NetGore

The main reason I would do free-form character building is simply because there is less content required. If you think about it, if you create 10 skills, free-form would allow 10 skills to choose from and use, however if you have 4 classes you would have to create 40 skills to provide the same amount of content.

This is also extends into things like class-based quests and so forth.

The advantage classes provide is depth in PvP and PvE, a party of all-round characters is just plain boring and requires no skill or strategy. However there is also such thing as too many classes.

In terms of lore, the easy way out of this is just to have less emphasis on class/race ability and more emphasis on class/race lore.

I don't think it would be too daunting at all. Just make sure you always provide awareness to the players on recommended skills, or allow them to reset stats. Just look at runescape, there are a ton of idiots on there and none of them seem to be having any trouble with free-form.

I'm more concerned about balance and multiple paths of viability, though, than the amount of "content", although I do see your point about the number of total skills being reduced in a freeform system, since redundancy is eliminated. Problem is, I'm thinking the time it takes to balance a freeform system far exceeds the time it would take to implement the increased number of abilities in a class-based system.

Still trying to figure out how to tackle this. I'm thinking tying stats/attributes (like strength, dexterity, etc.) very closely to efficacy of skills will help to create "soft classes" within the system. Essentially, you can try to be a mixed bag of tricks, but in doing so, your stats will be so spread out that you won't be as effective as those who specialize for certain situations/roles. Hmm...

You need contention between skill selection. When a player has to make a choice, what creates builds & counter builds is the contention, "what do i have to compensate for by selecting one ability over another". Its even better when both are equally as good but cater for different play styles, which means what is very strong for one player may not be for another, but then the other player might choose the other skill which works better for him and thus it becomes balanced.

This even transcends onto skill usage, for instance doing a nuke combo may temporarily make you vulnerable to a snipe by a caster. In certain situations using this would be overpowered, in other situations it may be the complete opposite. This then adds to the complexity of build order and contention.

Of course you would have to make obvious what type of player he/she is. So maybe casters should require equipment like robes and so forth so you can identify them, same as coats for rangers or something. I think the biggest problem with freeform would be intel, trying to find out the skill set of the other player so you don't end up doing anything stupid.

Sure, I think the theory behind the "contention between skill selection" is simple and logical conceptually. But in practice, it can be rough. Mathematically, let's say you have 20 total skills/abilities (a pretty small number, I'd say) and you only allow a PC to choose 5 of those abilities total (again small). The total number of permutations is:

20 * 19 * 18 * 17 * 16 = 1,860,480 permutations of power choice combination

Now, let's say you have 10 different classes, each with 8 possible abilities, again only allowing 5 choices total. That total number becomes:

8 * 7 * 6 * 5 * 4 * 10 (total classes) = 67,200 permutations

That's a huge difference in the number of possible builds that you'd have to try to balance against each other. And that's just from a statistics perspective. Add to that some other context, such as duplicate (or, at least, very similar) abilities between classes, and the more realistic number of "different" class-based builds is likely smaller. And this is in a comparison where you have 80 total abilities for a class-based system, and only 20 choices for a free-form system. I realize you can't go strictly off of numbers, but I still think it's a massive undertaking to balance a free-form system vs. a class-based system.

The only way I can see it being less daunting is to tie skill/ability efficacy very closely to base statistics/attributes (Strength, Constitution, etc.). If you give them limited development in that way (i.e., you can't max out all of your attributes, so you have to pick which ones to focus on), then you effectively create "soft classes" by forcing that "contention" between versatility and specialization.

Hmm, maybe I'll do a free-form system after all...

Yeah soft classes sound like a good idea, however limiting players with statistics attributes means you may as well have classes, i find stat requirements the most mundane form of specialisation as well. I think skill trees would be a better idea, however I find that most games that do this bring the level of contention basically down to "do i want to be strong now, or later?", which is really dumb imo. If there is leveling and PvP you will always opt for the late game passives tank build (torchlight comes to mind).

So i think things to look out for is fake contention, the illusion of contention when there really is none once you understand the game mechanics (i mean you want to avoid this). Skill trees simplify contention but balancing becomes recursively difficult the more tiers there are. Level requirements also simplify contention. I guess those things can make it easier for yourself when it comes to balancing.

Using maths for balancing in my experience has never worked for games. I remember when the new macro mechanics came out in SC2 for protoss, and some guy used maths to try and prove that the proton charge ability was completely imba at late game. Basically it boosts the mining capacity of probes in an area, so the rate of return is exponential depending on saturation. However he was completely wrong because there is a max saturation capacity, where having more probes mining after that point has a detrimental effect on your mining capacity (the ability has been removed fyi anyway). This happens all the time because of nerds and their spare time.

Instead using live testing has worked much better. You never know what is strong or weak until you see someone abuse it. Often some overpowered skills are completely overshadowed by other skills. The typical caster snipes tank paradox comes to mind. The next best thing is to just theorycraft.

well, you will definetely code some form of stat & skill resetting,

even if its for tests only.

have your testers test out pvp and pve extensively, arm them with unlimited resets for skills & stats.
this way, they can try anything that comes to their mind.

sure you wont have em test all there is, but a lot of builds!

highend imbalanced characters are mostly some abilities that can stack effects together.
but this can easily be reduced, or have this particular duo-buff weakened etc, non stackable, etc etc

WHAT I SUGGEST:
dont think away from classes.

you want your players to have freedom.
skill system => form your own class

wait a second there... "form your own class"

thats the sentenc eyou should think about imo.

in aleron, skills are bundled in skill sets of 4 (if you learn the set, you learn all 4 skills after your skillset reaches the lvl)
if you chose a fire magic skill set, you maybe get 5 intelligence per level.
if you select a sword skillset, you maybe get 3 str and 2 dex per level.

additionally you can add bonus stats in some ways, of course.
we have a total number of skillsets.
the player will build his own class together.
=> own class, easier balance as skills are bundled in 4, low level and high level use one skillset. so keep these skills at one usefulness level

hmm it actually has many pros.
we have a good game designer on that task

I originally wanted to do the "no classes" system, and allow people to be as flexible as they want. Though I decided against it in the end. It works much better in offline games than online games. In offline games, you have much more flexibility in how you change the world with your actions. More importantly, offline games are mostly solo games. So when you replay the game, you can take a different build, go for a "good citizen" instead of "ruthless murderer", etc.

With classes, you force people to play a certain way. While it is more restricting, it forces them to have a unique experience with each class (as long as you implement them well).

Though I do think going for the "no class" system as an indie developer is a good idea if you are up for it. It will definitely be more challenging (there is a reason why most all commercial MMOs don't use it, just like raping the fantasy theme to death) since you don't have many games to take ideas from. But if you try to make your indie game like the commercial ones, you are just going to end up failing big-time.

I'm planning on a combo of all of those things. Skill trees with dependencies (and being able to invest multiple points into a skill to increase efficacy), attributes that affect skill potency (which I think is pretty standard, otherwise why have attributes at all?), and possible level restrictions as well (at least for initially opening up higher-tier abilities, along with the dependencies on lower-tier abilities). I don't think there's anything wrong with using attribute reliance as a way to encourage/reward specialization in a more open development system. Something like "Intelligence provides a bonus to your spell damage" or "Dexterity increases the evasion bonus of your stealth skills". If anything, it gives meaning and value to those attributes, when a lot of games seem to downplay the importance of those attributes.

And I totally agree with Spodi that a classless system is much easier (i.e., more lenient) in single player games. Like I said, it's not "realistic" to expect all options to be equitably viable, but in a game setting, players demand an even playing field, and rightfully so.

if you have imbalance, it is "ok" if it is minor...

lets say this one class has some pros against many mobs and thus, can level faster...
but therefore his overall pvp is more difficult

definetely make your game SKILL DEPENDANT!
else its JUST balance on your side via spels vs spells.

if you have a good knight, let him use rush skills, have dynamic movement etc overall

keep action and thus set strategy in battle over skills.

so if your players stand 1on1 and only use spells, balance is CRITICAL

if a skill is too strong, but the player must6 aim good with it or so,
its slightly different. balance will not STING your game as hard.

diablo did have many imbalances due to the "endlessly high skill levels"

but up to a specific point it was quite ok.
if the game werent so dynamic in running around and aiming etc,
balance would be much more critical.

hope you get my point. if it is "stand there and use skills", skills is all that count.
else its running around, aiming etc, and balance will not be as CRITICAL as with "stand and type" battle systems.

thus the chances of you balancing out sufficiently is increased.

just a hint in that direction.

and yes, a hybrid (stats and skills affect eachother etc) is easier to balance out.

I think whatever happens, every skill should matter. I don't know how many games I've played (flyff, cabal) that have useless skills. (flyff especially) Whenever there is a skill, I expect it to be useful. Some skills are dwarfed by others (and they aren't in the same skill tree). This really has to deal with Contention (as bake pointed out).

To sum it all up, every skill should be unique and have a purpose. They should be dwarfed by other skills (unless their in the same skill tree). Why would I want to learn a series of skills in one skill path that are dwarfed by the "Super insane Slash" in another skill tree? That's just my opinion on skills.

My game team discussed (again) this thing some days ago. Our game have.. like, two types of class, like race and class (not really, our is nen and hunter type, because we're doing a hunter x hunter fan game, but for those who don't know the anime...), and each of these type of class will have different skill trees. But the skill trees of the two types of class will depend one with another, but just a little.
A type of class will have determined skills, but on the other type, depending on what class you choose to be, you can use some skills from classes related to yours too.

But the thing is that it will be impossible to keep 100% balanced, and we will need to make too much spells too. But because we are basing our game on an anime, we need to do it D:.

It looks like overly complicated too, but it's "just" a bunch of skills that are unlocked as you put points on another, like maplestory, and the player don't even need to understand it.

I know, it's not really good designed, and for the balancing we will need to receive a lot of feedback from the players. I hope they don't start cursing our mothers before we have time to fix the imbalances, and that we aren't dreaming to high.

if there is no "one build/class to rule them all" its actually okay.

if mages are better against normal enemies and assasins are better at pvp, thats okay.
but dont have any extremes there...

I have a game system already made and tested that is skill based, I'm kinda hoping that if there's any small game tutorials there's one with a basic skill system as well, even if it has 1 skill, that's cool.

I also have "classes" in a way. Basically, some skills are associated with your class, if your characters decides to be that class, and all it means is those skills go up a little faster than others and they have access to certain abilities linked to that skill. For example, anyone can learn the skill 'Block' but a "Warrior" can learn how to effectively 'Bash' with their shield when 'Block' gets to a certain number... just an example, I'm not looking for philosophy on if you should be able to bash if you can block ... I know how threads can go