Building better leaderboards
Made it, Ma! Top of the world! — Arthur “Cody” Jarrett, White Heat (1949)
One of the first full features of SimCity BuildIt that I had a chance to work on my own as a game designer is the Contest of Mayors. It was back in 2016, and it is still one of the pillars of the game. This feature contains a pretty elaborate leaderboard structure. I believe that this structure is at least half of the magic formula that guaranteed the longevity of this feature. It is also something that can be easily transferred to other games and game genres.
Understanding the leaderboard
Since the arcade days, leaderboards are one of the mainstays of video games. Even in a larger scope of the physical world, leaderboards are everywhere from the UEFA Champions League Standings to the Little League Baseball.
If you are designing a game in which a player can accumulate some kind of a score, chances are you will feel the urge to create a leaderboard. By definition, a leaderboard is an ordered list of players. It is usually presented in descending order with the player with the highest score sitting at the top of the leaderboard.
It might not be obvious at the first glance, but leaderboards are a first stab at designing a metagame. They exist outside of the core gameplay. Even in the arcade games in the ’80s, they were seen on separate screens shown either before or after the actual game would start.
KEY IDEA: A leaderboard is the start of a metagame!
A list is a simple structure, yet it holds powerful sway over the human psyche. For some reason, probably driven by imperatives of our evolutionary survival, we are driven to compete against each other. A leaderboard does just that. It provides means of comparison of our own performance at some activity against others. By some magic, it transforms a single-player game into a social competition!
KEY IDEA: A leaderboard turns a single-player experience into a social competition!
Leaderboards are especially efficient in doing this. Even this very rudimentary design offers a chance for players to instantly form several parallel goals:
- Reach the top of the leaderboard,
- If reaching the top is not immediately possible, at least try to climb up several places,
- If even this is hard, at least try to overtake the player immediately above you.
The time scale and effort needed to achieve these goals ranges in relative terms from immediate, overtaking the player immediately above, to long-term aspirational, reaching the top of the leaderboard. This is the nucleus of the Density of Goals I was mentioning in my previous texts.
This basic form of leaderboard does the job pretty well. As anyone playing any school game knows, these three goals already can generate quite a lot of drama and excitement.
You might be tempted to take the design of leaderboards for granted and stay satisfied by this basic template. Yet this simple structure has a couple of fundamental weaknesses, which need to be understood, and then taken into account if you want to maximize the impact of leaderboards in your game.
Pitfalls
The main problem with the basic leaderboard structure is that it doesn’t scale well. There are two scaling issues that can manifest themselves, depending on the type of game you are designing:
- time,
- the number of players.
Time considerations originate in the speed by which players can accumulate points. If your leaderboard is refreshed after every core gameplay session (or even several times during a session) it can be a focus of players’ attention. On the other hand, there is nothing worse than a stale leaderboard. If the time needed for any player to collect enough points to move up in the standings is too long, players will quickly lose interest in the leaderboard.
Think about the Eurosong voting. This is easily one of the most exciting contests on TV. Voting participants do not just give a single point/vote to the song they consider the most worthy. Instead, they get to distribute an increasingly larger number of points to several songs, starting from 1 and climbing all the way to 12. As a result, the standings on this leaderboard can change dramatically with each round of voting!
If you are designing a leaderboard, make sure to tie it to a score that can change significantly in an acceptable amount of time. The score that changes several times during one play session is a good starting point.
The second big issue in designing leaderboards is the question of scale, i.e., the number of players that can be simultaneously ranked in a single list.
Many games still contain some sort of a Global Leaderboard, i.e., a simple ranking of all the players that have accumulated any score within the game during a certain period of time. Given the number of players that even a mediocre game has nowadays, this list typically contains tens if not hundreds of thousands of names. This mistake is surprisingly common in game design, and it is hard to understand why.
As a player, I do not give a flying fuck about moving from place 156 342 to place 136 758 on some global leaderboard. The numbers are too large and too random to be understood on an emotional level. If anything, they are intimidating. The prospect of me reaching the Top 100 let alone Top 10 on this leaderboard is too daunting and too distant.
By the virtue of things, there can be only one person at the top of the leaderboard. Good for him, but everyone else, including the guy at spot no. 2 will eventually feel frustrated.
Both the time and the scale problem are interconnected. They can work in tandem to exasperate each other. Some of these leaderboards are based on some kind of a slow-changing lifetime score. This gives a huge advantage to players that have simply started to play the game earlier, making the experience even worse for each new player joining the game. No matter how much time and effort, I as a new player, sink into the game I will never be able to catch up with an advantage of millions of points that Top 10 players have accumulated simply because they had months or even years of a head start!
The end result is a stagnant leaderboard, with very little value for the players. The key question is why to have such a feature at all.
A better approach
If we are aware of these issues, we can design around them and produce better, more meaningful, leaderboards.
The first step is something that I already mentioned. The leaderboard should be tied to some relatively fast-changing score, ideally one that can change during a single gameplay session. This will go a long way towards creating emotional suspense.
KEY IDEA: Tie your leaderboard to a fast-changing score.
The second step has to do the scale of the leaderboard. Moving only one or two places on one leaderboard can feel infinitely more meaningful than moving 50 000 places in another. Divide your players into several small groups. The group size can vary, and depends on the type of game. However, it shouldn’t be smaller than 10 nor larger than 200 players. In my experience, a group size of 100 is a good number. Each individual group should have its own separate leaderboard instead of having a global one. These tight groups in combination with fast-changing scores should provide for intense competition.
KEY IDEA: Divide the players into groups of 100 and make a separate leaderboard for each group.
Numbers from 1 to 100 are small enough to feel intimate. Their meaning can be grokked immediately. Moving a couple of places on this small leaderboard feels intuitive and meaningful.
You can develop things further by building on top of this basic structure. You can construct a ladder of leagues. Periodically, for example, once a week, you can promote the top-ranking players from each of the groups into the next league. You can also downgrade the players from the bottom of the group to a league below.
KEY IDEA: Create a league ladder. Promote periodically top-ranking players from each group into the next league.
If you maintain the group size, as only a few players are promoted in every round, each subsequent league will have fewer groups. Your system will resemble a pyramid of player groups. You can tweak the parameters of the system to ensure that the top league always consists of only a single group. In this way, the top-ranking player in the top league will be the top-ranking player overall, with all the honors and bragging rights that this entails. The system will be equivalent to having a Global Leaderboard while avoiding its main shortcomings.
You can find a great example of the implementation of exactly this structure in Duolingo, a gamified language learning app.
Finally, you can overlay a reward scheme on top of this system. Reward players at the end of each round based on their standing. For example, the top player should get the Grand Prize, players in places 2 to 5 should get a Big Prize (whatever that might be), players in places 6 to 20 can get a Mediocre Prize and everyone between 21 to 50 could get the Consolation Prize. Players ranking below 50 get diddly-squat.
The Density of Goals in this structure is dramatically bigger than in the case of a simple leaderboard. The player can pick and choose his next goal based on his abilities and current standing on the leaderboard. He can decide to try to overtake another player, move up the leaderboard, or reach the top position. In addition to these, he can choose to reach the promotion threshold for the next league. He can also aim to move up the league ladder, or eventually reach the top league. He can even dream about reaching the top place in the world.
Since scores reset at the end of each competition round, everyone has a fair chance of catching up with the veteran players. Players can also aim for a particular reward bracket within their current leaderboard. At least they can try to avoid being demoted to the league below.
Keep in mind that in this context, climbing the leaderboard and a league ladder is a form of intrinsic motivation, while rewards granted are a form of extrinsic motivation. You can read more about different types of player motivation in my previous text.
Limitations
Of course, the structure that I just described is much more complex than the simple leaderboard. This complexity needs to be taken into account when planning the development of such a feature.
However, in addition to the higher development cost, this approach has some conceptual issues. What happens if the player reaches the top position? Well, OK, good for him. He is going to feel awesome for a little while. He might even feel the pleasure of fighting to maintain his top position. Also, each new victory will only yield diminishing returns. The game essentially ends for this player. The same problem applies also to anyone in the top league. The number of players in this group might be tiny compared to your overall player base, but these are most likely your most ardent players and best customers. Your game needs to cater to their needs.
There are two ways of solving this, and none is perfect. You can simply add a new league on top of the existing top league. You can continue to do this whenever there are enough players to fill up at least one competition group. This leads to a naming problem. You can’t call it the Top league if there is going to be another league on top of it soon. Some games resort to inventing names such as Gold, Platinum, Diamond, Obsidian, Deuterium, Elite, Legendary, etc. The problem with that approach is that you need to keep outdoing yourself with each new league name. As an alternative, you can start assigning increasing numbers to leagues. However, this is counterintuitive. Players expect that the First League is the top league!
Another approach is to periodically reset the league ladder and the scores of all players. This would result in all players having to reclimb the league ladder all over again, reliving the complete user experience. This has a disadvantage in that players would lose their top league status each time the ladder gets reset. This is a hard sell. To make it easier ensure that this is the part of the game from the very moment of introduction of the league ladder. Introducing this system later is definitely possible. However, it is much harder to explain to players than if it would be there in the first place.
Takeaways
- A leaderboard is a part of the metagame
- Leaderboards turn single-player experience into a social competition
- Leaderboards provide players with a set of clear goals
- The leaderboard should not feel stagnant
- Tie your leaderboard to a score that can change fast
- Divide the players into groups of 100 and make a separate leaderboard for each group
- Create a league ladder and periodically promote or demote players based on their performance
- Reward players at the end of each round based on their standing on the leaderboard
Links
- Density of Goals
- Player Motivation
- Duolingo leagues the essential guide
- EA Help article on SimCity BuildIt: Contest of Mayors
- Fan page about SimCity BuildIt leagues
- Eurosong voting system
