I was asked in this thread to comment on our balance decisions and how we approach balance in Path of Exile. I started writing up a reply, which turned out quite long before I even covered my first point, so I will make my comment in a multi-part post.
This first part covers the general balance philosophies of the game. In the next part I will talk about the passive tree. After that, I’d be keen to talk about progression, and balance for specific playstyles. Later on, I’d like to get into the specific numbers, and balance of specific game stats.
The free to play games that have been successful (like League of Legends or MapleStory, for example) and the games with online persistent economies (for example Eve Online or World of Warcraft), have been ones that have evolved over time. They have regular updates, additional content, and an evolving metagame. These changes keeps players involved in the game.
This is the kind of success we are interested in, so we do intend to add more content, more skills, and more keystones. We also expect the balance to change, with different elements of the game waxing and waning in strength over time.
We want decisions about your character build to really matter. Part of this stems from the game being, to a very strong extent, about items. For them to be important, the items your character wears need to matter. We don't want items to be strictly tiered and transitive in nature. In many games, you want the gear for your class, and then you want to keep upgrading the tier of those items, until you get the maximum tier item for your character. While this does work for those very character-centred games, in Path of Exile we want the items to take more of the limelight. Items are not just supporting pieces to a class progression you walk through.
We want items to have intransitive relationships, as well as transitive ones. Intransitive means that the items are not comparable by strictly looking for bigger numbers. Rock-paper-scissors is a game with intransitive relationships between the moves, as opposed to a game where a bigger number always wins. So, for intransitive relationships between gear, sometimes we will want an item with smaller numbers for specific purposes. The answer I most like to see to the question “Which of these two items should I wear?” is “It depends.”
If the gear your character uses matters, then how you build your character matters. We also want character build decisions to matter, because it fits the style of game we like to play.
This leads to few things that impact how we handle balance:
(1) Decisions matter. You can make your character better by making better decisions about its build. As this is possible, you make your character worse by the decisions you make. So, in this case, balance isn't about making everything equal. Having decisions matter means that randomly built characters tend to be much weaker than ones that are well thought out. (Although, it should be noted that we want to make it easy for people to see what they need for their character, and have early paths of strength obvious).
(2) Synergistic options. These create a great deal of potential for making a good character by choice. The game system has to allow for different things to work together in positive ways. This makes balance more difficult, as things cannot be balanced in a vacuum.
(3) Build versatility and emergence. We want players to have lots of tools available to them, and we want them to be able to be combined in different ways. A player has access to gear stats, skills, supports, keystones and flasks, all with interacting elements. This gives a lot of flexibility, and leads to players being able to discover emergent things that had not been considered. We design to make this possible. We try to make things in a way that allows for interesting things to happen. Things would be much easier to balance if we restricted players’ options.
As the game has this kind of emergence, many of our balance decisions have to be reactive. We have to react to balance problems that crop up due to how the game is played. This is not the only way we balance, but it has to be a factor. If something becomes too dominant, it has to be addressed.
An example of an emergent thing we didn’t expect making the game better would something like Ruby Flasks being used as a moderator for Righteous Fire characters. An example of something that needed to be balanced after being introduced would be the first versions of multistrike (and multicast), that dominated the game immediately after being added.
On the flipside of this, we sometimes balance things based on their perceived weakness, even though we know they are actually much stronger. We have to be careful with this, though, as it can come back to hurt us (and lead to reactive nerfs as the power gets discovered).
An example of us not buffing something "known to be weak" is Eldritch Battery. We had players demanding this be strengthened for a long time, but we did not change it, as we had seen it being used well. In the last few months, the community view of the power of this Keystone has basically reversed with no change to it. Burn damage also fits this; we were told for a long time that it was the worst status ailment, and that it needed to be buffed. We knew it was strong, and did not change it. Over time, it has gone from being considered very weak to very strong without us changing it. All that changed was the discovery of its usefulness.
(4) Build-based characters, not class-based. The synergistic options and versatility and decisions mattering are things we value. They don't make things that easy. The freedom players have in making their characters makes it very easy for players to converge on imbalances in the game. Very slight imbalances between options can result in characters gravitating to very similar builds very quickly. We work to avoid that, but our players are very canny. We are sometimes told that there are only a very few builds that work in high-level play, but when we look at high-level characters, there are more builds represented here than in many similar games. However, when players do converge on similar options, we need to make changes that can be less than popular.
Our players usually start our game thinking in terms of the class they choose first, and then work out what builds they can do. After playing for a while, they start to think in terms of build, and then work out which class best supports the build they want to make. Because of these two styles of play, it changes how we approach balance changes to the passive tree.
I’ll talk more about passive tree balance and development in a followup post.