04 Jun 2017
I haven’t written anything on Dota in a very long time. (Life is busy.) However, I’ve been meaning to comment on the changes in the Dota 7.00 update. As many players have already said, it is probably the biggest change to Dota 2 since it launched. There have been many important changes to the game, but I think the largest mechanic change has been the addition of Talents.
Talents - Do they improve gameplay?
Now that everyone has been playing for months with talents, it’s hard to imagine Dota without them. Talents have definitely added new and exciting elements to the game. Cooldowns reduced to a few seconds. Enhancements of key hero powers. Who can resist waiting for your own hero’s level 25 power spike? Certainly, some heroes have almost acquired new roles, with some supports acquiring heavy late-game damage abilities. Not only are there new abilities, but game-wide variables are being adjusted like respawn reductions (keep reading), additional gold per minute, and additional experience gain.
However, is this good for the game?
Talents - Concerns
My general opinion is that talents should not have been introduced. My main concerns are that some game variables shouldn’t be manipulated, and that talents make a game that is already notoriously difficult to learn, even harder. Assuming they can be balanced, talents are an inelegant game mechanic that could be better solved in other ways.
Changing variables that shouldn’t be changed
Most talents are a bonus to stats (str, agi, int) or hero variables (move speed, attack speed, range). However, one of the new aspects of talents is that they adjust game-wide variables, such as respawn rates, gold per minute, and xp per minute. Balancing these variables will be very difficult, and understanding these effects will be hard to see. Is a support hero stats or abilities too weak, or should they be expected to choose a gold or xp talent?
Adjusting variables like respawn rates eventually led to the imbalance in even pro matches. As a strong example, see Lina’s 79% win rate at the Dota Asia Championships. Thankfully, Icefrog has responded, and as of the 7.05 update, all of the respawn talents have been removed.
Making a steep learning curve even steeper
I started playing Dota in 2016, so I’m a relative newcomer. One very important topic I plan to write more about is the amount of knowledge to play Dota at an even basic level. The addition of talents now make new players have to acquire the knowledge of not only 140+ items, 110+ heroes, their 4 unique abilities, their unique playstyles, their surprising interactions with other heroes, but now also their unique talent trees with an additional 8 arbitrary variables. While some level 25 powers are rarely seen, other talents like Axe’s +75 damage at level 15 are generally too influential to not be aware of.
Because talent trees are specific to each hero, this is an additional 8 x 110 variables a new player needs to be aware of. When you play one hero it is not a big deal, but having to be aware of what talents your opponents and team members are choosing is important. If Dota takes 1000 games to become an intermediate player, with talents it now might take 1100 games.
Talents - an inelegant game mechanic
As a person interested in game design, the addition of talents seems like an inelegant way to add a new element to the game. From my understanding, talents were originally seen in Heroes of the Storm (HotS). In HotS, talents were a substitute for the complex item system that Dota and League of Legends have. HotS simplifies the economic side (or vertical power growth) to experience, i.e. there is no gold. In HotS, the way you adjust your situational hero build is by choosing different talents. In Dota, items (and gold) would theoretically play this role.
While there is a LOT to learn about for items, items are at least a general concept that applies to all heroes. Knowing how Radiance works with illusions allows you to apply that knowledge on all heroes, importantly Naga Siren or Alchemist.
Talents can be used to bluntly adjust a hero’s power curve. While I have full confidence in Icefrog and his team’s game balancing abilities, talents does take away from the emergent gameplay that Dota embodies. Talents can turn supports into semi-carries in a fairly arbitrary manner (Lich +120 damage, Techies +250 damage, etc.). They are not subtle linear improvements. Rather, they are often major alterations. Most of all, talents are not balanced on a horizontal basis between heroes. Some heroes have huge level 15 talents, while others have very modest bonuses. The talent modifications seem very arbitrary, and heavy-handed.
Talents - Already used to it?
Now that we’re already on Dota 7.06c, I strongly don’t think anyone at Valve is seriously considering removing talents. Talents are here to stay. However, I think it’s worth thinking about Dota before talents happened, and the direction of how the game is evolving. Dota is an amazing, amazing game. However, it suffers from high complexity, and the addition of any needless complexity should be avoided if possible. I’ll post more ideas on that later.
22 Oct 2016
Like all Dota players, I am always excited to see the release of an update to the game. Recently, the 6.88f update was released, and I’m sure it will be a step forward towards making an already amazingly balanced game even better.
As a person interested in game design, I’m always curious to think about how Icefrog approaches the question of hero balance. There are always numerous articles written on which heroes need to be buffed or nerfed (an example from Dotabuff).
I wanted to share some of my own thoughts and suggestions. I’ll fully admit that I’m not a very experienced a Dota player, and I don’t claim to have an expert opinion on every character’s exact gameplay. However, I think there should be general principles that game designers should pay attention to to decide which heroes to buff and nerf – and importantly how.
General Principles For Hero Balance
- Some level of equity in play rates is a goal (“hero diversity”)
- Relative popularity and power influences the type of buff or nerf
- Adjust for experienced players (but take into consideration new players)
1. Some level of equity in play rates is a goal (“hero diversity”)
I think all heroes should get some roughly equal level of picks.
To take an extreme case, a situation where one hero is picked 90% of the games, while other heroes are seen less than 1% of the time is clearly a negative outcome. The game could theoretically still be fairly balanced (win rates of 50%), but the game itself would suffer from a lack of variety. Hero diversity is important because it supports a wide range individual playstyles, and as a direct result, a wide range of team playstyles. This is truly one of the best aspects of Dota.
With 112 heroes, perfect equality in pick rates would be 8.9%. However, perfect equality is also not desirable, and there are many reasons for that (notably would require a rigid breakdown of heroes and their roles, support, carry, etc.). However, intuitively we should be able to say that a hero should not be picked 90 times more than another hero.
My sense is that the inequality in pick/ban rates for the heroes in 6.88 is still not equal enough.
Most picked heroes:
Least picked heroes:
Heroes like Pudge are picked in more than 35% of high level games, while the bottom 10 heroes are getting picked less than 2% of the time. This strikes me as wrong.
I think the designers should target closing the gap between the “richest” and “poorest” heroes.
There’s no scientific rule for the right ratio, but my sense is that heroes should not be picked more than ~5 times other heroes. This would mean the pick rate gap would be something closer to the least popular heroes being closer to 4% and the post popular heroes at 20%. Based on the 8.9% pure equality rate, this would imply that you’re likely to encounter some heroes twice as often, and other heroes half as often.
2. Relative popularity and power influences the type of buff or nerf
I think the two key factors involved in Hero Balance should be popularity and win rate. This might be obvious, but I think there are some nuances worth noting.
If you created a 2-factor graph, you could put heroes in different quadrants.
- Quadrant 1: Popular and High Win Rate
- Quadrant 2: Unpopular and High Win Rate
- Quadrant 3: Unpopular and Low Win Rate
- Quadrant 4: Popular and Low Win Rate
How to handle the heroes in Quadrant 1 and Quadrant 3 are fairly obvious:
- Quadrant 1: Popular and High Win Rate: Nerf
- Quadrant 3: Unpopular and Low Win Rate: Buff
These are the two situations that should be the easiest to handle. Hheroes like Outworld Devourer and Omniknight should be nerfed. Likewise, most of the heroes at the bottom end could use significant buffs, like Gyrocopter and Death Prophet. I think it’s fair to assume that there is a general tendency for heroes that get stronger to also get picked more often.
How to handle the heroes in Quadrant 2 and Quadrant 4 are less obvious:
- Quadrant 2: Unpopular and High Win Rate: Increase popularity, but reduce strength
- Quadrant 4: Popular and Low Win Rate: Increase strength, but reduce popularity
How to adjust the two factors?
Adjusting Win Rate (raising or lowering a hero’s strength) is relatively simple. There are many ways to achieve this, but the easiest thing is simply to buff or nerf the hero’s powers or stats. Buffs and nerfs should continue until the hero’s win-rate is within the boundaries (48%-52% as a suggestion).
Adjusting Popularity (unpopular to popular, or vice-versa) is more difficult. However, there are definitely gameplay elements that could be adjusted to make heros more attractive to play. Notably, heroes can be made more popular by making them more friendly to use, or reducing the level of micro required. Here are some possible examples:
- Reducing complexity on the effects of abilities (different effects on allies, enemies, or has potential negative effects - Example: Oracle, Chen)
- Changing an ability to be auto-cast
- Changing an ability to be area targeting or area of effect, rather than unit targeting
Likewise, to reduce the popularity of a hero, do the opposite.
Adjusting from the extremes
Adjusting Which Abilities - It’s also worth considering to start buffing a weak hero’s least used power. This is a way to cause major balance issues. Similarly to having hero diversity, it’s also more likely to lead to greater diversity in heroes builds over time. As powers within a hero become more balanced, there is more space for great players to build according to the game and situation, rather than follow a set path.
Adjusting Which Heroes - It’s also worth looking at the extremes for the heroes. Heroes that have close to 8.9% pick rates and 50% win rates can arguably be ignored. It’s the heroes on the edges (top 10, and bottom 20) that badly need the adjustment.
3. Adjust for experienced players (but take into consideration new players)
Last note, the game should be balanced for the professionals. These are players that should know how to play the game to the fullest. As a proxy, I am looking at the statistics from Dotabuff, and specifically at the 5000+ MMR group. These players should know the game, and their play should not be radically different from professionals.
What about low-level players? While there are definitely low-level pubstomp heroes that could receive buffs in this situation, this is theoretically something that players would learn to handle over time. What is not acceptable is the situation of many heroes even in the 5000 MMR bracket that have 1% pick rates and <48% win rates. This is not a situation where the player and the opponents do not know how to use the hero or how to counter it. This is where the game could receive the best returns on adjusting the balance.
Thanks for reading. Again, I am not an expert on Dota. I hope this generates some ideas and some discussions.