[MUD-Dev] Consistent Characters (Was Remote client connection)

[Paul Schwanz - Enterprise Services]

Phillip Lenhardt said:
>> I don't see how giving a limited menu of choices in a situation prevents
>> players from giving meaning to their characters. Who says that "say" and
>> "emote" aren't choices on that menu? If you're character is a paladin
>> and that paladin is talking to a good priest, why should that paladin
>> have the option of attacking the priest? After all, it is not good
>> roleplaying to do so since it is totally out of character. 

Travis Casey replied:
> How do you know it's out of character?  Are all paladins the same?
> Can paladins not be mistaken?  What if I want to play a fallen
> paladin, and am in the process of having him fall?  What if I want to
> play a flawed paladin -- a good man with a bad temper?  Simply by
> leaving off the "attack" option, you're both giving me information
> that I might not have had (that this is someone I'm not supposed to
> attack) *and* restricting the kinds of characters I can play.

Aye.  There's the rub.  But must we have one extreme or the other?  Do we only 
have the option to either rule out-of-hand any action that the designer decides 
may go against character or allow without consequence any action, no matter how 
out of place?

To me, consistency of character is essential in order to discern motivations and 
to provide deeper immersion.  If it is indeed desireable, is it possible to 
encourage it through game design without limiting (too much) the richness of 
character that requires a degree of flexibility?  Would it be possible to make 
actions which seem to violate character permissible, but not expedient?

I think that it is possible, but I'd propose three things which are necessary in 
order to make it work.

1)  Some method which gives players the ability to select from a robust list of 
values which will come reasonably close to defining their character in a way 
which the game engine can understand.  The usual "good" and "evil" are not 
adequate enough to describe how players view their characters.  The rogue who is 
known to pinch an item here or there, yet would never purposefully bring lasting 
harm to an innocent and is especially loyal to his friends is a common (and 
valid) archetype.  However, good and evil fall short when describing this 
archetype.  Even if we throw in "lawful" and "unlawful" it is not robust enough. 
 In designing a more appropriate method for defining characters, I'd suggest 
choosing values that form a natural continuum between extremes and lend 
themselves to being tracked through in-game actions.  I've identified these as 
possibilities:

Health-taking...........................Health-giving
(i.e. harming, killing).................(i.e. healing, nuturing)
Wealth-taking...........................Wealth-giving
(i.e. stealing, greed)..................(i.e. charity, generosity)
Power-taking............................Power-giving
(i.e. betrayal).........................(i.e. loyalty)
Information-taking......................Information-giving
(i.e. hoarding knowledge)...............(i.e. sharing wisdom)

To test your list of values, try describing common archetypes using those 
values.  If you can think of archetypes which cannot be described by your 
values, then you might need to rethink them.  Also, using layers to refine 
definitions of archetypes can increase greatly the number of character which can 
be defined.  For instance, you may need to lay something like order/chaos or 
passive/active on top of the above values to help define the different 
approaches one might take to health-giving.  The coscientious objector and the 
patriot both believe that they value life, but their approach is vastly 
different.  The key is to make it powerful and flexible enough so that the 
*player* feels like they can really nail down something of their character's 
world and life view.  The very act of actually working through these issues is 
the first step in consistent role play.

2)  Some method for tracking significant events in a manner that allows them to 
be compared to the player-defined values of the character.  So, knocking someone 
unconscious might be slightly health-taking.  Killing them would be very 
helth-takiing.  The developer would need to categorize events so that they are 
assigned an appropriate value.  Some values could be dynamic; calculated 
on-the-fly.  Other factors may be taken into account to determine where an 
action falls on the continuum.  For instance, if amount taken in a heist is 
compared to amount owned by the mark, then robbing from the rich might be only 
slightly wealth-taking.  Conversely, if amount given in charity is compared to 
amount owned by the receiver, then giving alms to the poor might be very 
wealth-giving.  So Robin Hood can be defined as an archetype who is, overall, 
more generous than greedy.

3)  Some method for comparing significant actions to the player-defined values 
of the character and encouraging the player to consistently take actions which 
reflect those values.  So the life-loving, health-giving, paladin decides to use 
a priest as an over-sized pin cushion for his large assortment of swords and 
daggers. That's not a very health-giving thing to do.  Now, other circumstances 
might have warrented the action.  Perhaps the priest was a spy?  If the paladin 
was even more committed to loyalty-to-nation/power-giving than to health-giving, 
then this action would still make sense, but the method used should not sit well 
with the paladin who is strongly health-giving.  Some algorithm needs to be put 
in place to simulate the character's feelings of guilt or self-loathing when 
they do things which are not in keeping with their values.  In general, actions 
which consistently reflect values will lead to quicker character development 
than actions which consistently contravene values, which in extreme cases could 
lead to a degradation in abilities.  In this manner, the design uses realistic 
consequences to encourage role play.  The paladin is allowed to have fits of 
temper, but he will suffer realistic consequences for acting in a manner 
inconsistent with his own world and life view.  The next time he draws his 
sword, he may be reminded of his character flaw.  He may be hindered by feelings 
of guilt and self-loathing over the unfortunate incident with the priest...not a 
good think to be thinking about in the middle of a sword fight.

> > And if the
> > paladin has a good in-game reason to attack the priest, then the menu
> > would present "attack" as an option.

 
> Of course, this requires keeping track of all possible in-game reasons
> why someone might do something, which is an AI-complete problem in a
> game world where players have their characters talking to each other
> freely.

And a fourth:

4)  Some method for allowing players to have direct, but limited, input into the 
game so that they can make up for the AI's inevitable lack of understanding of 
complicated situations.  The difficulty is that the input cannot be unmitigated, 
since it cannot be fully trusted.  If it has a real affect on other characters, 
then it needs to be balanced against trusted data.  One possibility is to 
balance player input against data gathered by the game.  If each event is 
tracked, then the event can be "observed" by other characters, evaluated against 
their own values, and relayed automatically to others with whom they have 
contact.  In this manner, reputation will begin to form automatically.  I see 
Bubba kill Buffy.  I disapprove since I am health-giving and Buffy was a friend. 
 I "talk" about the event or my general disapproval of Bubba to the shopkeep in 
a simulated conversation when I buy a new sword.  The shopkeep talks to others 
who buy swords.  A good amount of trusted data is collected and disseminated 
automatically by the game engine.  This gives us the option to open things up a 
bit to the player.  There may be extenuating circumstances that the game engine 
is not able to understand.  Players who understand these circumstances should be 
given a method for manually giving their approval of Bubba to the shopkeep and 
others.  Now believablility becomes an issue, but it shouldn't be too difficult 
for the shopkeep to check his reputation tables when evaluating conflicting 
information.  If he doesn't approve of the one giving him the information, then 
he is not likely to "believe" it.  In other words, his opinion of Bubba remains 
unchanged.

Obviously, this is a lot of work.  The question is whether or not consistency of 
character is important enough to the game to justify the effort and resources.  
I think that it is, but I am speaking as one who wants to play such a game and 
not as one who is likely to design it.

--Phinehas

------------------------------------------------------------------------------

		"All things are permissible,
			but not all things are expedient.
------------------------------------------------------------------------------






_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev