Factional Alignment system: Creatures

[Keldaryth]

In terms of creature personalities, the range is actually pretty darn high because of the interaction between coefficients and attributes. Assuming that we only go with attributes and coefficients to 1 dp, you still have a range of about 21 points for each variable and coefficient, across all factions (currently 4 meaningful) and alignment scales (2 or 3) given 21x2x6 or 21x2x7 (I think) which is 252 to 294 possible personality combinations (svenskmand, please correct if I'm wrong ). Assuming we add no further alignment scales but we do add further factions, the number of personality options would grow by what, 42 every time a new faction is added?

Or am I getting it wrong in that instead of multiplication it should be powers... o_O

Even if I'm not wrong, I think nearly 300 personality combinations is plenty, and if I AM wrong, then we have far, far more than that

[andrewbuck]

The exact number of personality types is huge, I don't know if your calculations are exactly correct or not but they seem reasonable. The number of possible personalities is of course somewhat subjective, as you pointed out with one decimal place of accuracy there are quite a few, whereas with only 3 or 4 "stops" on each axis the number drops significantly. Of course in actuality, the number of possible personality types is somewhat of a moot point, the only reason I mentioned it in my previous post was to reiterate the point that if you get too many axes, the number of possible types becomes so large it is almost meaningless since only a few personalities will be represented by creatures anyway.

This actually provides some guidance for when you are picking values for creature attributes. For example, if you go with the three axis system and use N positions along each axis then there are N^3 possible personality types. If we have ~50 creatures and assuming we want to have approximately one or two creatures representing each of the types then n should be approximately "the cubed root of 50" stops per axis which is about 3. So the creatures would be roughly divided into groups on each axis by the ones from (-1 to -0.3), (-0.3 to 0.3), and (0.3 to 1). This tells you that it is not really worth worrying about whether a given creatures alignment is 0.1 or 0.2 since they are basically the same anyway. It is for this reason that I was opposed to having too may axes as you have to cut the number of stops per axis down so much that you basically have just a stop at each end and nothing in-between really matters.

-Buck

[svenskmand]

My objection to having too many axes is based on the fact that the number of possible creature personalities grows exponentially with the number of axes.

Well as each dimension is on the continous range [-1;1] there is an unlimited number of different creature personalities with just one dimension. So the exponential growth does not really make sense.

It seems like svenskmand wanted to go with a fairly high dimensional system so this is an argument for the three axes system over the two axis one. Ultimately, I would be fine with either 2 or 3, it makes little difference from a coding perspective. I will leave the final to the two of you since it will be you that will likely be filling in values for each of the creatures.

I think that if we choose the axis labels right then having allot of axes will probably just improve the depth of the alignment system as we can then better describe each creature in that system. But I agree that we should not have so many axes that we just fill in arbitrary numbers because we do not know what to fill in. So in short the axis labels should be important for the game.

In terms of creature personalities, the range is actually pretty darn high because of the interaction between coefficients and attributes. Assuming that we only go with attributes and coefficients to 1 dp, you still have a range of about 21 points for each variable and coefficient, across all factions (currently 4 meaningful) and alignment scales (2 or 3) given 21x2x6 or 21x2x7 (I think) which is 252 to 294 possible personality combinations (svenskmand, please correct if I'm wrong ). Assuming we add no further alignment scales but we do add further factions, the number of personality options would grow by what, 42 every time a new faction is added?

Or am I getting it wrong in that instead of multiplication it should be powers... o_O

Even if I'm not wrong, I think nearly 300 personality combinations is plenty, and if I AM wrong, then we have far, far more than that

You are WRONG :P, we have far far far far far far more. If we just go with the 0.01 resolution of each axis of the 3axis system then we have at least (2*1/0.01)^3 = 8000000 possibilities for the alignment of each creature. But if you imagine that the ranges are continuous then you already have unlimited possibilities from one axis. But as we are working with floating point numbers this is of course not the case, but with a 64 bit double we still have something around 2^64 possibilities for each dimension (this is not exact, but it is very large so you get the idea). Just to get the big picture then it is estimated that there is around 2^200 elementary particles in the entire universe, so 2^64 is HUGE).

[Keldaryth]

Well then, I vote to keep things simple - although I think for a more realistic approach 3 axes are better than 2 (I can't think of a complete system for 2 other than LC/GE). If we have that many possibilities, I don't see the need to overcomplicate both the design process and the calculation process with more than 3 axes, based on the fact that an almost all encompassing model can be built around the 3 axis system.

[andrewbuck]

Although it is relatively easy to implement the coding to compute the creature arrival probabilities, the fact that the alignment numbers need to be embedded in the level file and other places as well makes it somewhat difficult to switch between them. Because of this I don't think we will be able to test both in-game and even if I did code up two branches, one with each system, the game is not really complete enough yet anyway to determine which system would work better in practice (we don't even have creature stats like hitpoints and damage worked out yet). As such I think it would be best to try to determine which system to use before it gets coded in.

-Buck

[svenskmand]

I would suggest the 3 axis system then, as I like the axis names better

[Keldaryth]

I would suggest the 3 axis system then, as I like the axis names better

Same here Okay. 3 axis it is then I guess.

[andrewbuck]

Sounds good. I started implementing the portals last night and I got a preliminary system pushed to git. Currently it just picks randomly with probability 1/N for each creature, and when it picks one it just prints to the terminal rather than actually spawning the creature. Tonight I will try to get the probability system working and then I will focus on the creature spawning second. I hope to get them both done tonight but no guarantees.

-Buck

[xahodo]

Why not leave the game mechanics to scripts? These scripts needn't be packaged with scenarios.

More scales mean more detail and diversity.

Here are some suggestions for an alignment, each having a potential range of 5 values (-2 to 2):
- Introvert/Extrovert (prefers to stay to itself/loves to get in contact)
- Egoism/Altruism (care only for self/care for others)
- Passive/Active (lazy or constantly busy)

This way you get a range of 125 potential results. Enough? Now, this will make sure some creatures do not get along with one another very well. Good and evil are also in there.

Things like spell- and health points have nothing to do with scales like these. But these scales could provide modifiers for tests and influence behavior (perhaps indirectly). A goblin might become lonely if it isn't part of a group, but it certainly is an egoistic creature.