Tobold's Blog
Monday, October 12, 2009
 
Skill-point systems vs. class / level systems

Theodorus the mage approached the clearing from the south. There he was, standing right in the middle of the clearing, the big grey wolf which had caused Farmer Brown so much trouble with his sheep. Theodorus was just at the limit of the range of his fireball spell, perfect. He mumbled the arcane incantations, the fireball zipped from his fingertips towards the wolf, and then …

Then what exactly?

At this point in any fantasy role-playing game, be that pen & paper, single-player video game, or MMORPG, a set of calculations starts. Does the fireball hit the wolf or miss it? If it hits, how much damage does it deal? And while in a shooter game hits and damage might be based on aiming, in a RPG they are solely or at least predominantly determined by the stats of our mage Theo and the stats of the wolf. From the stats, probabilities and damage numbers are calculated.

In a class and level-based game like World of Warcraft, the relative level of Theo and the wolf will already tell you a lot about the probably outcome. If Theo is much lower in level than the wolf, then the fireball will miss or just singe the wolf, and then the wolf will eat Theo. If Theo is much higher in level than the wolf, the fireball will one-shot the beast. In the most common case, with the mage and the wolf being of similar level, the wolf will lose a part of his hit points to the fireball, start sprinting towards the mage, and Theo will keep spamming fireballs on him until the wolf is dead, taking some damage himself in the process. How many fireballs Theo will need will depend on secondary things influencing stats: Gear, buffs, talents. But it is definitely the class and level of Theo which have the most influence on the fight.

Now some people dislike this sort of level- and class-based gaming, and prefer an approach based on skill-points. In a skill-point based game, Theo would not be "a mage", at least not written on his character sheet. But he would have a number of skill-points available to distribute between various game skills, and in this case would have put a number of points into either a general "magic" skill, or a specific "fireball" skill. It is obvious that ultimately the result is exactly the same as in a level-based game: Just instead of level and talents being used to calculate the probability to hit and the damage, we now calculate the same numbers using skill points as input.

The difference for the players is what options he gets in combining various skills. Our mage Theo, in most class/level-based games, would be able to throw fireballs, but he wouldn't be able to heal, and he wouldn't be able to wear plate mail armor. His class basically determines the "template" of his skills. In a game based on skill-points, he can choose that template himself. And some players like that choice.

Unfortunately having a higher degree of choice also has its disadvantages. Because as we said, in the end any template results in probabilities and numbers calculated. That is a mathematical problem, which can be solved for any given set of parameters. So after every patch changing the calculations, theorycrafters come and solve the mathematical problem, and calculate the optimal templates for every given role. That then becomes the "flavor of the month" template, until some developer realizes that something is overpowered, nerfs some ability, and changes the parameters in the next patch, resulting in new calculations finding new optimal templates. "Choice" for the player is an illusion, the best informed players will all "choose" the same optimum template. And the less well informed players who actually make choices themselves are referred to as "gimps", because they "gimped themselves" by making suboptimal choices. The well-informed meanwhile have chosen the optimum "tank-mage" template in which they are practically invulnerable while dealing maximum damage, or they specialized into one role for group play, either pure damage dealing, pure healing, or pure tanking.

Furthermore skill-point based games frequently have problems with how exactly you earn your skill-points. The basic idea is usually that you gain skill-points in the activity that you do. But often games have difficulty to determine whether what a player does makes sense or not. So Theo might be able to raise his fireball skill by launching fireballs at trees and rocks. Or a player might increase his running skill by putting a weight on the "W" key of his keyboard, have his character "run" against a wall, while he is afk to watch a movie.

Games with classes and levels basically provide players with fixed templates for skill-point distribution, and link the increase of those skills to a simple unified number, the character level, increased by experience points. Choice is limited to talent systems or to systems where you can specialize your class into some sub-class. On the surface it looks as if a player had more choice in a skill-point based game than in a class-based game. But in reality curiously the class-based games end up with having more variety. For example World of Warcraft has 30 different basic choices, plus variations thereof, in the form of 10 character classes with 3 talent trees each. Skill-point based games usually have a lot less, because there is just one optimum template for damage dealing, one for healing, and one for tanking. Extreme specialization usually is an optimum for any skill-point based system, which is why for example in Ultima Online everybody distributed his 700 skill-points into the maximum of 100 points into 7 different skills at "grandmaster" level. Putting 50 points here, and 50 points there was sub-optimal, so while mathematically there was a lot of choice, practically there wasn't all that much. Games with classes can offer choices like a healer still able to deal enough damage for soloing when he isn't in a group, which would be sub-optimal and thus not chosen if the players distributed the skill-points themselves. And they can make classes that are lets say sub-optimal for damage dealing attractive by adding other advantages and unique class skills.

So in summary I'd say that having some degree of choice is nice, and I certainly do like most sorts of specialization and talent systems. But I have yet to see a totally free skill-distribution system which didn't lead to flavor-of-the-month templates and ultimately less choice. Skill-point systems just disfavor the casual players who just want to play without first engaging in long theorycrafting and research of optimum templates.
Comments:
I agree 100%.
 
I'm glad you mentioned Ultima Online as it pretty much sums up your entire point for anyone that has played it in the past.

As always, Eve has a rather interesting spin on this - somewhat surprised you did not touch on it :)
 
Skill-point based games usually have a lot less, because there is just one optimum template for damage dealing, one for healing, and one for tanking.
Because class-based games are immune from having mathematically optimal choices, right?-)

If you keep the line of thought that there's only one optimal combination of choices for a role, then the only difference between class and skill-based systems is the granularity of the illusion of choice. Instead of "choosing" high defense and shield block, you "choose" a warrior. The end result is the same. Either you picked the FoTM or you didn't.
 
I've always preferred class/level systems to skill-based. I perceive "choice" as "work" and I don't play to work.

A modicum of choice for flavor can be "fun" rather than "work", though, so I'm open to making a few cosmetic decisions along the levelling path. Ultimately, though, I prefer a system like pre-AA EQ, where you pretty much know everything about a groupmate from his class, race and a quick inspect of his gear.

As for the "gimping" and "optimal" problems of skill-based systems or hybrid systems, I think that's largely a concern of the minority of players that post/blog/read forums. In-game I hardly ever meet people who even have a clear idea what "gimping" or "min-maxing" mean, let alone know what the current "flavor of the month" is.

Most players don;t read guides, forums, blogs or any other out-of-game sources. They just do what they feel like, or what their friends tell them, or their guild suggests or requires. So long as they are having a good time, they don't care how much "better" their character could be.
 
As always, Eve has a rather interesting spin on this - somewhat surprised you did not touch on it.

EVE doesn't have an overall cap to skill points, you can be maxed out in every skill there is in the game. Only that would take you literally years, and you could spend those years without actually playing, just logging on from time to time to select skills to train. This is a "rather interesting spin" I don't like at all.
 
I think that a lot of flavor of the month/cookie cutter choices are simply due a mindset problem.

I've for long been a fan of trying buids that are different to the mainstream and getting them to work. With varying levels of success.

I've always believed that in a well designed game, 'best' is situational. That is the closest any designer will get to balanced. Offering choices based on situation.

The problem comes in where it's often very hard to design two or more options that are A) radically different from each other and B) don't offer a clear-cut choice in which one is more effective in more situations or more common situations than the other/s.

The moment any choice offers even a cursory bonus above another, players all leap on it like so many rabid lemmings. Designers in the end are forced to nerf or buff things, not because one is necessarily so powerful it's affecting the game, but because every single player you see in the game has selected that choice, thinking themselves highly intelligent for reading Elitist Jerks and wowpopular. Soon enough those same players will whine and complain that the game is boring and offers too few choices.
 
The "interesting spin" of EVE isn't that it takes years to cap skills, that could easily change if you make a new game, letting people grind points instead of making it time limited... the real interesting spin of EVE is that "you are what you wear". You can have rank 5 in every skill associated with firing missiles while flying an amarr cruiser, but they mean nothing if you are in the cockpit of a caldari frigate armed with cannons. I'd love to see more games employ the "you are what you wear" design so that I don't have to make dozens of alts to play all the classes, instead I can keep my same identity (where all my friends can easily find me) and swap out my items to change the skills I'm employing to play the game in the fashion I wish to play it at that moment.
 
Out of curiosity, how does Magic the Gathering handle these issues?

It's effectively a skill-based game in that you can combine cards in various ways to create a deck. There are flavor-of-the-month decks, but you can't prove that a deck is optimal using a spreadsheet. And if one deck becomes really popular, people can design decks specifically to counter it.

If MMO combat systems can be solved using a spreadsheet, perhaps they're too one-dimensional?
 
Out of curiosity, how does Magic the Gathering handle these issues?

Well, MtG is a PvP game, and there is an element of rock-paper-scissors in the strategy. Thus if rock decks are flavor of the month, you'll do well with a paper deck, which is why there is never an absolute optimum.

You are right in saying that MMO combat is far more one-dimensional, as you can see with dps classes which basically end up defining themselves with a single number, the amount of damage they deal per second. But even if MMO combat were more complex, the fact that we are fighting static PvE mobs means that there is always a mathematically optimal solution. You would need mobs with neural network artifical intelligence changing their behavior to counter players when they use always the same strategy. Not likely to happen anytime soon.
 
It's effectively a skill-based game in that you can combine cards in various ways to create a deck. There are flavor-of-the-month decks, but you can't prove that a deck is optimal using a spreadsheet. And if one deck becomes really popular, people can design decks specifically to counter it.
The interesting part of Magic is that aside from the basic rules about how phases progress, 99% of the ruleset is in the cards. It's like trying to teach a computer how to play Go: It's simply unfeasible to conclusively calculate the best move within a reasonable amount of time.
 
You would need mobs with neural network artifical intelligence changing their behavior to counter players when they use always the same strategy. Not likely to happen anytime soon.

Actually, the PvP events in WoW are enough. like the on in the colloseum.

Imagine a game that consisted only of these non-tank encounters and all mages would be asked to skill more defense talents and kiting talents while also maintaining a decent dps.

You are right in that the reason for the ruthless min-maxing of a single number is the cause behind the cookie cutter builds (especially for dps, but also for healing and tanking).

But the real challenge is not to create encounters that circumvent this problem. This is easy.

The problem is to create encounters that circumvent these problems and are fun at the same time.

Blizzard repeatedly experimented with their pvp encounters; also in classic and TBC they had them. But they never found a way to make these encounters fun.

In my opinion, part of the problem is the fact that the encounters are predefined and thus predictable.
 
How about something similar to DDO levelling up system?

Have a number of predetermined paths, that have predetermined skills and points allocations etc, which the new or casual player can follow. For the initiated/hardcore, allow them to start a completely custom build, or break out of a predetermined build at anytime.
 
What 7 skills did everyone max out in UO? And can you provide a day-by-day training setup for EVE's FOTM? How about DF, do you have a skill-by-skill training guide to get into PvP asap?

Either you never really experienced a skill system far enough to 'get it', or you are oversimplifying this to try and make a point. The biggest joke is the number of 'options' you gave for WoW, which is beyond off. If you wanted to be a high-end (back when there was a high end) tank, how many options did you have? One, prot warrior with an exact talent layout and exactly the same gear, using the exact same buffs/pots/flasks. Anything less and have fun in Nax.

Class systems make it easy for players, reducing options to avoid confusion and people gimping themselves. They are also easier to balance, as you only have a few viable options that need to work in tandem. Balancing gear when you limit it to one specific class/setup is also much easier, as is designing any encounter. It's a 'safe' setup, both for the players and the devs.

I prefer to trade in some of that safety for player options and depth, but I'm in the minority on that one.
 
...the fact that we are fighting static PvE mobs means that there is always a mathematically optimal solution.

Well, an optimal solution for a particular encounter perhaps, but not necessarily an optimal build overall. If monsters varied more, then different builds might be able to handle different monsters better than others. That tank-mage might have trouble with a fire-resistant monster with armor-piercing lightning attacks.
 
If you wanted to be a high-end (back when there was a high end) tank, how many options did you have? One, prot warrior with an exact talent layout and exactly the same gear, using the exact same buffs/pots/flasks.

You should go and play some WoW, syncaine. All your information appears to be hopelessly outdated. There are several viable tank builds nowadays, I count at least 4.
 
Hence I said back in the day. Lower the difficulty bar far enough, and you could have a rogue tank and be fine. It still does not change anything, there is still a 'best' build, you just don't need that anymore because 99% of the content is 'accessible'.

Still waiting on the rest.
 
I still prefer the skill based system. While I agree that there is usually a FOTM skill set it usually only pertains to PVP and for PVE you can still mix and match your skills more liberally.

Despite the game being a flop I thought Matrix Online's skill system was really cool. Yes you had a cap on skill "points" but you could level them all up and swap them out for any combo you want. Im playing Fallen Earth right now which has no respecs and Im defintely missing that freedom.
 
A class-based system ensures that some players are gimped because the templates are fixed by designers. You pick a sub-optimal template, you aren't the best at anything. So, according to your "people will automatically optimize to healer, tank, dps," no one would play hybrid classes in MMORPGs, because hybrids can't do any one thing better than every other class.

Aside from that fact, your argument is based on the assumption that the default choice for a player is to optimize the hell out of their character. It's easy to see that less than 10% of players one-dimensionally optimize their build in any game if you consider the general lack of player knowledge in mass-appeal games like WoW. Players don't care anywhere near enough to figure out the optimizations.

I think the source of your distaste for skill-point systems arises from the ridiculously formulaic and simple combat rules in WoW. If there were actually more options available to the player in interacting with the environment, other players, and the enemies, a one-dimensional optimization would be overspecialization--you'd be ineffective in 99% of encounters but you'd be amazing in 1%. No one will choose that path over being viable in 60-80% of encounters.
 
Skill points will make sure that one build is best for dps, healing, tanking.

Classes are even worse. You've got one best class/build for optimum dps, healing, tanking. And it's not as simple to relevel a class then it is to change your points.

In the games I've played with a skill point system it usually doesn't matter that much. In dungeon siege one character would just get all his points dropped in tanking, one in nature magic, one in combat magic and one in archery. There's little incensive to make someone good at both archery and combat magic.
 
@Carra:
That is exactly the point. Because you cannot just relevel your lvl 80 char you do not to the cookie cutter flavor of the month specc.

Besides: The best raiding guilds available in WoW have all kinds of tanks. The differences are there, but they are too little to matter even for world class guilds. Most still use warriors although death knights are often better.
The differences are within a % or less.

In a skill point system people instantly reskilled to gain this %. But in a class based system they don't. Therefore there is much more variety.
If you think that the people are unhappy that they not have the perfect class right now, I can assure you: They don't.

The differences are too small to matter. The fun that comes from playing a different class and style dominates and btw: Balance shifts a little with every patch and whenever new content is added to the game.
 
I actually think that the amount of customization in WoW is pretty much optimal class-based but specializing via talents). I only wish that gear played a smaller role, but I guess that once the level cap is reached you need to have some way of increasing power.

Another issue with skill-based systems is that if you want to play optimally, you will generally go for a very specialized build. A class-based system will usually provide baseline classes that do include some diversity. For example, if I'm playing a warlock and I have to choose between, say, Water Breathing and +5 damage to my Shadow Bolt, I'm gonna take the Shadow Bolt buff every time. But give me water breathing as part of the whole Warlock package, and you bet I'm happy to have it.

You also get a larger range of different playing experiences, especially in combat. A WoW hunter plays very different from a WoW Warrior. But if you look at a game like GURPS (one of the ultimate examples of open-ended/skill-based PNP games), both basically work on the same principle of "roll to hit, roll defenses, roll damage", with very little in the way of "special moves" on either side. The GURPS warrior-type might decide to get fancy and hit his opponent on the hand, but he doesn't have anything along the lines of Sunder Armor, Rend, or Overpower.

Another example of open-ended/point-based PnP games is Mutants & Masterminds (superheroes). There, the issue is solved by putting in a general cap (based on overall power level) on your basic combat stats (attack bonus, damage bonus, defense, toughness), and providing enough points to easily reach those and have plenty of points left over for fun stuff.

Bah, now I feel I'm rambling on a bit. Anyway, very insightful post. Keep it up, cheerio, pip pip.
 
"I perceive "choice" as "work" and I don't play to work."

This is bizarre to me. Games are all about choice, play is all about experimentation and variation; if we didn't want to make choices, we're better off with a movie. The whole point of an interactive medium is making choices.

There is certainly a trend of "log in and mindlessly do your dailies", but to me, *that* is more like work rather than something fun and interesting. Funny how frame of mind changes things. To each their own, perhaps.

As to the stated topic, skill systems vs. class systems, I will always settle with the one that gives me more choices, the ability to change choices I don't like with minimal fuss and no cost, and that is the most fun to play and experiment with. In general, class systems are just too constrained, but there certainly are bad skill systems that don't work either.
 
I think someone said it already but I'm going to reaffirm the point that a big thing lacking from skill based MMOs that I've seen is flavor. WoW and Warhammer offer great flavor in terms of class differentiation even if it is only through cosmetic and lore differences, and I think certain people, myself included, crave these kinds of niceties. Whereas, in a skill based game like Darkfall, the nature of the system allows for large amounts of variation in the way you evolve your character through the skills you level, but not much in the way of flavor besides maybe racial differences. In the end everyone looks and feels samey, since developers of skill based MMOs can cheat on the flavor factor and make a few sets of armor and weapons that everyone can use. If everyone looks the same and has the same growth potential then I would have a hard time feeling invested in my character as a unique entity, which is weird because skill based MMOs are supposed to offer "near infinite" variation.
 
I would love to see a skill-point based game implement skill decay through lack of use. Why not explore the idea that a player builds their skill template to suit the way they play, and their skills normalise to reflect this.

Such a system need not decay based on time, but perhaps based on lack of use in combat. It could thus not penalise the time-limited casual.

What really bugs me is that it probably wouldn't be implemented simply because it emphasises the Role in Role Playing Game.

A Role implies specialisation; with Eve's train-anything approach, or Blizzard's dual-spec allowing some classes to spec both tank and healer, the specificity of Role has been thrown out the window. In its place is a weaker, flavourless substitute suitable for the masses.

Great for subscriber numbers, but often disappointing with respect to a game world's consistency.
 
@Flex: "A Role implies specialisation; with Eve's train-anything approach, or Blizzard's dual-spec allowing some classes to spec both tank and healer, the specificity of Role has been thrown out the window. In its place is a weaker, flavourless substitute suitable for the masses.

Great for subscriber numbers, but often disappointing with respect to a game world's consistency."

Can't speak for WoW, but in EVE's IP podpilots are a computer-enhanced immortal posthumans, in control of ships crewed by hundreds, thousands. They are the next best thing in the world to a god, limited only by the capabilities of the ships they fly.

EVE's skill training and class mechanics - time gating ships available, ships restricting scope - fit very well with the game world and the IP.
 
I think there's another dimension to this problem that you're leaving out. Even in WOW, you can have builds that are "optimal" only with a very high level of player skill. The old Affliction Warlock comes to mind (monitoring several DOT cooldowns and nukes) as a build that can only do top damage in the hands of a skilled player. At the other end of the spectrum you have builds like the old Shadowbolt spam Warlock. Mindless, easy mode dps on demand.

If we look at Darkfall, I'm certain that there must be skill combinations that are fairly easy to play and other combinations that require more planning/understanding of the game. So the "FOTM" build that works great for hardcore PVP'ers may, in fact, not be optimal at all for newer players.

If you play chess, it's equivalent to a new player learning Garry Kasparov's opening moves. They won't have any clue how to play the resulting positions and so they'd be much better off with a simple setup that kept them out of trouble.
 
You hit exactly what I don't like about skill based games. Deep down, in a place I try not to think about or admit I have, there is a munchkin living inside of me. If there are a bunch of options, I have the absolutely irresistible urge to spend hours researching the exact optimal way to distribute my points to get there. I can't help it. And so by providing "choice" I feel even more locked in than if I was given no choice at all.

Class-based games allow me to just relax and play the game more, just enjoy the experience rather than be given fake choices to make.
 
Having played quite a few talent tree games, diablo2 wow and a lot of the free ones on the net, and many of the more or less skill based ones morrowind, evercrack, etc. It occurs to me that the basic premise is usually correct, but not always.

I've ran across a few game that have multiple damage type with completely different damage registers other than the normal health/mana bars most games have.

when a game crosses into the areas of having say mental/body/blood/willpower stats for measuring "damage" calculations it opens up a whole new field of spell type and weaponry type options, these by their very nature multiply fotm builds because not one character type can "kill" everything, not one "healing" can heal everything.

the other thing i notice in one particular game that was interesting that all skills were left open for choosing reguardless of class with each skill having a different % modifier to turn experience into a skillpoint. this game also had a noticable lack of health or mana potions, this required the player to either choose to always play in a group, or to select a magic skill that could provide some healing. splitting choices up even farther.

the whole point of that was you could group play and get bosses really easy, or mix it up and solo and skill be ok in a party but excell at solo.

my point with all this being it is possible to design games that do not have a set fotm, however it requires a highly different take on the video game and any of the current ones in play that i know of.
 
Post a Comment

Links to this post:

Create a Link



<< Home
Newer›  ‹Older

  Powered by Blogger   Free Page Rank Tool