplomacy is that both sides know that ultimately they will have to turn on each other, so either alliances have to have well-specified termination clauses or there is no room for honour in such games. Even when two players are notionally allied, the game still pushes them to try and undermine each other.

Civilization III gets part of the way there - there are several ways to win without ever being openly at war with a previously allied civilization - on the other hand, there is still only ever one winner.

A game with potentially friendly AI and allied victories would definitely interest me.

As a side-note, the way I would define allied victory in a last man standing situation is that there is no pair of surviving factions which are not allied (strongly enough). For less clear cut victory conditions (eg dominance rather than extermination), an alliance would be any set of factions where every pair of factions is (at least strongly enough) allied, and any alliance which met the required victory conditions would be victorious, though this does allow the situation with 10 factions where everyone is allied to everyone else except for one pair of enemies who are allied with everyone except each other - meaning there are two alliances, each of which contain 90% of the surviving factions, and so everyone ends up victorious. The simplest work-around is to put conditions on the range of degrees of alliance a given faction can have with the various members of any given alliance - so with 3 factions, A,B,C, if A and B are allies, and C is allied to A, it has to be at least neutral towards B, and similarly if C is an enemy of A, it can''t be better than neutral towards B. If C were to ally with B whilst still an enemy of A, A and B would automatically cease being allies and become neutral towards each other instead. Of course, if you have more than three diplomatic states, then more varied ranges are possible.

Share this post


Link to post
Share on other sites
quote:

Original post by Obscure

It is certainly possible through diplomacy (team up with other AI). The old (but good) board game Diplomacy has you controlling a European country (1 of 7) in 1900 Europe. You need to capture the whole board but only start with three pieces (same as the other 6 players). Only by teaming up can you defeat enemies before betraying your allies and taking over the map.



Diplomacy is an exceptional game, but it is also very far from what I am thinking here. In Diplomacy, all players play to win - this is the core of the game, and all players must be very clever (I don't think Diplomacy can be properly played against computer players). I believe Diplomacy has the potential to revolutionize multiplayer games - all I hope with this thread is to describe a different kind of single-player computer strategy game.

My idea is to have many computer players who don't want to win - their actions are in fact designed to create a certain gameplay. They may even be dumb as rocks. I propose thinking of the AI of the computer players in the same manner one thinks of the upgrade path of a certain unit - or the building options of a building. Think of computer players as some sort of resources, pawns to use.

quote:

Original post by liquiddark

Just thinking along Civ lines, this could get very frustrating, and would have to be handled veeery caaarefully.
...
I find, however, that in games such as the Civ line it's just combat abstracted, since in the end diplomacy tends to be another means to total domination.



Yes, Civ has a lot diplomacy, and it is true that it is very hard if not impossible to play Civ without carefully considering the complex relationships with the other civilisations. However, the AI in Civ is again designed to win and the game only has a handful of factions. Like Diplomacy and Risk, Civ is a game where a few simetrical factions fight for world domination.

What I propose is to make the players goals different from the goals of the AI players (this makes the game specifically geared towards singleplay). Because only the player aims to dominate, his goal is a lot easier to achieve, and this allows an increase of the difficulty level by adding a lot more opponents. The gameplay focus changes from kill,krash&destroy (not necessarily removed) to maneuvering the different computer players.



[edited by - Diodor on October 18, 2003 4:14:50 AM]

Share this post


Link to post
Share on other sites
Interesting...

I had an idea for a space empire style game while ago(ala Masters Of Orion, but probably fewer menus than the latest incarnation) which sounds something like what you are suggesting.

Rather than focusing purely on building a political empire, you could focus on building any kind of empire you liked. So for example, you prefer set up some kind of trade empire, and eventually evolve into some kind of giant galaxy spanning corporation. Rather than forcibly take over planets, you'd kind of insidiously gain influence over them, establishing monopolies over their trade etc. They'd still be in control of the starships, but they'd probably find themselves relying on you to pay for them...

And that's just one area of focus. You could take the more traditional route of establishing political dominance, you could focus on technology etc. Or you could go for a more general strategy, and focus on multiple things.

The galaxy would be populated by a number of serious rivals, some of which would be focussing on the same things as you, others would focus on completely different things. More than one of the big players could even gain joint victory without even being allied - it is possible for one faction to build a trade empire without affecting another nation's ability to build a political empire. Furthermore, the galaxy would also be populated with a number of smaller rivals, ranging from smallish, non-expansionist empires of a few systems to individual systems. These guys would not necessarily be trying to compete with you, but are available as a potential resource - to be exploited. Too small to build any worthwhile starships? Buy them from someone elses shipyards. Not enough of a particular resource? Buy it from one of the independents.


[edited by - Sandman on October 18, 2003 5:28:13 AM]

Share this post


Link to post
Share on other sites
quote:
Original post by Diodor
What I propose is to make the players goals different from the goals of the AI players (this makes the game specifically geared towards singleplay). Because only the player aims to dominate, his goal is a lot easier to achieve, and this allows an increase of the difficulty level by adding a lot more opponents. The gameplay focus changes from kill,krash&destroy (not necessarily removed) to maneuvering the different computer players.


How would you address the need for limits on the player''s power? You talked earlier about wanting to limit them to a small fraction of the total mass of force on the board, so I assume there is some mechanism in your head for doing so, but if other players are waiting Kuwaits to the player''s Iraq, shouldn''t they be allowed to invade? And then wouldn''t you need an honest-to-dog superpower or two to stop them?

Hoping that''s more stimulating than it is pedantic,
ld

Share this post


Link to post
Share on other sites
quote:

Original post by liquiddark

How would you address the need for limits on the player's power? You talked earlier about wanting to limit them to a small fraction of the total mass of force on the board, so I assume there is some mechanism in your head for doing so, but if other players are waiting Kuwaits to the player's Iraq, shouldn't they be allowed to invade? And then wouldn't you need an honest-to-dog superpower or two to stop them?



There may be plenty of methods to limit power by direct game rules. For instance, corruption limits unlimited empire growth in Civ (or makes it useless).

In the context of this thread however, I'll note that it is perfectly possible to impose limits by designing the AI of the computer players. For instance, the AI players may respond by joining arms to destroy any one player that grows above everyone else by invading neighbours (pretty much what happened to Iraq). So the player won't win by invading every Kuwait around him, but perhaps may do so by inciting an Iraq to conquests and then fighting it down to maintain the peace.

[edited by - Diodor on October 23, 2003 8:35:00 PM]

Share this post


Link to post
Share on other sites
quote:
Original post by Diodor
For instance, the AI players may respond by joining arms to destroy any one player that grows above everyone else by invading neighbours (pretty much what happened to Iraq). So the player won''t win by invading every Kuwait around him, but perhaps may do so by inciting an Iraq to conquests and then fighting it down to maintain the peace.

I would note at this point that it is argued, often convincingly, that Germany and its allies could well have won the Second Great War were it not for the intervention of one of the great superpowers (the US).

Moreover, both sides formed coalitions. How do you circumvent the player forming similar alliances? Or do you? Is that one of the driving elements of the diplomatic model?

ld

Share this post


Link to post
Share on other sites
quote:

Original post by liquiddark

How would you address the need for limits on the player's power? You talked earlier about wanting to limit them to a small fraction of the total mass of force on the board, so I assume there is some mechanism in your head for doing so, but if other players are waiting Kuwaits to the player's Iraq, shouldn't they be allowed to invade? And then wouldn't you need an honest-to-dog superpower or two to stop them?

Moreover, both sides formed coalitions. How do you circumvent the player forming similar alliances? Or do you? Is that one of the driving elements of the diplomatic model?



There may be plenty of practical methods to limit power through direct game rules. For instance, corruption limits empire growth in Civ (or makes it useless). It may also be possible to limit power through AI design - perhaps the players overgrowth triggers a common reaction by all the AI players.

However I don't have a very good practical example for this thread. Just the idea that computer player behaviour has not been exploited as a source of gameplay as much as it deserved, and there's much thinking to do in this direction and many surprises to be discovered.

Actually I do have an example: in my Pax Solaris the hardest games are 3-player games (2 AI players + 1 human) - on the most difficult level, the player starts the game with lesser units than each of the AI players. The (very simple) AI tries to attack in any direction where it outnumbers its enemies. The player must therefore retreat to shorten his front lines and sway the AI players from attacking him by having higher forces on the borders. However this retreat must be controlled in such a way that the two remaining AI players are thereafter balanced and will fight each other without either of them winning (if one AI player grows over-large, the game is as good as lost). Also, when the player decides to attack, he should rather try to attack the biggest player - and when one of the AI players starts to grow too much, the player must try to fight it down - and therefore try to help the other AI player. Now Pax doesn't have any kind of diplomacy implemented, but I've felt more of an ally of these AI players than of any civilisation in MOO or CivIII.


[edited by - Diodor on October 28, 2003 3:27:31 AM]

Share this post


Link to post
Share on other sites
This is just my opinion, but I think you may be too close to that design, Diodor. I''ve not yet played the full version, but I never got that feeling at all from playing the game, and I think maybe in this case you''re seeing what you created as an author rather than through the eyes of a normal player. If the identification of the method is not explicit, you force the player to create their own meaning, and I don''t think it''s as clear-cut as all that whether they will take the same one as you intend.

I guess what I''m trying to say is this: be careful how much you try to make the player see what you see. The flip side of the games-as-art coin is that perception shapes the experience as much as anything. If you have a specific perspective that must be shared to enjoy a game, that limits what players can think in regards to the game. What effect will that have on your audience?

ld

Share this post


Link to post
Share on other sites
quote:

Original post by liquiddark

This is just my opinion, but I think you may be too close to that design, Diodor. I've not yet played the full version, but I never got that feeling at all from playing the game, and I think maybe in this case you're seeing what you created as an author rather than through the eyes of a normal player. If the identification of the method is not explicit, you force the player to create their own meaning, and I don't think it's as clear-cut as all that whether they will take the same one as you intend.

I guess what I'm trying to say is this: be careful how much you try to make the player see what you see. The flip side of the games-as-art coin is that perception shapes the experience as much as anything. If you have a specific perspective that must be shared to enjoy a game, that limits what players can think in regards to the game. What effect will that have on your audience?



It's not as much a matter of feeling and perception as one of strategy required to win levels on expert difficulty. Also, it's not something I designed, it is more of something that emerged. And it doesn't happen on very many levels - and not always - and only on very high difficulty.

I expect all players will come to learn the way the AI reacts and use it to their advantage, more or less. I don't know whether all players will find out that winning with White on the "tree" level may require an early attack that will draw Red forces that would have otherwise attacked and destroyed Green (on the other side of the map). But the very clear fact is that if Red destroys Green, White loses very soon afterwards, and that Red overwhelms Green sooner or later if White plays defensively. Seeing Green as a friend and trying to help it by drawing Red forces against oneself isn't that far-fetched anymore.

The point I'm trying to make is that if the player can trigger (in a predictable and repeateble and logical fashion) certain behaviours from the part of an AI player, he plays the game _differently_ It's no longer "what can I do to wipe this enemy out?", it's "what would happen if I do this?" I believe it is possible to design games that rely completely on this (as opposed to Pax - where this is just a side-effect).

Some ideas:
a beehive AI - attack it, and it will spawn forth its armies, destroying everything around
a rabbit AI - it will flee rather than fight as long as it has where to flee.
a herd AI - when attacked, it will launch a charge in the attackers direction - but it will continue the charge for the attacker even if in its path are found other AI players (whom the attacker wanted to destroy in the first place)
a bear mother AI - it will attack anybody who attacks its friendly AI players.

None of these ideas would be even considered when writing the AI for a strategy game. However, the resulting gameplay (such as getting two cubs of a mother bear AI to fight each other) may well be very interesting.

[edited by - Diodor on October 28, 2003 8:53:48 AM]

Share this post


Link to post
Share on other sites
Diodor I definitely had the experience you''re talking about while playing Pax Solaris, getting enemies to face off against each other and all that, and extending the idea further sounds cool. This is starting to sound almost like an a-life game, in fact my first visions when reading the start of this thread were of a god game or a puzzle game, rather than an actual strategy game.

A couple ideas in the spirit of your list...
sworn enemies - e.g. white and black AIs will attack each other on sight until annihilation, ignoring everyone else
kamikaze - opposite of rabbit

That''s all I have for now... back to lurking

Share this post


Link to post
Share on other sites