Sunday, March 4, 2012

Hero Academy First Turn Advantage

The advantage of going first in Hero Academy has been on my mind for the last week or so.  The topic has come up on the Robot Entertainment forums with a few solutions proposed.  I've been brainstorming up some solutions and would like to share some today.

For those who are not aware, there is a First Turn Advantage (FTA) for going first in Hero Academy.  Many games have this problem and different games have come up with different solutions to it.  In Hero Academy the person who goes first is always 5 actions ahead.  This 5 action advantage is more noticeable on some maps than other and is also more noticeable at higher levels of play.

The single large crystal map has a particularly strong FTA .  This makes sense, because the relevance of the FTA goes up for short games, and the single large crystal map usually has shorter games. Here is a worst-case scenario (but unfortunately not atypical) shot of a game I had this week.  My opening hand had two Knights which I used to secure the two purple assault tiles very quickly.  To make matters worse my opponent is playing Dwarves and has no ability to move my Knight (such as with Knight knockback or Impaler tug).

The map is bursty already - and the FTA exacerbates the issue.

With both assault tiles secured, I just start bursting down his crystal.  In this shot you can see I'm even using a priest now to attack the tile, who can deal a whopping 800 damage per attack thanks to control of both Assault Tiles.  On the subsequent turns I follow up with my Knight, Archer and Ninja -combined with the health potion and the scroll - I can charge straight at my opponent's crystal with reckless abandon.  My opponent clearly has a dominating army on the board, and would win in a comparable situation on any other map, but the combination of the FTA and this map are overwhelming.

So that's the problem - but what's the solution?  Some of the most challenging design problems are those with a large number of solutions, but every solution has a different drawback.  The job of the designer becomes choosing the solution that best addresses the problem while being in line with his or her overall vision for the game.


Starting Player has a Smaller Initial Hand
You could start the first player with only 2 units in their hand.  This heavily restricts the options they have for the first turn
Pros: Might actually make the new player experience easier and more streamlined.
Cons: May not actually solve the problem, 5 actions is 5 actions and some games I play I spend all of my first 5 actions getting 1-2 units into position anyway.

Starting Player Deploys from a Single Deploy Zone
You could restrict the first player to only deploying on their first turn from a single Deploy Zone.
Pros: Simple.
Cons: Might actually be confusing to new players.  May not actually solve the problem.


Second Player can Mulligan
A "mulligan" is where you reject your starting hand and draw a new one.  In Hero Academy terms, you could allow the second player have an opportunity to discard any number of cards from starting hand and then draw back up to full before beginning turn.  The first player would have the action advantage but the second player would be minimizing the effect of bad draws

Pros: Asymmetrical advantages can be pretty cool. Some players really hate RNG so may enjoy going second.
Cons: Asymmetrical advantages are harder to tune.  This solution is binary, if it fails to solve the FTA there is no granularity on the tuning knob, you'll basically have to go with one of the other solutions anyway.  Introducing a solution that basically remove RNG for one player isn't very sexy.  Having players choose between a concrete action advantage vs. not being screwed over by their deck is not an appealing choice. Increased burden for teaching new players.

Manipulate Starting Hands to Favor Second Player
One subtle technique that games that are player vs AI (or player vs. environment) often do is manipulate the random number generator to favor particular outcomes. For example, the first player might have a reduced (or zero) chance of getting swords in their starting hand, while the player who goes second might be guaranteed a sword in their starting hand.

Pros: Could probably manipulate the RNG to get 50%.  New players will never know.
Cons: Feels icky.  Betrays the trust of the players who are expecting true RNG. Part of the longevity and appeal of Hero Academy comes from varying starting hands being a random element that players have to accommodate.

Fixed Starting Hands
A variant of manipulating via RNG - each race could have a pre-determined starting hand, with the second player's starting hand being better.

Pro: Good deal of granularity for tuning to 50% win rates.  A really nice side benefit is the New player experience might actually be better.  Starting hands that are easy to understand could also be accompanied by tutorials that are designed to match the starting hand.
Con: The randomness of the draw is part of replay-ability formula - could lead to many games feeling similar to one another.  This may be mitigated in the future by having a large board pool rotation and lots of available races.

Extra Item for Second Player
This is a variant of manipulating the deck but does it a clear way so the hardcore players know it's happening. Since the deck contents are set, you could throw an extra sword, shield, or scroll into the second player's deck.

Pros: If you choose something like a scroll - that's probably equivalent to the FTA.  This modification does not affect the new player experience at all.
Cons: Is very hidden from the player, (unless Robot decided to show deck contents).  Feels like a less elegant solution (though I will admit I'm having trouble verbalizing why right now).

Add a Deploy Phase at Game Start
Much like Warhammer, you could begin with a deploy phase.  Rather than take their normal turn, the game starts with a deploy phase where players can place a few initial units directly onto the battlefield.

The design team could play around with the deployment rules.  For example, no super units - perhaps the player who goes second is allowed 1 extra deploy, or one extra unit to choose from.  You could declare that the starting player can only deploy units in the first 2 columns but the second player can deploy units in the first 3 columns.  There are so many deployment rule variations this could become a fun and interesting side to the game.

Pros: Some players may actually really enjoy having a deploy phase.  A deploy phase also helps to mitigate starting hand RNG. Tuning has a lot of granularity which makes it easier to get to 50% win rates.
Cons: Complicates the game with an entire new phase.  May add a barrier-of-entry for new players.  May make understanding the deployment squares in future turns confusing for new players.  Reduces some of the strategic options such as 1st turn Rambo Monks.  FTA with a deploy phase may have varying value based on the board, which increases the tuning complexity for future board designs.

Adjust Starting Crystal Health
In Go there is the notion of Komi - a point advantage granted to the player going second to compensate for the disadvantage.  When I started playing Go 15 years ago Komi was 5.5, but is currently 7.5 in the United Staes.

In Hero Academy the equivalent would be for the starting Crystal Health amounts to not be equal.  For example, instead of the standard 4500 health, the second player could have 4800 health per crystal.

Pros: Minimal impact for the new player experience.  Allows a very precise tuning knob that can be adjusted based on the map.  Elegantly preserves the symmetry of the rest of the game.  If a map ever needs more than about 400 crystal health adjustment (800 total) to bring the map to a 50% win rate, I would start to question the map design.  (For example, on the aforementioned single crystal map, I could see a crystal Komi adjustment as large as 2000 not being enough - which starts to hint that the map design itself needs to be revisited).
Cons: In some matchups you don't even attack the crystal until board control is established, and at that point it doesn't matter what the starting crystal health was.  It feels a bit strange that one player's crystal is not the same health as the other's.  This can probably be mitigated if desired by inserting some text or a mini animation in between the first and second turn "Bonus health for second player" - though that might just be even more confusing to new players.  In general - but particularly as a free to play game - I'm pretty sensitive to changes that makes the new player experience more confusing or intimidating.  Depending on the UI visualization, I could very easily see it as being awkward to compare "Who's crystal is at lower health?" unless one person's crystal health bar either started higher than 100%, or one person started lower than 100%.


First Player Starts with Less Actions
This is a fairly popular forum suggestion - since this is mostly about the first player always being 5 actions ahead, why not reduce the actions of the starting player?

Some people have suggested 4 actions but I think that's just silly - 4 actions is saying "We acknowledge there's a problem so we'll half fix it, but it'll still be a problem"

Reducing to 2 or 3 actions would be the correct fix if Robot chose to go down this road.
3 actions - This says "yeah there's an advantage but we've made it as small as possible" - and hey - maybe the second player's ability to deploy their army in response to the decisions made by the first player could be a small advantage of it's own.
2 actions - This actually gives the action momentum to the second player, but leaves the first player - arguably, with an "Initiative" advantage.

This is actually the solution I've personally adopted in my games. If I ever challenge you to a game, I voluntarily give up 2 actions now when I go first, accompanied by the following text:

I encourage anybody else reading this to consider adopting this convention as well.  Perhaps, much like "throwing the second round" in Street fighter or "Solo-Mid" in League of Legends, we can probably fix this via social convention for the people that care.  If you care about FTA, surrender 2 of your starting actions!  Or - if you want to be really boss - voluntarily surrender 3 of your actions and only start with 2 actions!


Pros: I think starting with only 2 or 3 actions will bring Hero Academy about as close to 50% as we reasonably need.
Cons: You may be detecting a pattern - but I think a forced 2 or 3 actions for the first turn will hurt the new player experience.  As a free-to-play game Robot needs to make sure when somebody downloads Hero Academy and challenges their friend - those first 30 seconds need to be representative of the game and feels satisfying.  The other possible con is that this solution probably won't bring the win rate to a theoretical exactly 50%,  but it would bring it a whole lot closer.

Conclusion
So there you have it - some solutions to the FTA.  I'd be curious to see if Robot does anything.  Personally I'd be just fine if they tweaked the single crystal map and left the rest to the community.  The community can use social convention to skip the first 2 actions among the hardcore players who care.  Organized league play can enforce skipping the first 2 or 3 actions very easily - it's akin to fighting game tournaments who will ban particular characters for that tournament under "generally accepted agreement" the character is overpowered or sports game tournaments who have at time banned the use of certain moves that exploit AI flaws beyond the control of the other player.

8 comments:

  1. I'm glad there are other people who take the game seriously enough to recognize this. I'm an old advance wars player and we've tried all kinds of things to deal with FTA.

    ReplyDelete
  2. I disagree . Certainly there is a certain FTA like in chess where white statistically has more wins over black but my opinion is that the second player has the advantage of reacting to the first players move. There are situations ( like a horrible hand vs a strong hand ) where the second player can get steamrolled. But that is mostly due to the random nature of the start . And then you can be creative and wait for the first pkayer to make a mistake abd capitalize on it .

    ReplyDelete
    Replies
    1. The update also helps opening up the game by nerging the knights and balances even more. Its all about using scrolls / rune metal at yhe right moment. I played 100+ games random/ random and maintain a win ratio of about 6/1. Since its random against good player it would be more 2.5/1 and I never feel going second is a disadvantage . My 2 cents :)

      Delete
  3. The importance of first turn advantage really depends on the game and the rules of that game. Every game will be different and some games give the advantage to the person who goes first, others to the person who goes second, and for some games it doesn't matter. In Hero Academy's case, the advantage goes to the person who goes first.

    I recognize that you may feel that it's not a disadvantage to go second, but unfortunately the statistics don't support that position.

    http://www.robotentertainment.com/forum/topic/Post-Patch-League-Won-Loss-Data-0

    Council:
    When going first: 51-28 (64.6% win rate)
    When going second: 53-41 (56.38% win rate)

    Dark Elves:
    When going first: 56-58 (49.1% win rate)
    When going second: 38-54 (41.3% win rate)

    Dwarves:
    When going first: 18-25 (41.9% win rate)
    When going second: 19-31 (38% win rate)

    Does the first player always win? Well of course not. But the first player is winning a statistically significant proportion of the time.

    ReplyDelete
  4. FTA is a competitive only thing. it does exist and is significant for people who take competition seriously. Those who don't (or just aren't that good) will never care, which is fine, they're just out for casual fun anyway.

    If going second isn't a disadvantage, then why not completely skip your first turn every game?

    ReplyDelete
    Replies
    1. Perhaps, but instead of just focusing on one game, make the league tournament games 2 of 3 or 3 of 5 etc with a random draw. For the forfeiting of the first turn why not. Challenge for 3 games and i will and we will see how it goes :)

      Delete
  5. It's something we're looking at very closely. I'd be interested in seeing the results of the next league tournament now that the Council's FTA was weakened in 1.1. We also have another change or two coming down the pipe for update 1.2. Thanks for another excellent blog post.

    --
    Marcin

    ReplyDelete
  6. The randomness of the items also give and take your advantages. If you're lucky, you get good items. Even if you are going 1st or 2nd, what you have in hand will affect the outcome. So if you want FTA, play 3 games whereby the last game first turn will be decided by a draw or a flip of a coin so it still depends on your luck.

    ReplyDelete