Putting the 'role' back in role-playing games since 2002.
Donate to Codex
Good Old Games
  • Welcome to rpgcodex.net, a site dedicated to discussing computer based role-playing games in a free and open fashion. We're less strict than other forums, but please refer to the rules.

    "This message is awaiting moderator approval": All new users must pass through our moderation queue before they will be able to post normally. Until your account has "passed" your posts will only be visible to yourself (and moderators) until they are approved. Give us a week to get around to approving / deleting / ignoring your mundane opinion on crap before hassling us about it. Once you have passed the moderation period (think of it as a test), you will be able to post normally, just like all the other retards.

TB Systems of combat

Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
Okay, so you are just talking about action queues, huh? Well I thought you meant a fundamentally different type of system, instead of a method of implementing different combat systems.

Yeah, I first came into contact with action queues when I was researching the mechanics of RTS games. They tend to use them extensively, but instead they aren't unified action queues. There tends to be an action queue for each entity and during each game frame all the entities get to run their actions. I haven't actually run into a system such as the one you describe with just a single unified action queue.

I think a complication with that method is the AI. It has to be modular enough to work either as either real time or as turn based. Obviously the more real time it is, the harder it is for the AI to have any real strategy while if it's turn-based the AI has more time during which to think. So you probably have to have a level of detail AI setting associated with the different methods, which causes more development time to go into the AI.
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
dojoteef said:
Okay, so you are just talking about action queues, huh?

I guess so. I'm not an expert in game development, only a hobbyist. I'm using it for my own CRPG project, but currently with a hardwired TB appearance.

I'm using a single action queue. It works fairly well so far. It's not been a problem to divide the AI actions into steps, at least not in the TB appearance. OTOH I'm only having a few dozen actors on a map, so it might be different (maybe unsuitable) for large maps.

From your messages I know you're an AI expert. My AI is most likely very simple compared to what you have in mind.

1) Determine current top-goal from a list of possible goals
2) Establish plan to reach the goal
3) Execute plan unless
3a) plan cannot be continued due to a change in environment
3b) another goal becomes higher priority

these steps can easily be modulariszed, and even be divided into smaller actions if execution-time-per-action becomes a crucual factor (which OTOH means the AI will need longer to think, but the game can run more AIs in parallel).

You're right if the execution time for each action varies strongly (e.g. the user changes from 2 seconds to 0.1 seconds) it requires the AI code to be flexible, which it can only be to a certain extend. There will be a lower border, defined by the slowest AI action.

But now we are finally completely off-topic.
 

ichpokhudezh

Liturgist
Joined
Jul 9, 2004
Messages
179
Location
germantown, md
dojoteef said:
Vault Dweller said:
What's an average target? RPGs usually feature impressive bestiaries and that would make any average ratings rather pointless.
Wow thanks for helping me prove my point about mathematics. You can't even figure out how to take an average. ;)
It looks like you're not too familiar with the process either (or way too oblique expressing yourself) (Hint: what kind of average you're talking about? A person with a 'degree of mathematics' you're surprisingly vague.)
Averages are well suited for calculations that don't matter - a friend of mine likes to give a spiel on importance and many uses of average body temperature per hospiltal to demonstrate this.

dojoteef said:
Take the popularity of FPS games. There is an initial learning curve for the controls, like any game. Beyond that it's point and shoot. There might be puzzles or other gameplay mechanisims, but the core aspect is very simple. The bigger the weapon the more damage it does. They are even sorted out for you by default; pressing "1" will give you the weakest weapon and "10" the most powerful. In order to get cRPGs to a larger audience you have to find out what sort of things deter these people from playing the games. You can easily find that out by talking to the average joe.
Are you being dense on purpose? That's totally opposite of truth. FPS (especially team-based) do require tactics (once you pass easy/normal). Weapons differ much more than by size/respective damage.

Consider this as well:
Vault Dweller wrote:
However, my point was that knowing the average is pointless in RPGs because the number would be too abstract.

dojoteef said:
It isn't too abstract using the percentage and statistical distribution method that I mentioned. In fact it makes complete sense. It's no different than the idea of THAC0. It gives you a frame of reference. For example if you are trying to hit someone with a weapon. They might have a percent change to dodge, a percent chance to deflect the blow, etc. The thing is, when you combine all that information together you come out with one value: the percent of incoming attacks that they can avoid. You then find the distribution of enemies compared to their ability to avoid incoming attacks and voila, you can get the average, and it's meaningful to boot!
As a person with a degree in mathematics, you're probably familiar with the fact that having averages for 2 sets of numbers, you cannot derive the average of the combined set (w/o additional info). Thus, the actual value of averages to any analysis averages <g> to 0.

Now, let's return to your explanation. Imagine you've decided to disclose 'the percent of incoming attacks that they can avoid' as a 'stat'. A gamer downloads a mod which contains a new super weapon/monster available. The average is going to change, so the 'stat' will change as well. How does it make any sense that stats change as a result of a download (the only thing that's apparent to the gamer)?

[EDIT]
And this is just laughable:
Just look at console/Japanese style RPGs. They tend to be more streamlined, with fewer choices because it allows people to experience the parts of the game that they find most interesting: the character and story interaction. Unfortunately those type of games don't tend to give you much freedom in meaningful choices (multiple ways to complete quests, multiple endings, etc). And before I get bashed for saying this, it tends to be true for the majority of those games, but not for all.
Japanese games do more often than not include branching in the storyline and multiple endings. Most of the games that I'm familar with (that means heavily advertized ones) are jam-packed with stats which usually have not-readily apparent effects on the gameplay (e.g: Final Fantasy games, Star Ocean: 'Til the End of Time)
 

Vault Dweller

Commissar, Red Star Studio
Developer
Joined
Jan 7, 2003
Messages
28,038
dojoteef said:
Wrong, you create a distribution for the stats as well. Everything number in the game has a distribution. You have exactly average agility... what would that relate to in numbers? 50%.
Could be somewhat misleading. Average agility implies that you are average. 50% implies that you've got only half of the expected value. It kinda looks like you are starting a game as a cripple after an accident.

Yes it can be, but developers have come up with numerous ways to get around it, though they might be a bit imperfect. Just look at pre-generated character builds and answering questions to determine your character stats.
JA2 and DF/MW. There is a reason why people always ask for explanation of the questions. There is a reason why JA2 UB came with an option to bypass the answering questions part.

Let's look at loot, there are different dice used for the weapons. Some are d6, d8, d10, etc. Then you have different numbers of dice being used for those, then you can have modifiers like +1, +2, etc. Well, that means you have to calculations each time you want to compare weapons, and I guanrantee that many people would get confused determining things such as whether a 2d6 + 1 weapon was better than a 1d10 + 2. I mean heck the 1d10 weapon has bigger numbers, it must be better!
Ok, that's specific enough, although that's not the part that could be expressed in percents. I agree that it could be very confusing for idiots and it would be nice if dice-based games spell that out in an easy to read format like 3-13.

It is the user that has to make sense of those calculated numbers in order to progress the character though.
Still disagree with that. In all games you can go by trial and error without reading the manual or be concerned about the numbers. How do you know what spell works better for you? You give it a try. What about that sword? Swing it a few times. Long time ago I found a unique scimitar in Diablo that was insanely fast yet had 30% damage penalty. I tried it, and it was a great weapon.

Every game has something that doesn't work "as advertised". Numbers could be misleading, calculations could be fucked, effects could be different, but "field tests" don't lie. In Fallout, AP ammo was useless. I tried it a few times, didn't work very well, and that was it. I couldn't care less about the exact calculations or anything like that. Same with everything else.

You make it sound as if Diablo 2 had such a difficult system to master.
It wasn't. Yet it was filled with calculations that, according to you, would give a casual player a heart attack.

In fact it was no more complicated for most people than the original really.
Now that would be a lie. Suffice to say, Diablo 1 didn't have any skill trees (!), not to mention a truckload of other things that made gameplay more complex.

Well, what does that have to do with Diablo 2? Thing is, determining which ability to choose next from a tree is no more difficult. You look at what you want to get (and with the Diablos there is no really picking a skill that can't be used "to kill stuff", so the choice in all honesty becomes nothing more than choosing the skill you think might be the coolest to use) and you determine which stats you have to increase in order to get those abilities.
It's a bit more complicated than that. Also, it's kinda interesting that you've reduced the complexity of building a character in D2 to merely choosing one skill and meeting the requirements, yet choosing between 2D6+1 and 1d10+1 is a show stopper. Comparing barb's lvl 30 skills Berserk (mana cost 4, attack bonus 100%, magic damage 150%) and Whirlwind (mana cost 25, attack bonus 0, damage -50%) is not a simple decision and it would require damage tables. Now add lvl 18 skill Concentrate which at lvl30 gives you 210% defense, 170% attack, and 125% damage, and choosing whether to invest another point into Concentrate or Berserk becomes a math problem.

It's not a terribly complicated system and it works well for anyone to be able to understand: keep increasing the numbers.
Isn't that true for every system? What seems to be the problem than? It doesn't really matter what this number is as long as it's attached to your primary skill.

The thing is, at least with the system I have, you don't constantly just keep increasing numbers, because in many situations it doesn't make a whole lot of sense, especially in social situations. It's highly unlikely that the people you meet later on in the game are going to be SO much more sophisticated that they can ferret out intentions, call bluffs, etc so well that you need 20 times the number of points in those social skills than you needed in the begining of the game.. That's why a system that uses statistical distributions could likely overcome that problem with a Diablo-like design.
With that I can't argue. You may have a better system. It's the claim that there are huge problems with other systems that I disagree with.

As for the "more sophisticated people" thing, I'm sure you'd agree that there are people who are much harder to convince (current discussion is one such example). There could be different reasons: mistrust, stupidity, blind faith, strong opposite position, etc; so stronger, more convincing arguments, specific facts, or a different approach could be necessary. That's what higher skill represents, imo

<Dungeon Siege related stuff>
It would have made a strong point, if Diablo 2 success didn't prove that point wrong.
See above.
See what? You have 2 ways to deal with my point: either to prove that the gameplay was simplistic or that it wasn't as popular as DS. I don't see either.

If a small amount of complexity must be removed in order for a more elegant design, then I think it's a good idea. See V:tM rules compared to D&D rules.
I dislike VTM rules as too simplistic. The output, I agree, should be simple, the mechanics shouldn't.
 

Loof

Novice
Joined
Jan 20, 2005
Messages
10
Hajo said:
dojoteef said:
Okay, so you are just talking about action queues, huh?

I guess so. I'm not an expert in game development, only a hobbyist. I'm using it for my own CRPG project, but currently with a hardwired TB appearance.

I'm using a single action queue. It works fairly well so far. It's not been a problem to divide the AI actions into steps, at least not in the TB appearance. OTOH I'm only having a few dozen actors on a map, so it might be different (maybe unsuitable) for large maps.

From your messages I know you're an AI expert. My AI is most likely very simple compared to what you have in mind.

1) Determine current top-goal from a list of possible goals
2) Establish plan to reach the goal
3) Execute plan unless
3a) plan cannot be continued due to a change in environment
3b) another goal becomes higher priority

these steps can easily be modulariszed, and even be divided into smaller actions if execution-time-per-action becomes a crucual factor (which OTOH means the AI will need longer to think, but the game can run more AIs in parallel).

You're right if the execution time for each action varies strongly (e.g. the user changes from 2 seconds to 0.1 seconds) it requires the AI code to be flexible, which it can only be to a certain extend. There will be a lower border, defined by the slowest AI action.

But now we are finally completely off-topic.

Im not sure I understand exactly what you have now, bit wouldn't it be cooler with a threaded system so that you one action queue for each combatant.
That way you could realy go from true TB to true RT, and also use induvidual executions times for actions to let the player go for speed or strength and so on...
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
Loof said:
Hajo said:
dojoteef said:
Okay, so you are just talking about action queues, huh?
I guess so.

Im not sure I understand exactly what you have now, bit wouldn't it be cooler with a threaded system so that you one action queue for each combatant.

Coolness is one thing. Threading means to care about thread syncronisation. Depending on the programming language and the environment this will be a more or less massive problem. Threading adds a whole new class of potential errors.

If your project is simple, I wouldn't go for the coolness factor, but for safety and ease of work. I assume that there are some limits for cooperative multitaksing systems, yet you need to weight that restrcitions agains the problems that you introduce by multi-threading.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
ichpokhudezh said:
It looks like you're not too familiar with the process either (or way too oblique expressing yourself) (Hint: what kind of average you're talking about? A person with a 'degree of mathematics' you're surprisingly vague.)
First of all my discussion isn't about averages, it's instead about distributions. With every distribution function there is an associated percent function. I'm refering to 50% as being average. Also if I started discussing the use of the different pythagorean means, the quadratic mean, or any other such average I would then start to lose people in the details. There is no point in deriving equations on this board, it probably wouldn't benefit more than one or two people and doesn't futher the discussion.



ichpokhudezh said:
Averages are well suited for calculations that don't matter - a friend of mine likes to give a spiel on importance and many uses of average body temperature per hospiltal to demonstrate this.
This is a bunch of nonsense when it comes to debating my stance. What your friend is likely discussing is the relationship of different statistics in being able to determine aspects such as causaulity. On the other hand what I'm proposing is the use of statistical relationships with direct causal links. A person who is of average (arithmetic mean) strength is not going to be able to carry the same amount of loot as someone with above average strength. But once again, I'm not talking about a system of averages, I'm talking about distributions which would give you all the information you require anyway.



ichpokhudezh said:
Are you being dense on purpose? That's totally opposite of truth. FPS (especially team-based) do require tactics (once you pass easy/normal). Weapons differ much more than by size/respective damage.
You're the one being dense. Let's look at these tactics that you have to employ in an FPS game that are just as difficult as most RPG mechanics, shall we?
1)Being able to aim using the mouse.
2)Deciding whether to use an area effect weapon or one with direct damage capability.
3)Deciding where to camp/whether to camp.

Then if you want to look at team based games such as RTCW: Enemy Territory that have different classes, all you have to do is add a couple more "tactics". If there isn't a medic on the team, get one. Put mines in front of goals; people on the other team need to get someone that can spot mines because they are going to be in front of the goals. Well I think I've pretty much just described FPS tactics.

Though all of this is ignoring my point. In order to play an FPS all you have to do is take that initial step of learning to use the controls, everything else is intuitive. On the other hand with RPGs you have to determine at each level up what skills and stats to increase, what weapons to use, etc. Often times the decisions are superficial or don't have an easily identifable relationship with success in the game. When the system is too oblique (as you like to put it), then people have difficulty making the decisions.



ichpokhudezh said:
As a person with a degree in mathematics, you're probably familiar with the fact that having averages for 2 sets of numbers, you cannot derive the average of the combined set (w/o additional info). Thus, the actual value of averages to any analysis averages <g> to 0.
Always with the averages, huh... that labotomy didn't go so well did it; you can't seem to read a post clearly. I said nothing about just using "averages", instead I said the use of distributions. I merely kept using the average in a skill/stat as an easy reference for people to understand. Once you learn to read and find out what a distribution is and what you can do with it, then come back and discuss the merits of the system I propose. I'm sure you're friend might know what he's talking about when it comes to mathematics, but you definitely do not.



ichpokhudezh said:
Now, let's return to your explanation. Imagine you've decided to disclose 'the percent of incoming attacks that they can avoid' as a 'stat'. A gamer downloads a mod which contains a new super weapon/monster available. The average is going to change, so the 'stat' will change as well. How does it make any sense that stats change as a result of a download (the only thing that's apparent to the gamer)?
Oh, no you've got me there! Imagine a person makes a mod and it changes something! That's obviously going to ruin the game! What kind of bullshit logic are you trying to employ? Yes the distribution is going to change, though the average might not necessarily change. If you put a strong weapon in and a weak weapon to balance it, then no change in the midline. Though the average is of very little consequence compared to the distribution function.

But anyway what's the problem with such a change? It's no different than if someone created a mod that added a weapon and they decided to then balance the game so that it is still fun. If someone created an uber weapon mod that suddenly makes one weapon the only one that really matters, then they might as well take the other weapons out because it ruins the balance of the game. I really don't understand what you're argument on this point is.



If you want to write a biting response, expect one in kind. Next time at least try to get your facts straight though.
 

Loof

Novice
Joined
Jan 20, 2005
Messages
10
Coolness is one thing. Threading means to care about thread syncronisation. Depending on the programming language and the environment this will be a more or less massive problem. Threading adds a whole new class of potential errors.

If your project is simple, I wouldn't go for the coolness factor, but for safety and ease of work. I assume that there are some limits for cooperative multitaksing systems, yet you need to weight that restrcitions agains the problems that you introduce by multi-threading.

Yeah I understand the problems that this could cause (although i have never tryed to implement anything like my sugestion). I just figured sisnce we where already dreaming about the optimal solutions why not go all the way.

So what I was realy after was your input on the extension of the idea if it could be implemented without said bugs?
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
"Multithreaded action queues?"

Loof said:
So what I was realy after was your input on the extension of the idea if it could be implemented without said bugs?

It depends. I'd say it pushes the limits a bit further, but it can's solve the problem.

Overall your PC has a finite computation power. You need to distribute it among all the things your game will need to do. Threading helps, but if the total amount of computation is larger than the power of the PC, your game will slow down or come to a freeze, reagrdless if you use threading or cooperative multitasking.

There might be a few additional things to consider like multi-core CPUs, hyperthreading CPUs and I/O heavy operations. In such environments multithreading your application is a big win.

But actually this is the point where a real game developer should jump in and report his experiences with such designs.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
There doesn't seem to be any reason to have multiple threads of execution for the action queues. It adds unneeded complexity that can be minimized through the use of simpler methods, such as having a list of all action queues and running them during each game tick. The only possible advantage to threading the action queues that I can see is the ability to have them run at time intervals that are independent of each other, though that can easily be simulated without the need for threads.

Now I guess if you look at performance issues you might be able to justify the use of threads for multi-processor systems or hyperthreaded cpus, but the payoff is usually not worth the complexity. The synchronization with the use of threads can be a bitch, especially the more complicated your code is.

I'm also uncertain how adding threads will let someone go from true real time to true turn based combat. The method that Hajo described will give you true turn based or true real time, so once again the threads add unneeded complexity. So I don't think you're approach is meaningful in anyway, it isn't an optimal solution.
 

Loof

Novice
Joined
Jan 20, 2005
Messages
10
In that case I must have missunderstood Hajo, I thought his idea only had one queue and all actions no matter who wanted to execute them would be placed in the same queue. In such a system I can't see how two different combatants could be acting at the same time, and hence I can't see how it could be used to make the system truely realtime.

About thereading cousing alot of complexity. I can see why you would complain about unessesary complexity. What I realy wanted to sugest was the use of one queue per combatant that execute in pararlel, and since I haven't tryed to solve this type of problem myself I had never thought about how to go about implementing somthing like it , threading was simply a word i choose from the top of my head.

Loof

(edited to correct the most glaring typos...)
 

Loof

Novice
Joined
Jan 20, 2005
Messages
10
I think Im failing horribly at expressing myself here...
I was aware of just about everything you just told me. I think that the problem is simply that i missunderstood your original sugestion, as it now sounds like what you had in mind is what i was after all along. I was never trying to turn this into a programing thory discusion.

What i was getting at was if in your system several actions coud be executing at teh same time, and i dont mean actions in the prograing sence but in the gameplay sence.
For instance say that the next action in the queue is you shooting me in the head. :wink:
While the compuutation for this action would be posible to do extreemly fast, we will want to present this action alot slower to the player so he gets to see all the gory details, so say the animation for you drawing your gun aiming and fireing should take maybe 1 to 3 seconds. Now say the next action in the queue was a random mutant ducking behind a rock, would the mutant be able to start ducking before you had fired?
In other words how would your system represent two actions being atempted at the same time in the simulated world?

(damn seems I faild to keep the programing aspect out anyway, oh well...)

Loof

EDIT: how on earth did my post come before the post I was replying to ?????
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
Loof said:
In that case I must have missunderstood Hajo, I thought his idea only had one queue and all actions no matter who wanted to execute them would be placed in the same queue. In such a system I can't see how two different combatants could be acting at the same time, and hence I can't see how it could be used to make the system truely realtime.

Unless you have a multiprocessor machine, all things will happen one after the other in your computer. And even with threads on a multiprocessor machine you'll not want to have true parallelism, but prefer a sequence of actions, that gives your deterministic and predictable results.

The apperance of flow is just a question of granularity.

To have the appearance of continuity you need to ensure that you can update your world every ~50ms. Even if you don't use threads, this can still be achieved. You don't have to execute all actions from the queue(s) within this timeframe, but you need make sure none of the actions will take longer than 50ms.

Almost all actions can be divided, so it's nearly always possible to get below the bounds.

A single queue ist good because it automaticaly snchronizes all actions. A queue of queues (one sub-queue for each actor) can be handy, but it's just a management issue, not a question of realtime or not.

Loof said:
What I realy wanted to sugest was the use of one queue per combatant that execute in pararlel, and since I haven't tryed to solve this type of problem myself I had never thought about how to go about implementing somthing like it , threading was simply a word i choose from the top of my head.

:)

Once you get a bit deeper into this, you'll see that true parallelism is a very dangerous area. What you usually want is a quick sequence, not parallelism.

A movie is made by showing a lot of images quick after each other. It's a continuous flow, unless you slow down, then it becomes a slideshow, maybe even with a trigger needed between two slides. This is the same with the action queue. If actions are executed fast after each aother it appears to be realtime. If you execute them slower (or make them longer), it appears more and more turn based. if you need a trigger to end a turn, you have reached full turnbasedness.
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
Loof said:
In other words how would your system represent two actions being atempted at the same time in the simulated world?

The answer may be unsatisfying: a queue, even a sorted one, has a natural order of entries that have the same priority. E.g. the queue in my H-World project will enque a new event of the same priority before an existing event of this priority. If the queue is polled for new events, the event that appears first is taken.

So events scheduled for the same point of time are forced to a sequence, last entry wins. (Probably it'd be better to have "first entry wins" of a group of equal entries, I just descibed the current implementation, not the perfect solution).

As a player you probably can't notice this, because you can't keep track of all the data, so it appears to be "good luck" or "tough luck" and it will average out during the course of the game.

You can point out that in case of instant death events, this system can be highly unfair. It is, but a RPG game should not have instant death events IMO. So I think we can (should?) solve this problem on level of game design and not on level of program design.
 

ichpokhudezh

Liturgist
Joined
Jul 9, 2004
Messages
179
Location
germantown, md
Mudslinging. Meh. Ok, I'm game.

dojoteef said:
ichpokhudezh said:
It looks like you're not too familiar with the process either (or way too oblique expressing yourself) (Hint: what kind of average you're talking about? A person with a 'degree of mathematics' you're surprisingly vague.)
First of all my discussion isn't about averages, it's instead about distributions.

With every distribution function there is an associated percent function. I'm refering to 50% as being average.
Now are you talking 50th percentile as your average? That'd be the median in your case, fyi, unless you can present me a usable function beforehand. Which wouldn't be "statistical" then, but rather "analytical". "Statical" results for a char could be (and most probably, would be) different.

dojoteef said:
Also if I started discussing the use of the different pythagorean means, the quadratic mean, or any other such average I would then start to lose people in the details. There is no point in deriving equations on this board, it probably wouldn't benefit more than one or two people and doesn't futher the discussion.
I would think, if your goal is really "introducing an element of randomness that might actually be even more intuitive" throwing in a few simple and intuitive formulas would not be so bad, now would it?
But, nevertheless, you are more than welcome to try to use a harmonic mean of something to express something, I'd be willing to poke around the result, when you get one. Just attach a human-readable, game-relevant description to that.

dojoteef said:
ichpokhudezh said:
Averages are well suited for calculations that don't matter - a friend of mine likes to give a spiel on importance and many uses of average body temperature per hospiltal to demonstrate this.
This is a bunch of nonsense when it comes to debating my stance. What your friend is likely discussing is the relationship of different statistics in being able to determine aspects such as causaulity.
Another brainslip? Damn aerosoles, they chomp your neurons away every time.
First, you've probably meant 'correlation' instead of 'causality', mentioning the latter is plain ridiculous.
Second, what's he talking about is inability of a aggregate characteristic to fully represent real picture or give out details. And "the devil is in details", you know. Your inability to recognize that
distributions
you're talking about are just another type of aggregate statistics is sad, really.

dojoteef said:
On the other hand what I'm proposing is the use of statistical relationships with direct causal links. A person who is of average (arithmetic mean) strength is not going to be able to carry the same amount of loot as someone with above average strength. But once again, I'm not talking about a system of averages, I'm talking about distributions which would give you all the information you require anyway.
BTW, Your use of "statistical" when you refer to something given just fucking drives me nuts, as you've noticed already. :)

If you have a distribution (not a single value) than it would mean that person 1 would sometimes be able to carry as much or even more than person 2 from your example.

Let me quote VD
Vault Dweller said:
The problem here is that your chance to succeed isn't the same for all goals, or at least it shouldn't be. You may have an attack skill 50 which would mean 80% chance of hitting a rat and 10% chance of hitting an orc. So, percentages could be misleading. Same applies to locks, traps, and anything else.
whose point you have masterfully riposted by
dojoteef said:
You could easily set it up so that it wasn't misleading. You could say that the chance of success is that for an average target. Then you can give difficultly of said target: say it is greater than 65% of all targets. Now during character creation you can explain that all that is needed to find the percent chance of success against an actual target is to multiply the two numbers together and let them know that this will be done for them automatically with the result visible for them on the screen
Now, here comes average into play. Did the whole system of distributions just go out of the window?

dojoteef said:
You're the one being dense.
:)) I dig your style, really.

dojoteef said:
Let's look at these tactics that you have to employ in an FPS game that are just as difficult as most RPG mechanics, shall we?
1)Being able to aim using the mouse.
2)Deciding whether to use an area effect weapon or one with direct damage capability.
3)Deciding where to camp/whether to camp.
...
Though all of this is ignoring my point. In order to play an FPS all you have to do is take that initial step of learning to use the controls, everything else is intuitive.
Damn those multiplayer games, I knew the whole 'skillz' thing was a hoax! :))

Ok let's start with 2). How about range/fire rate considerations? How about ammo replenishment?
1) How about the target not standing still? FYI, the projectiles in FPSes do have speed and results are not instantaneous.
3) Seems like an easy choice, doesn't it? :)

Missing parts:
How about logistics in general? Ammo/armor/weapons replenishment
How about fallback tactics? (Say an enemy did not die from your first shot.)
How about flanking and ambushes? How about covering fire (Tribes guys seem to use that a lot).
How about all these creepy movement styles (strafe jumps/crouching/circling)?

But hey, small matter.

How about some pointers for my winning US Tennis Open? You seem to be good with the strategical approach to things you have no clue about.

Always with the averages, huh... that labotomy didn't go so well did it; you can't seem to read a post clearly. I said nothing about just using "averages", instead I said the use of distributions. I merely kept using the average in a skill/stat as an easy reference for people to understand.
See my note on your style. :) Oh, boy... You did one too many of those "labotomies", kids should do that with carcinogens!
Let me quote your own tirade:
Vault Dweller said:
However, my point was that knowing the average is pointless in RPGs because the number would be too abstract.
It isn't too abstract using the percentage and statistical distribution method that I mentioned. In fact it makes complete sense. It's no different than the idea of THAC0. It gives you a frame of reference. For example if you are trying to hit someone with a weapon. They might have a percent change to dodge, a percent chance to deflect the blow, etc. The thing is, when you combine all that information together you come out with one value: the percent of incoming attacks that they can avoid. You then find the distribution of enemies compared to their ability to avoid incoming attacks and voila, you can get the average, and it's meaningful to boot!
How's your reading? Notice the word "average"? Notice "meaningful" as contrasted with VD's "too abstract"?

dojoteef said:
Once you learn to read and find out what a distribution is and what you can do with it, then come back and discuss the merits of the system I propose.
"Merits"? Hmm... You don't happen to like to stand before a mirror and touch various <censored> parts of your body, do you?
So far I was talking about deficiencies, mostly.

But let me get this straight:
You're the one who wanted to create an "intuitive" system for a gamers not inclined to crunch numbers.
Now you're sending me back to get some math/probability/statistics classes to even _start_ discussing this "intuitive" method.

Oh, I readily admit, anything beyond normal distribution (well, skew it here or there) and most I will be able to contribute to discussion are obscenities, but wouldn't that make me the above average of your "target auditory"?


dojoteef said:
ichpokhudezh said:
Now, let's return to your explanation. Imagine you've decided to disclose 'the percent of incoming attacks that they can avoid' as a 'stat'. A gamer downloads a mod which contains a new super weapon/monster available. The average is going to change, so the 'stat' will change as well. How does it make any sense that stats change as a result of a download (the only thing that's apparent to the gamer)?
Oh, no you've got me there! Imagine a person makes a mod and it changes something! That's obviously going to ruin the game! What kind of bullshit logic are you trying to employ?
Yes the distribution is going to change, though the average might not necessarily change. If you put a strong weapon in and a weak weapon to balance it, then no change in the midline. Though the average is of very little consequence compared to the distribution function.

But anyway what's the problem with such a change? It's no different than if someone created a mod that added a weapon and they decided to then balance the game so that it is still fun. If someone created an uber weapon mod that suddenly makes one weapon the only one that really matters, then they might as well take the other weapons out because it ruins the balance of the game. I really don't understand what you're argument on this point is.
See "style" :) You're a moron (moving average of your brain activity).
I dig your thoughtful insights (see "average might...") which are vastly outperformed by lack of understanding of your own words (see "it's no different...").
See, if there's a mod with uber-weapon or uber-enemy (and there's a chanche you'd never encounter these), you would be still playing the same game.
Re-balancing on the other hand means the game rules have changed.
But now, since you've understood the caveat in your system, let's continue: how about crafting(which creates new uber items)/monsters that can get enhancements?
Suddenly, you 'distibution function' wobbles all the time.

Even if you were going to represent that instead of strength/agility/lockpicking/whatnot (which is more than questionable by itself), now _stats_ change whenever I summon a monster?

Let me give you an example from the RL: if person A uses doping in competition, it doesn't make person B lose his strength. B's chances to win might decrease somewhat, but they're not directly dependent of A's gains, there much more factors (A can get busted, die from OD, etc)
In short: conbined "success distribution" might depend on much more than a single factor and there's no way one would be able to calculate that as they are able now with just a few arithmetical operations.

Let's move into the base.
You have been arguing that a die, i.e uniformly distributed random value, independend of the game rules/item set/monster set is too complicated to comprehend.
Now you're presenting as a solution for everything a normal (at best) distribution with complex dependencies requiring heavy analysis as a viable alternative?
And you allow yourself to get offensive on that?
Puzzling.

dojoteef said:
Next time at least try to get your facts straight though.
Although it took some heavy thinking, I resolved not to discriminate and still use gay facts occasionally.
 

ichpokhudezh

Liturgist
Joined
Jul 9, 2004
Messages
179
Location
germantown, md
dojoteef said:
There doesn't seem to be any reason to have multiple threads of execution for the action queues. It adds unneeded complexity that can be minimized through the use of simpler methods, such as having a list of all action queues and running them during each game tick. The only possible advantage to threading the action queues that I can see is the ability to have them run at time intervals that are independent of each other, though that can easily be simulated without the need for threads.
Not so easily, apparently :) (just look at progress of multi-tasking on Win platform).
Multi-process/threading is quite easy to program, once you grasp the concepts. Making it work is much harder :). Debugging is tha bitch. Resource contention is no easy nut also.

The problem with ticks/queues is that (once you 've found your lowest common 'tick' denominator) you still need to manage object states by yourself and do a full-blown resource management, resorting sometimes to half-assed methods as fifo/lifo (see Hajo's post) instead of being fully parallel (this is not critique of the method, which is rather a consequence of the approach). You get this for free with threads. You still need to know about it and take care of it, tho.

Another thing is that event-based systems do not integrate well with action queue management.
Yet another thing is that actions/states could change faster that your system can display, so you'd need another thread and syncro points for drawing anyhow. Sleeping/waiting is a pain in the neck as well.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
ichpokhudezh said:
Mudslinging. Meh. Ok, I'm game.
I didn't start it, you did. I just told you I will reply in kind. If you will notice Vault Dweller and I might have differing opinions, but we didn't insult each other (though we did take some playful jabs). Instead in your initial post you had comments such as: "Are you being dense on purpose?" without any provocation.



ichpokhudezh said:
Now are you talking 50th percentile as your average? That'd be the median in your case, fyi, unless you can present me a usable function beforehand. Which wouldn't be "statistical" then, but rather "analytical". "Statical" results for a char could be (and most probably, would be) different.
Maybe this might help you. Goes into the definition of the statistical median and how for a normal distribution, this is in fact the mean, i.e average. I guess I could be more clear, but I find the fewer the terms I bandy about, the easier it might be for people understand the discussion, though I guess that only works up until a certain degree.



ichpokhudezh said:
I would think, if your goal is really "introducing an element of randomness that might actually be even more intuitive" throwing in a few simple and intuitive formulas would not be so bad, now would it?
But, nevertheless, you are more than welcome to try to use a harmonic mean of something to express something, I'd be willing to poke around the result, when you get one. Just attach a human-readable, game-relevant description to that.
I think you are not understanding my point. The underlying system might be complicated, but the final output for the player would be simple. If you remember what I stated at the beginning of the discussion, I wish to keep as much of the underlying complexity as possible, while making the numbers easier to read and understand.



ichpokhudezh said:
Another brainslip? Damn aerosoles, they chomp your neurons away every time.
First, you've probably meant 'correlation' instead of 'causality', mentioning the latter is plain ridiculous.
Second, what's he talking about is inability of a aggregate characteristic to fully represent real picture or give out details. And "the devil is in details", you know. Your inability to recognize that
distributions
you're talking about are just another type of aggregate statistics is sad, really.
Once again read up on statistical causality before speaking. Correlations aren't very useful. For example, during the summer months there is a pretty high correlation between the number of ice cream cones sold and the number of violent crimes*, would I thus be able to say the ice cream has an effect on violent crimes? If you think that's true, then their is no way I can reason with you. That's why I refer to causality, it holds more significance for the discussion. That way I can say because a person has a certain strength, they are able to carry more.

*For a further, yet fairly short and simple explaination check Section V. in this course lecture, or the "Analyze Data and Identify Patterns" section in this course lecture. Sorry I couldn't seem to pull up any nice information on the subject...maybe my search parameters weren't very good.



ichpokhudezh said:
BTW, Your use of "statistical" when you refer to something given just fucking drives me nuts, as you've noticed already. :)
Explain. I don't see what you are refering to.



ichpokhudezh said:
If you have a distribution (not a single value) than it would mean that person 1 would sometimes be able to carry as much or even more than person 2 from your example.
You need to explain this point better as well. I would love to see how you arrive at such a conclusion. Let's look at IQ; it is considered to have a standard distribution centered at an IQ of 100. So if one person has an IQ of 100 and the other has an IQ of 120, which person would be more capable of solving logic problems? The person with an IQ of 120, correct? Same goes with strength, so please explain your position a little bit better, because it doesn't seem to make sense.



ichpokhudezh said:
**NOTE: See his two quotes above, rather than making this post longer than necessary.**

Now, here comes average into play. Did the whole system of distributions just go out of the window?
No it didn't. I base my chance of success on the average target, ie. 50%. If you noticed I state that the target is 15% above average, ie. 65% (according to the probability density function). This all has to do with the distribution, not with a system of averages. To use the example of IQ, a person with an IQ of 100 has an average intelligence, meaning the person is smarter than 50% of the population. A person with an IQ of 120 instead is smarter than roughly 91% of the population. Can you understand it now?



ichpokhudezh said:
**Some stuff regarding FPS games.**
I'm simply not going to argue this one anymore. Anyone that thinks FPS games are as complicated, or even more so than RPGs, is just beyond any comprehension.



ichpokhudezh said:
**More lengthy quotes**
How's your reading? Notice the word "average"? Notice "meaningful" as contrasted with VD's "too abstract"?
If you notice I explain that the average is meaningful in a context of the distrubtion function. If you can't see beyond the word average when it comes up, then this debate is pointless.



ichpokhudezh said:
But let me get this straight:
You're the one who wanted to create an "intuitive" system for a gamers not inclined to crunch numbers.
Now you're sending me back to get some math/probability/statistics classes to even _start_ discussing this "intuitive" method.
As I stated above, it's to supposed to have intuitive numbers that a player can decipher, though it won't necessarily have an easy to follow formula underneath. Just like Vault Dweller said:
Vault Dweller said:
I dislike VTM rules as too simplistic. The output, I agree, should be simple, the mechanics shouldn't.
Which is what I would hopefully be going for with such a system.



ichpokhudezh said:
Oh, I readily admit, anything beyond normal distribution (well, skew it here or there) and most I will be able to contribute to discussion are obscenities...
That's why I'm trying to be less technical. Though it seems the only real way to demonstrate such a system for diehard skeptics without complicated math here would be to actually code up a system such as this (which I intend to do; it's the reason I started the discussion about this in the first place, to get feedback).



ichpokhudezh said:
See, if there's a mod with uber-weapon or uber-enemy (and there's a chanche you'd never encounter these), you would be still playing the same game.
Re-balancing on the other hand means the game rules have changed.
But now, since you've understood the caveat in your system, let's continue: how about crafting(which creates new uber items)/monsters that can get enhancements?
Suddenly, you 'distibution function' wobbles all the time.

Even if you were going to represent that instead of strength/agility/lockpicking/whatnot (which is more than questionable by itself), now _stats_ change whenever I summon a monster?
Let me address the different points.
  • Consider my system an "auto re-balancing" system. Maybe I want to make it so that a person that makes a stupid mod doesn't ruin the game for people. But the other thing to note, is that with a distribution, adding in one outlier isn't going to drastically change anything. Let me go back to the IQ discussion. Since a distrubtion has already been set, there is nothing stopping a person from having an IQ of 200, though the likelihood of that happening is very small.
  • Crafting does not necessarily create uber items, nor will it likely change the distribution very much. See the fact that the population of the world is continually increasing (for now), though the IQ distribution is not changing.
  • Once again, stats will not change do to the summoning of a monster. Summoning a monster or two will likely have little to no effect on the distribution.



ichpokhudezh said:
Let me give you an example from the RL: if person A uses doping in competition, it doesn't make person B lose his strength. B's chances to win might decrease somewhat, but they're not directly dependent of A's gains, there much more factors (A can get busted, die from OD, etc)
Though it might not seem intuitive, A's use of doping has an effect on B's chances to win without having to consider other factors. If you are up for the reason why, check out Bayes' Theorem.



ichpokhudezh said:
Now you're presenting as a solution for everything a normal (at best) distribution with complex dependencies requiring heavy analysis as a viable alternative?
And you allow yourself to get offensive on that?
Puzzling.
The dependencies are only complex underneath, though they would seem simple at the surface level. Someone with more strength, but lower sword skill will likely do the same amount of damage as someone with less strength, but more sword skill. Though all that the player will see is the fact that the amount of damage that can be dealt in the same amount of time is equal. So there might be several ways to reach to attain the same ability score, but it's not something the player actively has to worry about.



ichpokhudezh said:
dojoteef said:
Next time at least try to get your facts straight though.
Although it took some heavy thinking, I resolved not to discriminate and still use gay facts occasionally.
Cute. :)



Now you see how I can be civil. If you would return the favor that would be great. If instead you still wish to get into a flame war, we can do that as well.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
ichpokhudezh said:
Multi-process/threading is quite easy to program, once you grasp the concepts. Making it work is much harder :). Debugging is tha bitch. Resource contention is no easy nut also.
That's very contradictory considering debugging and making it work is part of programming it. ;) The thing is, if you execute the actions fast enough, it doesn't matter whether you have concurrent threads or a single path of execution, it will appear the same.



ichpokhudezh said:
The problem with ticks/queues is that (once you 've found your lowest common 'tick' denominator) you still need to manage object states by yourself and do a full-blown resource management, resorting sometimes to half-assed methods as fifo/lifo (see Hajo's post) instead of being fully parallel (this is not critique of the method, which is rather a consequence of the approach). You get this for free with threads. You still need to know about it and take care of it, tho.
You'll have to do resource management regardless. In fact it is not free in threads and can be much more difficult, considering you have to lock shared resources anytime you want to use them for fear of them being changed in the middle of a calculation by another thread. Once again, that adds complexity.



ichpokhudezh said:
Another thing is that event-based systems do not integrate well with action queue management.
Quite the contrary, event-based systems are ideal for action queue management. If you recieve an event for an action, add it to the queue. And since the action queue is looked at every game tic regardless, the action queues get executed as soon as they fill up.



ichpokhudezh said:
Yet another thing is that actions/states could change faster that your system can display, so you'd need another thread and syncro points for drawing anyhow.
I'm not sure what you are getting at here. All you have to do is ensure that your display is updated at the end of each game tick. The player is unlikely to be able to generate events quickly enough to overload the queues, and the other elements such as AI, physics, triggers, etc have to be balanced anyway so that they don't slow down the overall performance. That happens regardless of threads or not.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
I have a new question about TB vs. RT. Has anyone ever played the game Space Hulk? It wasn't an RPG, but it had an interesting tactical combat system where you controlled several characters. The game was essentially a RTwP type system, but it had a catch. You had a slowly replenishing bar that determined how much time you had to pause the game. This allowed the game to let you have control issuing orders, but lent a sense of immediacy to the combat.

I'm guessing most of you wouldn't like it because of the speed of having to choose weapons, spells, attack targets, etc in a short amount of time, but I thought it sounded like an interesting twist on the RTwP type system. What do you guys think of such a system for a more action oriented RPG?
 

Hajo

Liturgist
Joined
May 19, 2003
Messages
283
Location
Between now and then
dojoteef said:
You had a slowly replenishing bar that determined how much time you had to pause the game. This allowed the game to let you have control issuing orders, but lent a sense of immediacy to the combat.

Battle Isle IV had a similar system, but I assume the input time has been much longer than in your example.

I think it works well for games that emphasize strategy over tactics. It's an intersting twist to know not being able to cammand all units each turn burt only some - you'll have to make decisions.

I usually don't like games that force a certain pace on me, but the system in Battle Isle IV worked quite pleasantly and therfore I'm not immediately biased against it.

If you are planning your next project, and it will have this feature, I'll give it a try. Just make sure it workd well :)
 

Otaku_Hanzo

Erudite
Joined
Oct 19, 2003
Messages
3,463
Location
The state of insanity.
dojoteef said:
Someone with more strength, but lower sword skill will likely do the same amount of damage as someone with less strength, but more sword skill.

Just about any skilled swordsman would argue this statement fiercely. First off, the fighter with more strength and low skill is going to be slow. So slow in fact that he will most likely be easily readable and therefore easily defendable against, thusly doing less damage in the end. Even when he manages to hit, he does so sloppily to the point that he is not getting maximum efficiency out of his weapon.

Okay, sure he's getting a strength bonus to damage. But if you're going to be realistic about it and use a sword skill, then he's not going to get maximum damage from his weapon at low levels. At higher levels of skill though, he now starts learning the most effective way to land his blows to cause more damage, getting maximum effect from his weapon and even exceeding that with higher levels.

Okay, let's break this down in gaming terms (loosely, mind you):

Strength Damage Bonuses:
---------------------------------
16: +1
17: +2
18: +3

Weapon Skill Damage:
---------------------------
Low Skill: Base Damage
Medium Skill: +10% to 50% Damage
High Skill: +51 to 100% Damage


So, an 18 Str fighter with low skill swings a longsword (1d8 Base DMG) and manages to land a blow. He rolls an 8 for damage and does a whopping 11 points with his +3 modifier. Not bad.

Now, the 14 Str fighter steps up to bat with his medium skill and swings the longsword, landing a hit. He rolls an 8 for damage and adds 50% again to that for a whopping 12 points. Okay, not much more, but it IS more.

Of course a fighter with high strength and high skill by this system would do some serious damage, but that's not the point here, is it? ;)
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
Well, Otaku, I understand your concern, but you have to understand where I'm coming from for this argument.. First this was just a quick example off the top of my head. I didn't look at things such as balance, or whatever; notice I use the word likely. The point I was trying to make is that there would be multiple ways to reach any such ability score. Second thing of course that we all know is that realism doesn't always translate into fun. Once you have the system in place then you start balancing it for fun. It might work out that the way you do things is fun or not, that's just something that's too hard to actually determine before getting to play the game.



Hajo, I will be doing a mod that has that feature in it. I had initially planned to make a turn-based RPG, but in the time I have allotted for myself, I don't think it's possible. So, I've modified my initial idea. The game will be an action game. You will control a squad of marines through an overhead tactical screen, though you only have a limited amount of time to use the screen. Once you have finished issuing orders (or run out of time), you will pick a team member to "possess" and switch to a first person view and actually take part in the battle. There will be different classes, with different weapons and abilities. For those familiar to Battlezone you will find the game to be similar, but if I'm not mistaken there was no pause feature for Battlezone. The game is much more like Space Hulk, which is a part of the Warhammer 40K universe.

Since I'm using the Doom 3 engine to make the mod, you will be fighting the hordes of Hell (I'm not an artist, so I'm not going to be making my own models, skins, etc). Eventually I would hope to redo the whole game, with a dialog system, some quests, and a decent storyline, but for the intial release it would just be a tactical sqaud-type shooter. I think it's best to get things out there early and gather support for the mod, so I can hire good talent later on. So that's the way I'm planning it so far.
 

ichpokhudezh

Liturgist
Joined
Jul 9, 2004
Messages
179
Location
germantown, md
<OFFTOPIC mostly...and BORING, even for me>
dojoteef said:
ichpokhudezh said:
I just told you I will reply in kind
First, what's this, an elementary school? You are not going to read, think and argue, just "reply in kind"? Then the whole skit is nothing but self-aggrandizing.

Pardon me, but I've needed to edit the flow to make it more coherent (oof mark).

Instead in your initial post you had comments such as: "Are you being dense on purpose?" without any provocation.
...
<oof>Anyone that thinks FPS games are as complicated, or even more so than RPGs, is just beyond any comprehension.</oof>

This was a meaningful remark which expressed my opinion on your means, not an insult.
If you are insulted by this, avoid using this approach in the future.

Please re-read my remark on "US Tennis Open" (did you discard it for being a joke?). (I'll try to limit myself to shorter quotes, since the distribution of the 'chance' of you reading them deems my efforts useless)
Think about it.

Let me spell this out for you:
Anyone who says that an involved physical activity could be mastered by recognizing three or four general rules is either hopelessly dumb or pretends to be so. The former you aren't, hence to the latter you belong. Questions?

ichpokhudezh said:
That'd be the median in your case, fyi
Maybe this might help you. Goes into the definition of the statistical median and how for a normal distribution, this is in fact the mean, i.e average.

<oof>That's why I'm trying to be less technical.</oof>
That's fucking laughable. Chew on this: your plain-jane median. This short one glosses over the details and has no enchanting sigma signs either, but gives out the fact in plain english that for skewed distributions median is not nesessarily equal to the mean. (and yeah, normal is not skewed %), in case you didn't figure it out, and the opposite of 'skewed' would be 'symmentrical'). And game distros are going to be skewed, aren't they?

Now, since you were so coy although brisk and indiscreet but didn't provide anything that contradicts my conclusion let me pause and do some 'graph' analysis here.
You were angry at me because (in the order of 'probability' :)
1. you didn't know what were you talking about and I figured it first
2. you think it's so obvious that anybody who doesn't get it deserves to be sent the most roundabout way (remember those quadratic means? Can you say "tangent remark"?)
3. you wanted to conceal you 'very special' formula
4. you were running a long riddle and I've guessed it right before the punchline.

What the fuck is going on here, oh kind bearer of the mathematic degree?

ichpokhudezh said:
BTW, Your use of "statistical" when you refer to something given just fucking drives me nuts, as you've noticed already. :)
Explain. I don't see what you are refering to.
Since we're on referencing spree, here's mine:
Read on
a. differences between statistics and probability theory. Especially note differences in methods and initial assumptions.
b. definition of probability distribution

Now, a question.
Which one are you referring to when you mention "distribution [function]"?
Since there was no data for you to build your distribution on, and there's no way you can "choose" statistics openly (happens in RL, I know) your suggestion of "choosing a better statistical distribution" is funny at best.

You've made a lot of progress by referring to "probability density function", tho.

I guess I could be more clear, but I find the fewer the terms I bandy about, the easier it might be for people understand the discussion, though I guess that only works up until a certain degree.
I totally agree and you should do both...

I think you are not understanding my point. The underlying system might be complicated, but the final output for the player would be simple. If you remember what I stated at the beginning of the discussion, I wish to keep as much of the underlying complexity as possible, while making the numbers easier to read and understand.
And you choose to not understand neither mine nor VD's (as far as I can understand his words). Mechanics can be comlex while underlying rules can be simple enough. What you propose is wrong since it presents values that do not allow easy operations with them and the baseline of your system is at the prohibitive level of required comprehension.

Note though, that the output is no more easy than the output of the original system and much less 'predictable' without special means.

Compare:
Monster: AC 15, Weapon A: 2d6, thac0 0, Weapon B: 1d10 thac0 5. Baselines are obvious, the distribution is known beforehand and common (yah, the uniform one). Will take some number crunching but very possible even without your TI-83, wouldn't you agree?
Monster 78%, PC Strength 33%, PC Dex 88%, Weapon A: damage 7 44%, Weapon B: damage 5.5 55% (success and stat rankings are bogus). Which is better? I feel my insides cringe even for uniform distibution, even worse with normals. And where's the baseline? It looks even less approachable with skewed distros (I imagine that would require multiplying def distro by damage-dealing distros by 'extra-hit-ability' distros but that's a shot in the sky, really, and I cannot imagine doing this on a piece of paper).

Once again read up on statistical causality before speaking.
I was going 'WTF?' but then I saw 'paths', 'graphs', 'inference' and said 'Aha!' :)
You don't happend to be one of them Bayesians? Bah, small matter. Just don't go preaching that numbers are subjective, that blows off my top.

Correlations aren't very useful.
This comes from the same person who references the method of analysis of systems of correlating values? Did you actually read those articles yourself? Are you familiar with what those methods do (which is, it doesn't find the 'cause' (as in A->B) exactly, but figures out a common factor dependency, so _methods_ changing parameter A would with great likelyhood change parameter B)?

For example, ...[icecream-crime example follows]...
Before you go on with your crusade, apply some analysis to that controversial statement (ice->crime). If you feed your system only with samples that confirm that correlation, your favorite Bayesian theorem would 'confirm' that your belief is fucking true. Moral: confirm what you state to be obvious with a sanity check.

I would suggest you go and read my friend's joke again. It appears you don't get it. At all. A pity, really, 'cause it's so applicable.

ichpokhudezh said:
If you have a distribution (not a single value) than it would mean that person 1 would sometimes be able to carry as much or even more than person 2 from your example.
You need to explain this point better as well. I would love to see how you arrive at such a conclusion. Let's look at IQ; it is considered to have a standard distribution centered at an IQ of 100. So if one person has an IQ of 100 and the other has an IQ of 120, which person would be more capable of solving logic problems? The person with an IQ of 120, correct? Same goes with strength, so please explain your position a little bit better, because it doesn't seem to make sense.
Ref spree again: probability, IQ

First, your complete statement is a burp out of your ass (as in "not true") which is a direct consequence from your misunderstanding of the term "IQ" and it's applicability.

Second, let's look at your example even on the valid domain (same age group): IQs (whatever their validity may be) do not guarantee that a particular 'problem' would be solved by a person with lower IQ and not by the other one.

average target, ie. 50%.
....
This all has to do with the distribution
...
Can you understand it now?
You baffle me with your persistence. Do you understand what are you talking about? Your 'distribution', henceforth the average, depends on the sample.

A person with an IQ of 120 instead is smarter than roughly 91% of the population.
Stupid notions like this bite off behinds of adventurers into wild Amazonia. Mostly since they misuse their terminology ('smarter' in this case).

**More lengthy quotes**
Neat. I put those to facilitate re-reading of certain passages. It doesn't seem to work, sadly. That's for plebs, isn't it?

ichpokhudezh said:
How's your reading?
Notice the word "average"? Notice "meaningful" as contrasted with VD's "too abstract"?
If you notice I explain that the average is meaningful in a context of the distrubtion function. If you can't see beyond the word average when it comes up, then this debate is pointless.
It seems your keyboard skills are so good your words don't even register in your own brain.

If you think that I should burn a candle next to your avatar every time you use distribution your hopes are greatly exaggerated.

Read the passage again (that lengthy quoted one). There's no "context<...>" when you present those numbers to an unsuspecting customer.

Vault Dweller said:
I dislike VTM rules as too simplistic. The output, I agree, should be simple, the mechanics shouldn't.
[EDIT: somehow a quote slipped: dojoteef was saying 'me too' kind of thing
There's a recognizable difference between rules, their application (mechanics), and result. If you need me to, I can send you some links with reading material on that. Look at the 'emergent behavior' systems favorite example - termites - simple rules, complex mechanics, simple outcome. Here's a good link with amusing picture in it.

Though it seems the only real way to demonstrate such a system for diehard skeptics without complicated math here would be to actually code up a system such as this
(which I intend to do; it's the reason I started the discussion about this in the first place, to get feedback).
That's right, bimbo: to demonstrate something you've got to present something. No jerking off around the bush.
No need to code too - get the fucking formulas out.

<my example on "no reverse dependence", smartly 'retorted' by not reading it, not understanding it and then rehashing my own words and referencing the holy theorem/>
Bayesian sucks, what's new? Your misapplication of it blows twice.

The dependencies are only complex underneath, though they would seem simple at the surface level
...
but it's not something the player actively has to worry about
I doubt this. Want a cookie?

Cute. :)
...
Now you see how I can be civil. If you would return the favor that would be great. If instead you still wish to get into a flame war, we can do that as well.
No way! Score!!! So you're ok with them hot games either ;)

But let's finish with your bullshit shchienze first.
 
Self-Ejected

dojoteef

Self-Ejected
Joined
Oct 26, 2004
Messages
970
Wow, ichpokhudezh, I honestly don't know where to being. You're writing style belies your vacuous intelligence. There is seemingly no order or coherence to most of your comments. Certain sections were so incomprehensible that I can't even begin to imagine what sort of drugs you are on. You definitely show your lack of mathematical knowledge, especially in the realms of statistics and probabilities.

I don't know why I'm bothering to inform you considering your inability to grasp even the most simple of explainations, but who knows, maybe one of them might latch onto an area of your brain that isn't dead from being drop on your head as child.



ichpokhudezh said:
Since we're on referencing spree, here's mine:
Read on
a. differences between statistics and probability theory. Especially note differences in methods and initial assumptions.
I'm not sure what you are trying to prove, but if you want I can give you a fairly simple layman's definition. The difference between statistics and probability can really be seen by the purpose of each. Statistics are gathered and aggregated in order to give an understanding to a certain problem, while probability uses those statistics in order to make an "educated guess". This is why statistics and probability tend to go hand in hand.



ichpokhudezh said:
b. definition of probability distribution

Now, a question.
Which one are you referring to when you mention "distribution [function]"?
Since there was no data for you to build your distribution on, and there's no way you can "choose" statistics openly (happens in RL, I know) your suggestion of "choosing a better statistical distribution" is funny at best.

You've made a lot of progress by referring to "probability density function", tho.
If you will refer to this page. You will understand what the distribution function is. I'll copy the relevant sections for you just in case you are too lazy to hit the link:
Mathworld said:
The distribution function D(x) (also called the cumulative density function (CDF) or probability distribution function)...The distribution function is therefore related to a continuous probability function P(x) ... so P(x) (when it exists) is simply the derivative of the distribution function P(x) = D'(x)
And if you check out the probability function it states:
Mathworld said:
The probability function P(x) (also called the probability density function (PDF) or density function)



In the system I propose, the statistics are calculated by compiling each parameter of every object in the world and can be used to determine the probability of success for every action. A way of choosing a better distribution is then simply a transformation of the cdf (cumulative density function) or pdf (probability density function) since one is the derivative of the other. That means you could change a normal distribution into say, a half-normal distribution.

So
nimg1670.gif


becomes
himg348.gif



So simply because of your lack of knowledge you think you have some sort of a case against my arguments? Most of your rant has no relevance except to show you are making attacks out of ignorace with no basis in regards to the discussion. I guess you feel threatened when it comes to discussing topics which you could never hope to understand. I find no point in discussing any of your other assertions. They hold no merit, and only hold up legitimate discussion rather than furthering the debate process.



ichpokhudezh said:
But let's finish with your bullshit shchienze first.
So you know what, I'll let you finish it without me. I've got better things to do than to reply to a lackwit the insists on wasting my time. BTW, "shchienze" ?
 

ichpokhudezh

Liturgist
Joined
Jul 9, 2004
Messages
179
Location
germantown, md
dojoteef said:
Wow, ichpokhudezh, I honestly don't know where to begin.

How about start with reading the thread again, especially your own posts?... That's all in vain, I know.

Apparently, you're here to boast your l33t statistical skills, not to discuss your methods and/or get some pointers/critique.

Ok, the most of personal remarks/cross-referencing/explanations will go PM (probably).

dojoteef said:
In the system I propose, the statistics are calculated by compiling each parameter of every object in the world
Oh, I figured that one out long time ago. To reiterate myself: how are you going to do this?
What are you going to do with 'potential' objects (summoned/crafted)?
How about zero/'infinite' objects (gods ('infinite' defense and attack, or walls/chests with 'defense' rating, but zero attack possibility) which are changing your 'statistics', affect some particular encounters but have next to no relevance to other game events?

A way of choosing a better distribution is then simply a transformation of the cdf (cumulative density function) or pdf (probability density function) since one is the derivative of the other. That means you could change a normal distribution into say, a half-normal distribution.
Please do read the links you're referring to. If you will still insist that transformations change functions (please refrain yourself from rehashing by 'explaining' and giving out 'helpful links' on the topic of "distribution IS A function") you've got to take your math classes again.

I've got better things to do than to reply to a lackwit the insists on wasting my time.
Yep, please do feel yourself 'statistically literate' and '50% witty' in your valuable time.
 

As an Amazon Associate, rpgcodex.net earns from qualifying purchases.
Back
Top Bottom