Advertisement

Generation: Character (The Next Generation of MMORPG's)

Started by May 28, 2009 11:40 PM
50 comments, last by Griffin_Kemp 15 years, 8 months ago
Quote:
Original post by Wai
I think the logic for this particular example is totally reversed. If you are the Emperor's Wife, assassins would offer you more to kill the Emperor. You are so close to the Emperor that if you flip side, you will succeed.

I wouldn't say it's totally reversed, but I would agree that creative offerings like you show an example of are something that adaptive systems could offer using log memory and recognition.

Both are accurate, but of different parts of the same concepts.
One is reacting and appealing to the professional trend of the character, while the other is reacting and appealing to the social trend character.

By social, I mean personal or ethical.
By professional, I mean impersonal and just a "job".

I like your example better for displaying adaptive systems though as it more accurately shows how the system can take something that is not necessarily quest related and turn it into a quest, and in so doing cause a character development.

Quote:
In the current dynamics, if you have never used a bow, and you start using it, your ranged accuracy starts low because you had never leveled it up. The value of using bow is already automatically low because it takes your character longer to kill something due to the low accuracy. You don't need an adaptive system on top of it to make it less valuable in the begining. The effect is already there. It is of lower value because your character simply cannot use it right. What is the role of adaptive system here?

Well...not really.
In most systems today, you simply cannot pick up and use things outside of your profession or class.
In the ones that you can, it remains useless until you seek out the skill training (through some spending option) to spend X on and slowly build points up in that way.

But largely at the moment, class structure is the ruling format and not alterable skill systems.

However, it was showing a very small piece of the adaptable system...probably one of the smallest forms actually.

Quote:
Isn't it more elegant to implement it using the existing action and competence pair?

I wouldn't say it takes these concept away, as just has the option of utilizing these already existing concepts and furthering them in ways by adding dynamics to them.

When you scale down to these minor levels, it's difficult to find where the current skill options systems vary from adaptable systems.
I think it is more obvious when it is involving concepts where the system is responding more to the character than direct items or statistics.

That said, it really depends on what the bow represents to an adaptable system.
If, for instance, a game considers bow users to be quiet and sleek users and you currently use an axe, then consistently using a Bow will offer more options to you to spend or train (depending on how the system works) on more quiet and sleek skill sets.

So picking up that bow and getting better at it isn't just about the bow...it's about what that bow represents to the system.

If the system only sees the bow as a bow, then it would be no more different than, as you point out, some systems in place today.

Quote:
If I want immersion, I would simulate the internal thoughts as the player attempt to choose the option [Lie]:

The first time the player clicks on it, the character would respond: "Lie! I will not do that!" And nothing would happen. If the character click on it again, the character would have a respond that considers it more, and on the third attempt the character would lie. Depending on the intelligence and emotional control of the character (stats) and the listener of the lie, the lie may either succeed or fail. In both cases, the character would earn some intelligence point and some emotional control point, and the lying affinity of the character would increase. This means that the next time the player chooses to lie, the character would not respond as skeptically.

This mechanism is no diffient from that determines the miss or hit of an arrow. The difference is that the game needs contextual responds so that the player won't get bored reading the same text over and over. So it is easier to get implemented on actions that aren't verbal. But what is the role of adaptive system here? Or is this the kind of adaptation you were talking about?

Now you are hitting on the idea of why I said, "Generation: Character" and spent most of the time talking about how it will recognize the character choices of the player's character and not just it's statistics.


Now further this just a bit more in imagination (of which you have a very good one):
You may be on a quest and in a village to do whatever you need to do there (pick up supplies, rest, drop off x thing, pick up x thing, etc...) and some(artificial)one in the village that you talk to may end up "reading" your log and because of their ai character design react to your log (rumor, the word, overheard, etc...) and instead of just telling you information or selling you something, they may suddenly offer to help you pitching that you could use a good Y type of character for the quest you are on and don't seem to have one.

It recognizes you, what you are doing, what you do not have that may help in doing it, and then offers to interact along with you in your playing.

When you ask questions, the answers will be more regarded towards your log, and as such, fully developed, would cause more seemingly real character responses and interactions with the system.

---
I like your ideas though, you show some great methods of how adaptive systems can work.

Quote:

When you ask questions, the answers will be more regarded towards your log, and as such, fully developed, would cause more seemingly real character responses and interactions with the system.


Seems like a metric buttload of work for what? Mildly more interactive NPCs? I'm not sure that talking with fake people is really what people are looking for out of MMOs.

Quote:

2) The altered situation:
o Characters that perform out-of-character actions will have their characters shift to that character automatically
o Players have no freedom to dress one way and consistently behave differently
o The game enforces in-character behavior


This seems not fun. Also 'out of character' is going to be essentially impossible to codify without forcing your thousands of users into a few stereotypical character buckets.
Advertisement
Quote:
Original post by Telastyn
Quote:

When you ask questions, the answers will be more regarded towards your log, and as such, fully developed, would cause more seemingly real character responses and interactions with the system.


Seems like a metric buttload of work for what? Mildly more interactive NPCs? I'm not sure that talking with fake people is really what people are looking for out of MMOs.

And yet, it is what is constantly being worked on.
AI reaction and interaction is a leading research field in the gaming industry; as I've shown in just a few examples of papers on the subject matter.

And I'm not suggesting that people prefer AI to players, but AI is inevitable, and represents the world and the system; it is part of what defines what the system is.
The more responsive such things are to players by recognition of the players, as opposed to simple "wall" answers and programing, the more a player will think of the world as a place they are in and not on.

It will be less obvious that they are on a server, and more feeling like they are in a world.

So it may seem like a "buttload" of work for seemingly little difference to you, but that difference is actually massive in effect and extension.

For example, it's a ton of work to nail down an AI system for an FPS game that behaves as humans do with movement.
It's far easier to just tell the AI it's parameters and path, and then tell it it's protocol that will cause it to engage with the players.

Yes, it's harder to have the system monitor human movement and create a complex algorithm that "pushes" the AI around the map as the AI moves in a "search" pattern with only it's goal defined and not how to get there, so to come within 90% of the accuracy of how a human player moves in the game.

Yes that's harder, but it's also simply better.

Typically speaking, the first steps towards such concepts are incredibly difficult and do not offer much return compared to their later incarnations.

That said, I don't think it's as painful as you may be thinking it is when we look at it being done 3 to 5 years from now and not today.

Firstly, you have things on the developers side already moving in this direction of log memory in the fight against player-bots within 15 minutes.
So the technology to read the log and reuse it as AI memory is something that is possible, and thanks to the folks at MIT, the technology for behaviorally based and biased AI reaction is possible as too the ability for unique character qualities between AI's based on their "growth" period.

These things do exist now, but are just in their infancy of use.

So I don't really think that it will be a next to impossible task of creating and assimilating artificial intelligence from scratch for a mitigated return.
I think that it will be a large job of learning and adapting various systems together and then tailoring them to the interests of the game itself.

Eventually, such functions will just be an industry pack, just as today the various game engines are.

To be honest actually...the amount of work that goes into the ever increasing system of graphic engines is about equal to the amount of work that continues to occur in artificial intelligences, so to say one is less worth it than the other is ultimately to stop progress.


Quote:
Original post by Telastyn
Quote:

2) The altered situation:
o Characters that perform out-of-character actions will have their characters shift to that character automatically
o Players have no freedom to dress one way and consistently behave differently
o The game enforces in-character behavior


This seems not fun. Also 'out of character' is going to be essentially impossible to codify without forcing your thousands of users into a few stereotypical character buckets.

No one suggested this was a good thing.
Wai thought it was part of a construct of what I was talking about, but it was not; not in these restrictive manners.

We've since moved past that part.
What NPCs need is to be restricted by the same rules as players, which inherently forces NPC AI to not be nonsense.
Re: The Embodiment of the Causes of Adaptive Variations

Quote:
2) The altered situation:
o Characters that perform out-of-character actions will have their characters shift to that character automatically
o Players have no freedom to dress one way and consistently behave differently
o The game enforces in-character behavior


These were derived from some of the descriptions of possible applications. It is everywhere in the thread. For instance, in the original post:
Quote:
This (the use of adaptation in next generation game) will mean that Mario will make a choice of where to go on the level and find that his superpowers shift accordingly to that choice, on a basic level.

It is subtle, but it implies that Mario cannot choose to go to a place and play using his current superpowers unchanged. Mario has no freedom to consistently go to a certain stage while keeping the old set of superpowers. I assume that the motivation behind it is to enforce immersion. The way Griffin envision how adaptation could be used is different than the way I envision them.

Quote:
Different material to wear, use, and own will cease and become available accordingly to these choices as well.
On the surface, this statement seems to say that, if you have an evil character, you cannot put on a good person uniform to do your evil things, because you have are on the evil path. If you ask me to guess the reason behind it, I would say that Griffin was trying to enforce immersion.

However, the same statement could be rewritten to reflect what Griffin really meant. One concept is to introduce the agent that embodies the discriminatory reactions in the statement:

Instead of writing

"Different materials to wear ... will cease ... accordingly to ... choices..."

It could have been written as:

"Vendors will cease to sell certain material to characters they don't trust."

Suddenly the statement makes total sense. You go to a proper store to buy a gun, the shop keeper ( in the interaction) asks to see your gun license. If you don't have a license, you can't get a gun at from the store. You need to go to the black market. On the other hand, if you just choose to get a gun from the black market, you break the law, and you will get into trouble if caught. Let:

A1 = Buy gun from black market
O1 = The police hires you

If your character had done A1, you lose the opportunity for O1 to happen.

This sounds completely reasonable. Now imagine that the game is set in medieval times. There is "weapon license" and there is no "database of felons". The shop keeper (a good shop keeper) can only guess whether you are a criminal from looking at your appearance and from hearing news about your recent actions in town, or the recent events in town in general. The AI of the shop keeper makes these decisions:

D1 = If customer looks shady, don't sell stuff
D2 = If customer has a reputation of being shady, don't sell stuff
D3 = If bad things have been happening in town and you don't know the customer, don't sell stuff

These decisions could use help of an adaptive system. For instance, to tell whether a customer looks shady, the game observes player-player interactions to reflect the rules on how a typical player would judge another player to be shady. By all mean, this could be done by manual--the developer could have set a bit such as, "If character is equiped with a dagger, the character is shady" -- but with the adaptive system running that automatically, the game no longer requires a group of designer to sit down and decide how the game AI would detect "shadiness"; the game would do all of that, and make changes to the rules as the trend of shady appearance changes. That is the point of this form of adaptation -- it adjusts itself when the features of what needs to be detected changes over time.

Why does this description makes so much more sense and so much easier to accept?

The reason is that the set of adaptive behaviors are embodied to an agent. When the player encounters this dynamic, the player is no longer thinking,

"Ah! the game designers want me to form groups and play a certain way, therefore there are all these restriction and rules and I must follow!",

but

"The game designers pay much attention to protrait what the in-game character would do, and the characters behave really realistically! I feel that my character is really alive in the game world! It is watching and reacting according to my behaviors!"

The player will accept the situation when they perceive the role of the designers are not as someone that enforces the behavior of the player, but someone that enforces the behavior of the game world.

You also see here why this thread was about immersion, and rightfully so.

Because the world reacts realistically, the player must think in-character in order to accomplish anything.

"Oh! If I plan to get a weapon from the store, I must be a good person, because the owner will not sell otherwise!"

If the player is resentful, instead of resent a designer (who is outside the game world) on the inconvenience, the player can now recent an agent who is inside the game world. Which is good because that is an in-character emotion.


* * *

So you are a bad person.

You went to the store and the store doesn't sell you a weapon because the owner finds you shady. So you decide to go to the clothing store to dress differently and try again. You bought new cloths, you changed, and you walk in the weapon shop. The owner still refuses. Why? Because the owner had already seen your face, he recognized you! Impression matters!

Lesson learned.

You go to the next town with your good clothing and got weapon there. Mission accomplished.

[Edited by - Wai on June 4, 2009 5:52:38 PM]
Quote:
Original post by loufoque
What NPCs need is to be restricted by the same rules as players, which inherently forces NPC AI to not be nonsense.

That would first require AI to have the imaginative ambition and emotion that players have.
Because they do not, slapping only restrictions onto AI would be akin to telling a stop sign that it must stop and wait before proceeding just like drivers must.
Of course this is silly since the stop sign won't ever move in the first place.

Same problem exists in AI.
They are barely "moving" at this point. Before we start restricting them, we first have to develop them well enough into becoming a problem that requires restriction.

Essentially; the day that we have to make rules to restrict AI like we do players is a great day because it means we've developed AI to a behavioral simulation comparable against the observed actions and behaviors of human players.

The silliness that you see in AI, is usually the same result as that found in robotics today.
A conflict of rule sets instructions, or a gap between them.
AI tends to do strange things when these happen.

----
Wai...you say the concept with a much more beautifully articulated tongue than I.
*applaud*

Once again, I love the way you think about this stuff; it's continually interesting to read.
Advertisement
Quote:
Same problem exists in AI.
They are barely "moving" at this point. Before we start restricting them, we first have to develop them well enough into becoming a problem that requires restriction.

They don't move because they have no reason to.
If they're restricted by the same rules as players, that means they've got a finite inventory. Therefore they can't give the same quest to everyone and give them rewards without gathering those rewards, or the materials to do them, beforehand.
That means NPCs will have to participate in the world economy, work, earn money, get supplies, express needs as quests to the players (yes, i'm talking of dynamic quest generation depending on the AI needs which themselves depend on the state of the world), etc.
You end up with NPCs being proper agents of the world, not more special than players, which is what they should be.

There is nothing difficult about programming that kind of thing.
Quote:
Original post by loufoque
Quote:
Same problem exists in AI.
They are barely "moving" at this point. Before we start restricting them, we first have to develop them well enough into becoming a problem that requires restriction.

They don't move because they have no reason to.
If they're restricted by the same rules as players, that means they've got a finite inventory. Therefore they can't give the same quest to everyone and give them rewards without gathering those rewards, or the materials to do them, beforehand.
That means NPCs will have to participate in the world economy, work, earn money, get supplies, express needs as quests to the players (yes, i'm talking of dynamic quest generation depending on the AI needs which themselves depend on the state of the world), etc.
You end up with NPCs being proper agents of the world, not more special than players, which is what they should be.

There is nothing difficult about programming that kind of thing.


Forcing supply and demand upon NPC's is never a bad idea, but that isn't quite the same as a universal slapping of the AI with the rules players are bound to.
That's just creating a dynamic supply and demand system for a dynamic player/npc economy.

Some games actually do things similar to this, but generally speaking, it hasn't been popular for reasons related to each games designs by the developers.

But like I said, that's a bit different than slapping rules on them like players have.

But it's a good idea never-the-less; one that I have always agreed with.
Quote:
But like I said, that's a bit different than slapping rules on them like players have.

How so?
Because that's only supply and demand.

Players have far more limitations on them that simply supply and demand.

This topic is closed to new replies.

Advertisement