Reviving the "Roleplaying experience" thread
by Alfred Norris · in Game Design and Creative Issues · 06/05/2005 (11:47 pm) · 110 replies
Wow. I just read the thread down there and it had some interesting throughts. Hopefully people would still like to discuss those issues?
1) Live people playing in-game monsters:
I believe you will see this more and more as long as the finances can support it. The Themis Group (MMO econ think tank) believes that the future of MMOs will be guided excursions.
In our project we are implementing 2 interfaces for GMs (who at release will be my online gaming guild that has 50+ members and plays SB and WoW)
a) An RTS-like interface - so that one person can control in-game mobs with one interface for much more realistic encounters. GM spawns 10 mobs of level 35 next to 6 characters average level 30. They send some mobs for what they think is the leader and scattered mobs for various other players. In the end its real people reacting to what they see.
b) Acting as NPCs - the ability to jump into any NPC and speak, attack, move around as that character. This can lead to highly customized reactions or conversations about a multitude of things. I can see that a GM might have to cook dinner and play at the same time in real life. Ok. Jump into a guard who is patrolling an area. The guard is scripted to move around, but when the GM can be at the computer he sees a player trying to sneak into the control room.
2) Do stats break immersion?:
There's a post about designing MMOs called "Mu's unbelievably long rant about designing MMOs" or something like that. He wrote it several years back and called for a no-stats game. The idea is tempting...but I just dont think it should be done.
a) people will find a way to make their character more powerful
Try as hard as you might to vary xp, randomly change xp based on lvl of mob, location and other factors...some geek is going to sit there with a pad and a pencil and find out that 1000 kills of Mob X = 32 skill points. You will get found out.
b) people LIKE to min/max stats - It lets them feel more in control and that that can really have a good character
c) can you realistically give non-numerical information that is as good?
Ok. So your character limps when his leg is broken and is a pretty good lock pick. But how does a character know that he only has 30 hp left out of 3000? If hes only pretty good at lockpicking...wont he get tired of trying to open an uber-hard safe...and just get frustrated?
Perhaps you can come up with an entire info feedback replacement for numbers, but I feel they are necessary. Besides PnP roleplaying is even MORE mathmatical but you dont hear people saying they roleplay LESS in PnP.
3) Player-run quests:
Im not opposed to this..but most of the examples given could have been fixed by going to the bazaar or teaming up with friends. You dont have the Ogre Staff of Pummelling? Why post a whole quest when you can buy one? (Personally Id rather go get it myself but Im being realistic about how MMO players work) Why post a quest when your buddies will help you get it? To be fair...perhaps you could work out a system where they post a quest that their buddies can accept, which gives more xp than merely going and killing the mob, but there has to be more than that...otherwise its just extra xp for clicking a button. You would need to link it to getting increased faction with the wizard guild (if the quest maker is a wizard) or increased standing in the city nearby.
4) Unreal Gameplay:
Someone spoke of Onslaught gameplay earlier and you must be looking at our game doc ;)
Why dont more MMOs (FPS or not) include UT2K4 gameplay? Assault maps in UT have multiple objectives like quests do. In fact they feel a little more immersive. But they fall flat because planting the bomb is just holding the "E" key down for 10 seconds. Why not link that to an RPG skill to make the whole thing more immersive? Or how about having Invasion gameplay...your clan town is invaded by a nearby religious cult? Or Capture the flag? Youve stolen a virus that the guards want to get back?
1) Live people playing in-game monsters:
I believe you will see this more and more as long as the finances can support it. The Themis Group (MMO econ think tank) believes that the future of MMOs will be guided excursions.
In our project we are implementing 2 interfaces for GMs (who at release will be my online gaming guild that has 50+ members and plays SB and WoW)
a) An RTS-like interface - so that one person can control in-game mobs with one interface for much more realistic encounters. GM spawns 10 mobs of level 35 next to 6 characters average level 30. They send some mobs for what they think is the leader and scattered mobs for various other players. In the end its real people reacting to what they see.
b) Acting as NPCs - the ability to jump into any NPC and speak, attack, move around as that character. This can lead to highly customized reactions or conversations about a multitude of things. I can see that a GM might have to cook dinner and play at the same time in real life. Ok. Jump into a guard who is patrolling an area. The guard is scripted to move around, but when the GM can be at the computer he sees a player trying to sneak into the control room.
2) Do stats break immersion?:
There's a post about designing MMOs called "Mu's unbelievably long rant about designing MMOs" or something like that. He wrote it several years back and called for a no-stats game. The idea is tempting...but I just dont think it should be done.
a) people will find a way to make their character more powerful
Try as hard as you might to vary xp, randomly change xp based on lvl of mob, location and other factors...some geek is going to sit there with a pad and a pencil and find out that 1000 kills of Mob X = 32 skill points. You will get found out.
b) people LIKE to min/max stats - It lets them feel more in control and that that can really have a good character
c) can you realistically give non-numerical information that is as good?
Ok. So your character limps when his leg is broken and is a pretty good lock pick. But how does a character know that he only has 30 hp left out of 3000? If hes only pretty good at lockpicking...wont he get tired of trying to open an uber-hard safe...and just get frustrated?
Perhaps you can come up with an entire info feedback replacement for numbers, but I feel they are necessary. Besides PnP roleplaying is even MORE mathmatical but you dont hear people saying they roleplay LESS in PnP.
3) Player-run quests:
Im not opposed to this..but most of the examples given could have been fixed by going to the bazaar or teaming up with friends. You dont have the Ogre Staff of Pummelling? Why post a whole quest when you can buy one? (Personally Id rather go get it myself but Im being realistic about how MMO players work) Why post a quest when your buddies will help you get it? To be fair...perhaps you could work out a system where they post a quest that their buddies can accept, which gives more xp than merely going and killing the mob, but there has to be more than that...otherwise its just extra xp for clicking a button. You would need to link it to getting increased faction with the wizard guild (if the quest maker is a wizard) or increased standing in the city nearby.
4) Unreal Gameplay:
Someone spoke of Onslaught gameplay earlier and you must be looking at our game doc ;)
Why dont more MMOs (FPS or not) include UT2K4 gameplay? Assault maps in UT have multiple objectives like quests do. In fact they feel a little more immersive. But they fall flat because planting the bomb is just holding the "E" key down for 10 seconds. Why not link that to an RPG skill to make the whole thing more immersive? Or how about having Invasion gameplay...your clan town is invaded by a nearby religious cult? Or Capture the flag? Youve stolen a virus that the guards want to get back?
#82
07/16/2005 (8:23 am)
Regarding P to P MMO security, that is a big concern that I thought of. How do singleplayer games with multiplayer components handle cheating? I have no interest in doing an MMO myself, so this is all academic to me, but it is an interesting problem.
#83
Allowing almost anyone to create a guild and have the ability to build and upgrade/personalize their own Hall could be helpful in helping guilds and guild members to feel more like they were a part of the world in addition to giving them an anchor point from which to start.
07/18/2005 (11:11 am)
Another option would be to make the Guild Hall or what have you a common respawn or bind point, so that on player death or use of a "town portal" item, they would end up in there. This way they could still log out in the field and return to where they logged out.Allowing almost anyone to create a guild and have the ability to build and upgrade/personalize their own Hall could be helpful in helping guilds and guild members to feel more like they were a part of the world in addition to giving them an anchor point from which to start.
#84
Brian "Kenryoku" Hunter's post on Jun 17, 2005
I think this statement is heading down the path of eventualities for RPG games in general. If our goal is to provide an environment that facilitates role-play, then the best way to do this would be ot have npc's speak and react to pc's in a highly realistic manner. If 80% of the characters you run into are npc's and all the npc's act upon information they would reasonably know (such as the character is wearing clothing of a farm worker) and npc's can initiate events/actions suitable for the situation (pc farmworker walks around non-chalantly in the presence of the npc provincial lord causing the lord to have the pc thrown into jail for the night.) then players would be hard pressed to not role-play.
To say it simply: Realism facilitates role-play.
I've considered utilizing chat bot code (such as ALICEbot). To describe it simply, the "brain" of the chat bot converses by determing what [u]topic[/u] the human is aparantly talking about and will respond with a stored reply to that topic. Basically the bot is a huge database that will dynamically learn new topics and record responses for them. Topic information can also be stored by "personality", letting one bot know something that another bot doesn't. Using this, I would create one generic personality that is filled up with the common lore of the game world. Then npc-profession (shop keeper, blacksmith, farmer, etc) specific personalities that would combine the generic personality with profession specific topics (if you ask the blacksmith how bronze is made he would tell you just how it is, but if you ask a farmer he would tell you to go ask a blacksmith).
The major disadvantage to this is that the server load is greatly increased simply by players chatting, as nearby npcs will have to track the topics and reply when suitable. Also, after this initial step is taken, the next step will be to allow topics to trigger actions by the npc (tell the blacksmith that he sucks and he might just throw a hammer at you). Creating that engine would be quite a difficult and time consuming process.
07/18/2005 (12:16 pm)
(Digging from way back here...)Brian "Kenryoku" Hunter's post on Jun 17, 2005
Quote:I think the first step that would go a long ways to helping the roleplaying aspect of MMORPGs and taking some of the spotlight away from the treadmill would be for the NPCs to dynamically react to persistance elements throughout.
I think this statement is heading down the path of eventualities for RPG games in general. If our goal is to provide an environment that facilitates role-play, then the best way to do this would be ot have npc's speak and react to pc's in a highly realistic manner. If 80% of the characters you run into are npc's and all the npc's act upon information they would reasonably know (such as the character is wearing clothing of a farm worker) and npc's can initiate events/actions suitable for the situation (pc farmworker walks around non-chalantly in the presence of the npc provincial lord causing the lord to have the pc thrown into jail for the night.) then players would be hard pressed to not role-play.
To say it simply: Realism facilitates role-play.
I've considered utilizing chat bot code (such as ALICEbot). To describe it simply, the "brain" of the chat bot converses by determing what [u]topic[/u] the human is aparantly talking about and will respond with a stored reply to that topic. Basically the bot is a huge database that will dynamically learn new topics and record responses for them. Topic information can also be stored by "personality", letting one bot know something that another bot doesn't. Using this, I would create one generic personality that is filled up with the common lore of the game world. Then npc-profession (shop keeper, blacksmith, farmer, etc) specific personalities that would combine the generic personality with profession specific topics (if you ask the blacksmith how bronze is made he would tell you just how it is, but if you ask a farmer he would tell you to go ask a blacksmith).
The major disadvantage to this is that the server load is greatly increased simply by players chatting, as nearby npcs will have to track the topics and reply when suitable. Also, after this initial step is taken, the next step will be to allow topics to trigger actions by the npc (tell the blacksmith that he sucks and he might just throw a hammer at you). Creating that engine would be quite a difficult and time consuming process.
#85
I think that research in the direction of truly conversation capable code is quite a worthy cause, but basing your game design on the assumption that such a technology is going to be available would be very dangerous. We just aren't to the point yet where you can have the universal Turing-tested AI GM just yet. Though, truly, that will be an awesome day :)
07/18/2005 (3:32 pm)
Quote: then the best way to do this would be ot have npc's speak and react to pc's in a highly realistic manner.While I agree that this would truly be the optimal solution for any computer based RPG, MMO or no, I really do not think that it can be reasonably accomplished with current technologies. While something like ALICEbot is fairly clever, it would require a huge amount of fine-tuning to actually make it feel as though the bots were reacting in a convincing way - and even then, they aren't tht smart, and players will end up spending their time trying to get the bots to react in amusing ways rather than actually roleplaying (though that could be an entertaining game, it is not the purpose of the bots)
I think that research in the direction of truly conversation capable code is quite a worthy cause, but basing your game design on the assumption that such a technology is going to be available would be very dangerous. We just aren't to the point yet where you can have the universal Turing-tested AI GM just yet. Though, truly, that will be an awesome day :)
#86
Story-generation engines...and the NPCs that play them out... ARE doable.
Check out this thread on GameDev.net
Click Here to read it.
About half-way down I give 2 links to two gentlemen (one of which we are working with) who are creating just such a system.
MMOs are going to have to become more scalable and yet more tailored at the same time. A hard bit to swallow, but the technology IS there. We just have to make it happen.
07/18/2005 (3:51 pm)
Being the one that (re) started this thread...I suppose I should chime in sometime.Story-generation engines...and the NPCs that play them out... ARE doable.
Check out this thread on GameDev.net
Click Here to read it.
About half-way down I give 2 links to two gentlemen (one of which we are working with) who are creating just such a system.
MMOs are going to have to become more scalable and yet more tailored at the same time. A hard bit to swallow, but the technology IS there. We just have to make it happen.
#87
07/18/2005 (4:35 pm)
Cool! Very interesting thread to be sure. I look forward to seeing/playing the results of such research as it progresses - it will be interesting to see if it is something that will be able to gain widespread acceptance among RPG developers and player, or if it something that is more appealing to the more "hardcore" RPG crowd (or both.)
#88
It's how I've planned everything, even the conversation. Everything that is interpereted into context could arise in conversation, and it'll most likely be a NPC that starts it! You could be walking through a town and a NPC might approach you to try and buy off your shield, or sell you some goods from a storefront. They might recognize you from a past meeting or even from tales of your adventures they've heard, asking,"So, whatever happened to that mean ol' dragon ye smited?!"
Even the PC characters get into the conversations with pre-empted response, set by the player. If you are a thief and decide to browse a shop for a 'discount' you might get away with capturing the shopkeeper in a good topic as you casually swipe that dagger you were eyeing!
So many different contexts to include though, a subject about, for instance, a hook might go into different paths depending on who you ask and how - maybe a fishing hook will come into context, maybe a hooked claw, maybe a hook you could place in your home to hang items on, etc.
Everything is context based as well as based on the mood of the personalities in question. A store keeper might be willing to give you a discount one moment, get robbed, then angrily throw you out the store!
What's more is, anything in context could be made into an adventure or quest, so you might be selling your shield to the guy above one minute, looking for a shield the next, then finding out the local smith is looking for some special 'plates' to make some new shields out of! Guess where this is headed?!
Again, it's some work, but fully possible.
- Ronixus
07/18/2005 (5:59 pm)
Derivitive Data Structuring - Structured data that derives from parent data which is context sensitive to the situation.It's how I've planned everything, even the conversation. Everything that is interpereted into context could arise in conversation, and it'll most likely be a NPC that starts it! You could be walking through a town and a NPC might approach you to try and buy off your shield, or sell you some goods from a storefront. They might recognize you from a past meeting or even from tales of your adventures they've heard, asking,"So, whatever happened to that mean ol' dragon ye smited?!"
Even the PC characters get into the conversations with pre-empted response, set by the player. If you are a thief and decide to browse a shop for a 'discount' you might get away with capturing the shopkeeper in a good topic as you casually swipe that dagger you were eyeing!
So many different contexts to include though, a subject about, for instance, a hook might go into different paths depending on who you ask and how - maybe a fishing hook will come into context, maybe a hooked claw, maybe a hook you could place in your home to hang items on, etc.
Everything is context based as well as based on the mood of the personalities in question. A store keeper might be willing to give you a discount one moment, get robbed, then angrily throw you out the store!
What's more is, anything in context could be made into an adventure or quest, so you might be selling your shield to the guy above one minute, looking for a shield the next, then finding out the local smith is looking for some special 'plates' to make some new shields out of! Guess where this is headed?!
Again, it's some work, but fully possible.
- Ronixus
#89
A text-parsing system would be tough, but potentially more rewarding in the end. But are players interested in that mode of interaction? The option system is easier, but I always thought it led the player too much.
07/18/2005 (8:00 pm)
What about the method of interaction with NPCs? Are some of you talking about moving towards the old text-parsing systems of past adventure games, or sticking with selection options, ala Neverwinter Nights?A text-parsing system would be tough, but potentially more rewarding in the end. But are players interested in that mode of interaction? The option system is easier, but I always thought it led the player too much.
#90
07/19/2005 (8:15 am)
@Mike Rubin: Check out ALICE bot (http://www.pandorabots.com/pandora/talk?botid=f5d922d97e345aa1)
#91
It was scalable, could do multiple languages - translations, accents, geographically based slang and vocal inflexions.
It was (apparently) set up so that you could tweak your own voices.
It occurred to me that something along these lines would be pretty cool in an RPG - actually have the NPC's speak in distinctive voices without having to store 8 billion megabytes of sound data on the user's computer.
Now if I can just remember the name of that product and post a link. =\
My head is so full of 1's and 0's lately that I'm running out of 'virtual memory' =P
07/19/2005 (8:46 am)
This may be a bit off the beaten path, but I recently got an email regarding a text-to-voice engine that, after listening to their samples, was pretty draned impressive.It was scalable, could do multiple languages - translations, accents, geographically based slang and vocal inflexions.
It was (apparently) set up so that you could tweak your own voices.
It occurred to me that something along these lines would be pretty cool in an RPG - actually have the NPC's speak in distinctive voices without having to store 8 billion megabytes of sound data on the user's computer.
Now if I can just remember the name of that product and post a link. =\
My head is so full of 1's and 0's lately that I'm running out of 'virtual memory' =P
#92
07/19/2005 (9:35 am)
@Michael: Good stuff, but I think the question is, are players really interested in that mode of interaction? If it's done well, I think it could be really interesting. However, it would mean a lot of work making different personalities, and players would need to do a lot of typing. I wonder if there would be a way to tighten it a bit, without making the kind of system where you choose among three or four possible questions or responses...
#93
@Alfred: The dynamic story generation probably the best next step to take. That whole thread is exciting and inspiring to read!
07/19/2005 (10:42 am)
@Mike: Using a speach-to-text converter would (potentially) aleivate alot of the typing, and the text-to-speach converter would add an extra perk to the game. Both of these options, however, put a large demand on the client machine.@Alfred: The dynamic story generation probably the best next step to take. That whole thread is exciting and inspiring to read!
#94
It did run on an old visual chat client called Palace that I hung out on when I was younger. Speaking of online roleplaying, that program was a great venue for just that. It had a pretty flexible scripting language that you could write decent dice rolling and other stat scripts in, and everyone could have a visual representation of their character. Plus you could switch room backgrounds as an admin (from nice images of scenery to a battle grid for example - on the battle grid we'd all use little FF3 style avatars :)
The program also had the ability to draw on the screen, so the DM could sketch out all the scenery and action fo the battle that way.
It was an interesting experience, because different people had done such drastically different things with a piece of software originally intended as a chat client. Currently, Second Life comes to mind as a similar sort of highly scriptable "sandbox" environment. Though I've never had time to check it out in depth. Ah for the days of highschool and infinite free time...
07/19/2005 (12:09 pm)
Apple has had a fairly robust text-to-speech system as part of their OS for nearly 10 years (anyone remember Hyper Parrot?). It is not very good with inflection, but it does have a few accents and tweakable speed/pitch parameters. Funny that no one has ever used it for a game as far as I know.It did run on an old visual chat client called Palace that I hung out on when I was younger. Speaking of online roleplaying, that program was a great venue for just that. It had a pretty flexible scripting language that you could write decent dice rolling and other stat scripts in, and everyone could have a visual representation of their character. Plus you could switch room backgrounds as an admin (from nice images of scenery to a battle grid for example - on the battle grid we'd all use little FF3 style avatars :)
The program also had the ability to draw on the screen, so the DM could sketch out all the scenery and action fo the battle that way.
It was an interesting experience, because different people had done such drastically different things with a piece of software originally intended as a chat client. Currently, Second Life comes to mind as a similar sort of highly scriptable "sandbox" environment. Though I've never had time to check it out in depth. Ah for the days of highschool and infinite free time...
#95
It would be quite interesting to use the option system in combination with the AI text creation system - in effect, having 4 or 5 text options for the player that are crafted by the AI based on what it knows of the character's goals and personality.
It would essentially be two bots talking to one another, but with the player having some control over the flow of the conversation. Conversation trees could be dynamically generated based on the plot elements that both the character and NPC were aware of, in addition to other secondary points that might be of common interest to both chracters.
07/19/2005 (12:15 pm)
Quote:A text-parsing system would be tough, but potentially more rewarding in the end. But are players interested in that mode of interaction? The option system is easier, but I always thought it led the player too much.
It would be quite interesting to use the option system in combination with the AI text creation system - in effect, having 4 or 5 text options for the player that are crafted by the AI based on what it knows of the character's goals and personality.
It would essentially be two bots talking to one another, but with the player having some control over the flow of the conversation. Conversation trees could be dynamically generated based on the plot elements that both the character and NPC were aware of, in addition to other secondary points that might be of common interest to both chracters.
#96
07/19/2005 (12:20 pm)
@Alex: That sounds like it has interesting potential... Do I smell a prototype?
#97
I am very impressed with how far this thread has evolved - really a lot of well reasoned discussion of some of the big design/gameplay challenges of the RPG genre. Keep it coming!
07/19/2005 (12:53 pm)
@Michael: No prototype from me anytime soon, I've been quite busy with things here at GG - and I'm primarilly an artist. I would love to see someone else take on the task though.I am very impressed with how far this thread has evolved - really a lot of well reasoned discussion of some of the big design/gameplay challenges of the RPG genre. Keep it coming!
#98
I wonder, though...would people really be interested in an RPG that involved pure text-based interaction with NPCs, the way old text-based adventures or ALICE systems do?
07/19/2005 (4:00 pm)
@Alex: Now there's an interesting idea...dynamically creating possible text options for the player. Very creative, but the system would still have to account for each of the possibilities in generating a response. That would potentially take away a bit from the way option systems lead the player.I wonder, though...would people really be interested in an RPG that involved pure text-based interaction with NPCs, the way old text-based adventures or ALICE systems do?
#99
07/19/2005 (4:07 pm)
@Mike: I wouldn't envision it as pure text based, but rather as controlling the dialogue options in a graphical game like Fallout or Neverwinter Nights. Or did you mean as opposed to voice acted?
#100
07/19/2005 (5:04 pm)
@Alex: No, you got it right. I was wondering if people would put up with a pure text-based interactive feature within an RPG, as opposed to a system where the player chooses dialogue options (like NWN). I thought NWN was a reasonably good system, but there are problems with it. The main problem I have with the NWN system is that the player is basically led by the options, limiting the overall experience. But it seems to work well enough. I suppose if you could dynamically generate the different dialogue options, then it could be more interesting... 
Torque Owner Michael Brewer
Community feel is most certanly what MORPGs need to shoot for. The most fun I've ever had while palying games is when I joined a clan - not because we were any good (cause we wern't) but because I knew the people and their personalities. That humanity made the game fun. Being part of a tight knit team (read-"community") is fun all on its own, regardless of what type of game you are playing (or work you are doing!). Now that I've realized how my FPS game clan experience really does relate to RPG games, I feel that I understand the importance of having a smaller player community.
One thing that comes to mind that I'd like to just throw out there is how do you balance having a large world that provides lots of content for players to experience and explore (I LOVE exploring game worlds, finding new creatures, neat places, etc) yet keep the player community in close enough proximity that they have opportunity to socalize and adventure together?
Christoper just touched on this:
I like having players able to join some guild/faction right away, however if the game has the standard division of character levels then noobs won't have any reason to really socialize with high-levels. New characters will need to be capable of directly participating in [u]important[/u] and [u]useful[/u] activities and functions for the guild/faction.
Having a common meeting area is also a good idea. I think that would be particularly useful if players always started from their guild/faction's starting area every time they joined the game - the more often players have to come across the same group of people the more likely they are to initiate some socialization. This will have to be ironed out, as simply popping back to the local inn every time you disconnect and reconnect would cause major game play issues.