Wednesday, April 19, 2006
Multiplayer vs. Single-Player Design
Designing any game is challenging (if you are doing it right). But designing a game that is playable both with multiple players and by a single player is one of the more difficult design struggles any game developer must face (though it is probably second to designing a major massively-multiplayer game that attempts to break the mold of established MMORPGs and MUDs).
It's not just the technological issues (though these can be challenging). It's trying to create a game that serves the needs of a lone player, a group of players grouping together for a cooperative experience, and the need to provide an interesting battlefield for competitive multiplayer play. Trying to accomodate all these modes results in what I half-jokingly refer to as getting "two games for the price of two," in terms of development cost and effort.
Several friends and I (and my wife) have been playing multiplayer Neverwinter Nights almost since its release. Neverwinter Nights has an incredibly awesome module creation suite, which has resulted in somewhere north of 5,000 player-created adventures appearing on the web. Neverwinter Vault has a form you can fill out which includes information on whether the module is multiplayer or single-player. Because Neverwinter Nights handles multiplayer pretty automatically, the technology is not a limitation. But there are still the design issues of creating a game geared for multiple players versus one designed for one person.
Most fan-created modules are never even tested in multiplayer (I'd guess that less than 25% of Neverwinter Nights players have actually played online for more than just a single exploratory test). But even without testing, they still check the "multiplayer" box because they figure there's no reason why a party of six people with lower-level characters can't play the module just as easily as one.
Some of the design problems we've found in "multiplayer" modules:
* Really long conversations with NPCs - one player ends up talking while the rest of the party wanders around bored.
* Quests that only give rewards to one party member rather than the group.
* Quests that give rewards to the entire party when one member completes it, but the quest can be completable by the entire party (thus we finish the quest SIX times for six times the reward).
* The party ends up with different faction ratings from each other - some of the party members end up being attacked on sight by monsters that are 'friendly' to the rest of the group.
* "Boss Monsters" fall easily to a party of six, whereas their minions (which spawn dynamically based upon the size and strength of the entire group) are actually harder to fight.
* Areas that can only be reached by a player doing something which can only be done ONCE - meaning one player can go and the rest of the party gets left behind.
* Storylines that are focused around a single-character fail to take into consideration the fact that a group is participating now. Characters who never spoke to an NPC are greeted as friends with an assumed history that never actually occured.
* Quests "break" because a player who started the quest on one night are unavailable for a later play session.
There are far more subtle design issues, as well, which are only noticeable over time. In one module that we've currently been playing, there seems to be relatively fixed treasure: The monsters dont' generally drop loot, and the treasures are always the same. Many areas of the module are also only reachable if you have a special magic item that can be purchased at the town at a hefty but seemingly reasonable price.
The trick is that in single-player, the player ends up with way more treasure than he needs, and sells the rest. He can very quickly afford this magic item and visit all these additional areas.
However, in a party of six players, there's very few "excess" items that we can sell (thus there is less liquid currency as a whole!), and what we do sell has to be split six ways. By the time we could afford to buy SIX of these items, we had already explored well over half the module, and now we have to back-track to find these half-dozen areas that we missed (which may be far too underpowered for us by now).
Probably the easiest (I won't say 'best') way to handle both the multiplayer and the single-player experience in the same game is to make the single-player game mimic the multiplayer game as closely as possible, as id Software and Epic did with Quake 3 Arena and Unreal Tournament, respectively. This puts a lot of demands on the AI to behave like other players, but it solves a ton of design problems.
Labels: Game Design
- posted by The Rampant Coyote @ 7:44 AM (Permalink)
Comments:
Links to this post:
<< Home
Some games can afford to build group-only content. The MMO games do it all the time, making areas where you must group up to succeed.
Some games work really well as single player content, where there is much greater control given to the designer.
As you mentioned, the difficult part is when you want it to work for both a single player and a group. Designing for both is much more of a challenge than designing for one player or designing for a group.
Another part of the issue is just plain attention to detail, or the ability to spot potential trouble spots. It's really true that the devil is in the details.
Some games work really well as single player content, where there is much greater control given to the designer.
As you mentioned, the difficult part is when you want it to work for both a single player and a group. Designing for both is much more of a challenge than designing for one player or designing for a group.
Another part of the issue is just plain attention to detail, or the ability to spot potential trouble spots. It's really true that the devil is in the details.
Though I can assure you that I have spent no actual time designing games professionally. I do tend to spend an inordinate amount of time considering the challenges of improving MMO's and various other games to a lesser extent.
My personal opinion on how to take care of designing for both parties and solo is, to be general, improved AI. More specifically dynamic sensing and reaction to various party sizes, levels, and compositions.
Changes in fundamental tactics between fighting one, two, or more players would theoretically allow the experience to be balanced seperately for each situation.
Constrained examples (for the sake of brevity) might be
-using heavy damage singular attacks against one foe to eliminate him quickly.
-When fighting two players disabling one while focusing damage on the other.
-Against more than two using area of effect attacks, especially disabling ones and targeting important members of the party (for instance clerics) would be viable.
-For the issue of player levels being higher than enemies, some enemies might simply run away from you or try to stay out of sight while others (especially if you are invading an area marked as "their territory" might enlist the aid of allies.)
I'm generally thoroughly disappointed with the AI offerings of nearly every MMO currently running. I don't see much hope for the next generation either. My only guesses are that processing/network transmission speeds just aren't fast enough to provide the data transmission needed to maintain more sophisticated AIs, people are trying to cash in on a trend rather than innovate, and/or people are just lazy (aka the design challenges inspire trepidation.)
Like most things I'm sure it is a combination of them all.
My personal opinion on how to take care of designing for both parties and solo is, to be general, improved AI. More specifically dynamic sensing and reaction to various party sizes, levels, and compositions.
Changes in fundamental tactics between fighting one, two, or more players would theoretically allow the experience to be balanced seperately for each situation.
Constrained examples (for the sake of brevity) might be
-using heavy damage singular attacks against one foe to eliminate him quickly.
-When fighting two players disabling one while focusing damage on the other.
-Against more than two using area of effect attacks, especially disabling ones and targeting important members of the party (for instance clerics) would be viable.
-For the issue of player levels being higher than enemies, some enemies might simply run away from you or try to stay out of sight while others (especially if you are invading an area marked as "their territory" might enlist the aid of allies.)
I'm generally thoroughly disappointed with the AI offerings of nearly every MMO currently running. I don't see much hope for the next generation either. My only guesses are that processing/network transmission speeds just aren't fast enough to provide the data transmission needed to maintain more sophisticated AIs, people are trying to cash in on a trend rather than innovate, and/or people are just lazy (aka the design challenges inspire trepidation.)
Like most things I'm sure it is a combination of them all.
# posted by 
: April 21, 2006 5:43 AM
: April 21, 2006 5:43 AM
The purpose of AI in games isn't really to DEFEAT the players. It's easy to create killer AI. I mean, I can play about any computerized chess game and have my butt handed to me even at a low difficulty level. The trick is to create AI that will keep the player ENTERTAINED. Challenge is only part of that equation (but an important part! You are not wrong!)
Creating AI that is "believably stupid" yet challenging is the big trick. What usually happens is that the AI is good in one aspect of the game, but unbelievably mechanical in another, which completely spoils the illusion and allows the players to come up with cheap counters.
The AI in combat flight sims is one example. In most of the better flight sims, the AI at higher difficulties is EXTREMELY good - but also extremely predictable. It's like the Terminator in a cockpit. They are perfect, "by the book" robots - which gives the player a couple of tactics which he can repeat endlessly to defeat them (so long as he has the initiative to bring it to bear).
But the faults you are addressing are very real. I mean, in MMO's (and in other RPGs), the monsters pretty much stand around waiting to be killed, and then fight to the death and only rarely attempt to call reinforcements (I think DDO is the first MMORPG I've seen where the AI will actively ring an alarm to summon more monsters to help). It WOULD be more fun to deal with more interesting, organic tactics.
But it's not so much fun to deal with constantly fleeing creatures, or creatures that are consistently making "hit and run" attacks. While it's good, challenging AI - it's more frustrating than entertaining to the players.
So while I agree that better AI is *one* tool to help balance the multiplayer / single-player equation, it's not the whole answer.
Creating AI that is "believably stupid" yet challenging is the big trick. What usually happens is that the AI is good in one aspect of the game, but unbelievably mechanical in another, which completely spoils the illusion and allows the players to come up with cheap counters.
The AI in combat flight sims is one example. In most of the better flight sims, the AI at higher difficulties is EXTREMELY good - but also extremely predictable. It's like the Terminator in a cockpit. They are perfect, "by the book" robots - which gives the player a couple of tactics which he can repeat endlessly to defeat them (so long as he has the initiative to bring it to bear).
But the faults you are addressing are very real. I mean, in MMO's (and in other RPGs), the monsters pretty much stand around waiting to be killed, and then fight to the death and only rarely attempt to call reinforcements (I think DDO is the first MMORPG I've seen where the AI will actively ring an alarm to summon more monsters to help). It WOULD be more fun to deal with more interesting, organic tactics.
But it's not so much fun to deal with constantly fleeing creatures, or creatures that are consistently making "hit and run" attacks. While it's good, challenging AI - it's more frustrating than entertaining to the players.
So while I agree that better AI is *one* tool to help balance the multiplayer / single-player equation, it's not the whole answer.
Couldn't agree more, it just seems to be the most glaring flaw in most multiplayer games that aren't straight pvp. Also excellent point on "believable stupidity". I meant to mention it as a counterpoint to improving AI tactical efficiency but somehow managed to forget it.
Yes, constantly chasing fleeing enemies would be annoying, but constantly encountering low-power enemies that nonsensically fight to the death is just as annoying. In my mind you should almost never NEED to kill an enemy that is so weak that he simply runs from you.
I simply think that it would introduce some variety to the game without extrapolating to the extreme of huge heuristic architectures to emulate human behavior. Which, while a noble goal, is a hell of a project in and of itself. Its just a step not the landing.
In further explanation, rather than having tactical variety at random, It would be better I think to apply "AI templates" to (generically speaking in mmo terms) specific races and specific classes. For instance you might apply to an Orc Shaman the Orc template that causes him to tend toward melee combat and be relatively fearless against superior numbers. You would also apply the Shaman template that would cause him to tend to use spells and fight defensively when forced to. A rather simplistic example but I think it covers the basics of the idea.
A potentially more efficient solution would be to associate AI behaviors with the attributes of a given monster, large size causes it to be confident in combat and employs routines that take advantage of its greater reach and strength. Low health causes it to try to stay out of range of attack.
Somewhat tangently, I've noticed that designers tend to limit player freedom, both in actions and movement as a means of increasing difficulty. One of the most frustrating experiences for a player is to lose control of his character's actions. Especially if he has had more freedom earlier in the game.
Barriers to his free movement through a map (especially invisible or nonsensical ones, though if they add unique gameplay mechanics such as "rubber barriers" or they close off the way to somewhere that the player intuitively knows constains nothing interesting then this doesn't really apply), inability to do something he thinks he should be able to, or actual loss of direct control such as the character being knocked out or restrained (though if it happens in a "cut scene" the effect of that is mitigated), all decrease fun factor of a game.
Post a Comment
Yes, constantly chasing fleeing enemies would be annoying, but constantly encountering low-power enemies that nonsensically fight to the death is just as annoying. In my mind you should almost never NEED to kill an enemy that is so weak that he simply runs from you.
I simply think that it would introduce some variety to the game without extrapolating to the extreme of huge heuristic architectures to emulate human behavior. Which, while a noble goal, is a hell of a project in and of itself. Its just a step not the landing.
In further explanation, rather than having tactical variety at random, It would be better I think to apply "AI templates" to (generically speaking in mmo terms) specific races and specific classes. For instance you might apply to an Orc Shaman the Orc template that causes him to tend toward melee combat and be relatively fearless against superior numbers. You would also apply the Shaman template that would cause him to tend to use spells and fight defensively when forced to. A rather simplistic example but I think it covers the basics of the idea.
A potentially more efficient solution would be to associate AI behaviors with the attributes of a given monster, large size causes it to be confident in combat and employs routines that take advantage of its greater reach and strength. Low health causes it to try to stay out of range of attack.
Somewhat tangently, I've noticed that designers tend to limit player freedom, both in actions and movement as a means of increasing difficulty. One of the most frustrating experiences for a player is to lose control of his character's actions. Especially if he has had more freedom earlier in the game.
Barriers to his free movement through a map (especially invisible or nonsensical ones, though if they add unique gameplay mechanics such as "rubber barriers" or they close off the way to somewhere that the player intuitively knows constains nothing interesting then this doesn't really apply), inability to do something he thinks he should be able to, or actual loss of direct control such as the character being knocked out or restrained (though if it happens in a "cut scene" the effect of that is mitigated), all decrease fun factor of a game.
# posted by : April 26, 2006 5:38 PM
: April 26, 2006 5:38 PMLinks to this post:
<< Home
