Monday, April 21, 2014

Leaderboards, Loopholes, and Lessons From Our First Experience with RTS

Leaderboards and Rankings: two of the main foundations for any competitive, social game. When we at Kiwi decided to take a crack at making a mobile RTS (Real Time Strategy) style game with our title Enemy Lines, we knew that leaderboards would be part of our formula for making the title work. Players who enjoy the RTS genre all know of Starcraft 2’s ladder ranking as well as other leaderboard systems in comparable games. We decided to aim for that feeling and experience in mobile form.
For Enemy Lines, we decided to base our leaderboard off of a fairly simple token system: medals. All players started with 1000 medals when they became eligible for player vs. player. Players could then win more by successfully attacking another player’s base or defending their own from an enemy’s siege. Conversely, an unsuccessful attack or failure to stop an enemy’s offense resulted in a loss of medals. At the launch of Enemy Lines, this medal system was a closed loop. If you lost an attack the medals you sacrificed went directly to the opponent for a successful defense and vice versa. This worked well for a while, until the leaderboard began to stretch itself too thin.
The simple search criteria for an opponent was +/- 30% medals. As users rose up the leaderboard, the number of opponents they could find within that medal band (+/- 30%) became smaller and smaller. This created a negative experience for the players in the top ranks of the game. They’d worked hard to gain all of those medals, yet once they got to the top, they started to get “Enemy Not Found” (ENF) messages as matchmaking began to fail. Not only was the search pool smaller at the top, a lot of high-ranked players would leave their games on indefinitely. This prevented them from being attacked and kept them out of the matchmaking pool, helping them individually but hurting the system as a whole. Frustrated top players began to “dump” medals (purposely losing in order to find easier enemies to attack). Soon, we found that very powerful, high-level players were decimating new players at low medal bands. We had to find a solution to both the ENF and Dumping problem. The dumping problem was addressed fairly easily. We added a +/- 5 “player levels” to the matchmaking search. This meant that high-level players who dumped medals still couldn’t be matched against a disproportionately new player. These players would either fight others who had dumped their medals, or they wouldn’t find opponents at all. The ENF issue for higher level players was a bit more complicated to solve. We decided on a solution where we would “clone” a player base that would be an appropriate match, upgrade all the defenses to max, and fill the base with troops. When a player would have gotten an ENF message, they were instead presented with the “cloned” base. This eliminated the ENF problem, but a HUGE new problem reared its head. Players were having a blast with the clone bases, but we underestimated the strength of some of these players. With no limit on the medals or resources looted from these clone wars, high-level players took just hours to begin abusing the new system. Two days in, a player had racked up more than a thousand medals from the clone maps and perched himself at the top of the leaderboard. We quickly released a patch capping both medals and resources that could be gained from cloned bases per day, but the damage was done. The players who had abused the early stages of the clone system had cemented themselves atop the leaderboard. The first time you branch into a new genre, things are going to happen that you can’t anticipate. Once a game is live it evolves in ways you would never expect and you have no choice but to roll with the punches. Reflecting on Enemy Lines and the RTS genre as a whole, there was one feature that would have made all of the problems easier to handle: Leaderboard/Rank Resets.
Leaderboard/Rank resets are actually common in all top RTS and competitive titles. Starcraft 2, Hearthstone, League of Legends, World of Warcraft, even Clash of Clans utilise resets in some shape or form, often calling them “Seasons”. During a Season, players climb the leaderboard and compete for top ranks. Once the Season is over, rewards are distributed and ranks are reset to a baseline. The next Season starts, and the cycle repeats. Enemy Lines needed this. While the addition of Seasons wouldn’t have solved all the game’s problems as they sprang up, it would’ve provided an opportunity to undo them soon after. Lets take a look at what happened in Enemy Lines:
  • Top ranked players unable to find matches become frustrated and want to “dump” ranks to find competition.
  • Players dump medals and ranks to find easier opponents.
  • Certain players abuse a loophole, planting themselves at the top of the leaderboard.
Each of these problems are fixed when the Season ends and all players are brought back down to the baseline.

I’ll end with the key lesson we learned from Enemy Lines, one we learned too late: Seasons are the key to any successful competitive game. Not only does it discourage users to see one player infinitely at the top of a leaderboard, having a way to wipe the slate clean if [read: when] something goes wrong is invaluable. Every leaderboard needs the NES reset button, don’t forget yours. - Nevin Vorfeld Associate Product Manager, Kiwi, Inc.

No comments:

Post a Comment