Wednesday, July 8, 2009

Balancing with Tiers

Melf_Himself and I have been having a discussion in the comments of one of my DotA MMO entries.
I've not played DotA, but it's a bad idea to include a large percentage of classes that would never be chosen by competitive players. 10% is a very low number and is much lower than what you'll see in well balanced fighting games. It really makes things hard for new players, making the amount of information they have to possess before they can start playing the actual game a lot higher. Ideally any choice that you make before a match starts should not affect the likelihood that you will win the match, only how much fun you will have.

I'm not sure what you mean by "painstakingly difficult to balance because there are so few". The more classes there are, the harder it is to balance them all.

I draw a lot of my design and game-playing principles from David Sirlin. Not only his Playing to Win articles, but also his Multiplayer Game sections.

DotA has over 90 heroes. I would say about 20 (22%) of them are viable for competitive play. My favorite fighting game of all time is Naruto GNT 4 and only 9 out of 25 (36%) characters are viable competitively. A more Western SSB:Melee has 5 characters out of 26 (19%) at the top tier.

DotA isn't a very newbie friendly game, partially because of the reason you described with tiered heroes. That doesn't mean DotA isn't fun or isn't competitive--quite the contrary.

If DotA had fewer heroes, say 10, then all of them need to be viable. That is much more difficult than making 50 heroes and letting the players figure out which ones are the best. The designer obviously needs to make sure that none are dominant and none are dominated, and over time he can try to pull their differences in power closer together. But with "tiered-balancing" the success of the game isn't dependent on the viability of every character.

You are quite right when you say that choices made before the match shouldn't affect your chances to win, which is why MMOs are very difficult to balance. Every class needs to be viable because players have invested resources into the character and are locked in to that choice. Compare this to DotA or a fighting game where players can see what character their opponent (or team) has chosen and react appropriately. I love drafting in DotA and MTG for this very reason. There is a game that takes place before the game starts.

Many moons ago when Blizzard announced that switching gear in Arena was going to be disallowed, I was heartbroken. I have never played Arena, but here was a counter to team composition, and they completely removed it! They had valid reasons for doing that, but it removed an entire level of depth to Arena PvP. Players were not locked into their stat choices and could react appropriately to the opposition.

So for the DotA MMO, we can have hundreds of abilities and scores of ability groups. Allow the player to react to his opposition (either through the meta-game or with a pre-match draft system). Abilities and builds will fall into tiers (WoW talent builds are tiered, by the way). As long as no one combination is dominant, and players have counters to abilities, builds, and team compositions, then the game will be fun and competitive.


Tesh said...

MTG drafting works nicely as a skill tester, but you also only wind up playing with your choices for a couple of hours. Not so with an MMO, which is why you either need to keep balance very tight, or allow players to freely and easily change when they find their particular build is broken or gimped.

Melf_Himself said...

Sirlin is one of my favourite writers on game design. I don't recall (nor can I see from a quick skim) anything he wrote detailing the actual percentage of characters that should be competitive. If you read his series on how he balanced SF2 Turbo HD remix, it sounds like the majority of them are tournament viable.

I think anything less than 50% sounds very low. Your method for determining the best characters (let the players sort it out) sounds good if you have a large number of players of varying skill levels during the alpha testing of your game. After that, if you don't want to spend time tweaking the balance then you should just cut those characters from the game.

Otherwise the thing is just less accessible in exchange for no real advantage (and it took you a lot of development time to make those extra 60-odd weak classes that nobody uses). The ridiculous number of classes etc available in DotA contribute to its lack of accessibility for new players, which is the biggest criticism I've seen leveled at the game.

I'm a big fan of not needing a diploma in forum research for a new game before becoming competitive. Anything that you have to learn that could only ever apply to one specific game is undesirable. Skills that you can take between games (whether it's aiming skills, micromanagement, a tactical mind, reading people i.e. "yomi", etc) are much more interesting.

I'm not sure about DotA, but in most fighting games you can not see what character your opponent chooses before the match starts. This gives a disadvantage to whoever picks first. Sirlin even lamented that this was not the case in his review of SF4 I believe.

The draft situation in MtG is different because there are such a large number of cards that will be chosen, so the disadvantage is minimal.

The 'skill drafting' situation you suggest for the DotA game sounds attractive. Although you really want to make sure that part of the game is as fun as possible - people might get bored if they have to wait 10 minutes before a game that lasts for 20 minutes, for example (I have no idea how long DotA games go for, but you get the idea). You might consider the 'drafting' actually occurring inside the game, as the game progresses (isn't that kind of how DotA plays out anyway? You choose what to power up based on what your opponent is doing).

That's all the time we have for today, but for your next post maybe we should discuss the issue of class vs skill based systems ;)

motstandet said...

Eric Heimburg posted a wonderful article on his blog today:

There is one passage I'd like to highlight:

"Fortunately, I had a decade of engineering experience and understood how to tune complex systems. I wrote analyzers, modeled usage patterns, and made corrections.

Unfortunately, my approach did not take the “human equation” into consideration very well.

I found that the Feral Intendant class was 30% overpowered, and that’s why so many people were playing a Feral Intendant. Yet somehow, reducing the power of the Feral Intendant to the correct level did not suddenly make the game more fun… thousands of players were complaining and nobody was telling me they were happy about the change. Weird! I double checked my calculations. They were correct. So what had gone wrong?

Turns out that the people who played the other classes available to that race had taken on an “underdog” mentality. The people who played Claw Bearers liked that they were woefully underpowered compared to Feral Intendants. It was like playing the game on Hard Mode. And the people playing Feral Intendants liked playing on Easy Mode. In balancing the game I had failed to understand the needs of the people playing it. I just ham-handedly fixed the equations, instead of solving the problem with the finesse it needed. It was one of my more serious missteps. (And it’s a great example because I think it’s pretty obvious in hindsight. Most mistakes were much more subtle.)"

The point I was making by posting those percentages was that here are either very popular or very competitive (or both) games where not every option presented to the player is balanced. There is enough variety of play that every character doesn't need to be competitive or optimal.

And there are times where players will pick lower-tier characters because they want a challenge, some fun, or a few laughs, as Heimburg points out.

With regards to DotA, learning the game is a massive undertaking. I've been lucky enough to have played it through the years as it has grown and switched designer hands. But I frequently have to coach friends new to the game; every thing from heroes, to items, to team composition, to map control, to proper micro.

DotA is complex. This is why so many players love it. Mastering a single hero can take weeks. Then you can try a new one. I agree with you that the casual gaming market (a force to be reckoned with) wouldn't like this game at all.

Some technical details: teams of 5; a standard pick-up game lasts 40-60 mins; an "easy mode" (-em) game lasts 25-45 mins; the game mode I am talking about is random draft: 22 heroes randomly selected, teams alternate selection with players "drafting" their hero in sequence; league drafts could take up to 10 mins.

I'm thinking about picking it up again. Perhaps I can help you ease into the game ;)

Melf_Himself said...

Oh, I know some people prefer games that are quite complex, it's always a trade-off. I used to prefer them that way, but after learning the intricacies of all the DnD RPG's with various rulesets, several cycles of Magic expansions, Diablo 2, and Guild Wars, I'm kind of a bit burnt out on the whole process. I have to know that something's going to be really special before I do that again.

I guess it all depends on what you think the audience of your game would be like. If you're trying to make a first game to garner some widespread internet popularity and make more cash on your first game, an accessible game is probably the way to go. If on the other hand you know you can hype up a reasonably dedicated hardcore audience like the DotA crowd is then that could be more rewarding. I suppose the Source modding community does get some good attention so there's some good potential there.

That stuff about data mining is kind of a separate issue I guess. I think designers really just need to play the game and watch other players play the game rather than rely too much on numbers and formulas... it's more of an art than a science.

I am semi-tempted to learn about DotA, but I secretly hated WC3. Perhaps if you cover your falling in love with the game all over again here on the blog it will inspire me to jump in with you ^^