In the first half of this post, I described three types of complexity:
- Comprehension Complexity (how does this component work?)
- Board Complexity (how do these components work together?)
- Strategic Complexity (how will these components work together in the future?)
The first two types are ‘bad complexity’ – they’re very visible to new players, and too much of them creates a large barrier to entry before the game can start to be fun.
The last type is ‘good complexity’ – it doesn’t come from the rules or immediate actions so it’s much harder for new players to see, but it gives experienced players something difficult to master and keeps the game interesting for a longer period of time. To distinguish it from the other two types, I’m referring to it as depth.
Ideally, we want to craft games that have as much depth as possible while keeping the ‘bad complexity’ within appropriate limits for your target audience. How would we go about doing this?
In 2014, Magic head designer Mark Rosewater wrote another article about what he refers to as lenticular design. The name comes from lenticular printing, a technique that allows a single image to appear differently when viewed from different positions – it’s often used to make posters appear ‘3D’ or to allow for simple animations on postcards:
The idea here is that it’s possible to design components and rules which are perceived differently depending on the skill level of the observer. We can use this to make things that are engaging and accessible for all players, but contain hidden depths that keep players interested as they learn more about how your game works.
An alternate analogy might be focus. In my experience, it’s not so much that new players and veteran players see different things, but more that veteran players see the same things in more detail. Without my contact lenses in, the world is a blurry mess – I’m still able to perform basic tasks like shopping, but I sure as hell shouldn’t drive. With my vision restored, complex tasks are now available to me, and I might notice details around the basic tasks that makes them more productive or efficient (e.g. finding better quality items to buy, or noticing a special offer).
Rosewater’s article concentrates on Magic card design. However, I believe the underlying differences in player perception are universal, and can be made more general:
New players tend to react to events as they happen rather than anticipating and planning around them. This can be exploited in the following two ways:
Firstly, it can be used to ‘backload’ game rules. If the game is structured such that new components are encountered during play, then the rules for these components can be hidden away in a glossary or on the components themselves (e.g. cards). Newer players will be oblivious to what they don’t know and adapt on the fly. Advanced players can use their knowledge of the hidden components to plan for possible futures.
Secondly, the probabilities for random events are often hard to calculate accurately, but easier to approximate. New players will ignore these and play reactively. More experienced players will have a vague idea about the likelihood of certain events, and roughly plan around the main ones. Very experienced players will have a firm grasp on the likelihood of certain events, and accurately plan around most of them.
For example, in The Castles of Burgundy, players will initially roll their dice and see what actions they can perform with the numbers rolled. With more experience, they’ll play in such a way that more dice numbers are beneficial to their current plan, meaning fewer rolls ‘miss’ and are wasted. Additionally, players don’t need to know what each tile’s special ability is until they’re able to purchase and place it – often, placing the tile is the main goal, and the placement effect is a nice bonus.
New players tend to use components in the most obvious way first. Once they’ve figured out what that is, they’re happy to move on to figuring out the next component. Adding subtle ‘secondary uses’ to components is a good way to exploit this.
For example, in Dominion, players will initially use the Chapel card to only remove harmful Curse cards from their deck. With more experience, they’ll also use it to remove weak positive-effect cards (Copper, Estates) from their deck as well, so that they’re more likely to draw powerful hands.
Note that if a component doesn’t have an obvious use, new players will get confused and/or frustrated, as they’ll correctly realise they’re missing something. Sticking with Dominion, the Chancellor card is similar to Silver, except it costs you a (scarce) action point to use and has the mysterious ability to put your whole deck into your discard pile when played. This indirectly allows recently purchased cards to be drawn sooner, as your discard pile will be turned face down and shuffled to replace your empty deck, but this isn’t obvious and confuses new players.
As games are complex systems, it takes a while to understand the relative worth of game components. How many ‘resource cubes’ a ‘card in hand’ is worth depends a lot on how all the game components interact, which will usually take multiple plays to puzzle out. Additionally, the value of components may shift over time – a red cube at the start of the game might be much less valuable than it is at the end.
If the true valuations of a set of components are roughly similar, then an easy approximation for new players would be to treat them all as being worth exactly the same. More experienced players can take advantage of the subtle differences and shifts in value.
For example, in Catan, players initially want to build roads and settlements (requiring Brick and Wood), and later on they’ll want to build the more expensive cities and development cards (requiring Stone). Each resource is obtained in the same way, so trading them one-for-one with the other players as they’re needed isn’t far wrong. However, savvy players can make better trades by paying close attention to supply and demand as the game progresses.
New players don’t look many moves ahead, as they’re still trying to learn how the game works at a base level. If they can do something positive right now, they’ll usually do it – even if they might be better off waiting to do that thing at a more opportune time. Making actions optional is a simple way to exploit this.
For example, in Hearthstone you put minion cards from your hand into play by spending mana, a resource that grows slowly over time. New players will often play minions as soon as they can afford to do so, which is usually the correct strategy. However, the game has a lot of effects that damage or destroy all opposing minions at once. Playing more minions than you need to puts you at risk of being blown out by one of these ‘sweeper’ cards, so sometimes you want to not play your minions immediately if you’re under no pressure. Knowing when to continue developing your board and when to hold back is an interesting decision that’s invisible to newer players.
Sub Terra tries to implement these lenticular design techniques in the following ways:
- Tiles and Hazards (causality)
Sub Terra revolves around two ‘decks’; the stack of tiles which players uncover as they explore the cave, and the deck of hazard cards that cause bad things to happen at the end of each turn.
New players don’t need to know what all the tiles and cards do when they start playing. The tile details can be looked up from the rulebook or reminder sheets during play, and the hazard cards have written instructions on them to tell players how they work. It’s possible to play the game reactively without needing to keep the full range of possibilities in your head at once.
Experienced players will know what all the tiles and cards do, as there aren’t that many unique types. They’re then free to pay attention to the probabilities that a given tile or card will be drawn next. At a simple level, they can just look at the base frequencies – one in five hazard cards are ‘cave-ins’, for example. At a slightly less simple level, they can consider what tiles and cards have already been revealed to get a feel for the current probabilities – if six of the eight gas tiles have already been placed, it’s unlikely that a new tile will also be gas, and exploring during gas turns is slightly safer.
- Backtracking (causality)
Sub Terra takes place in a random tile-based cave system. Players can move to adjacent tiles as long as there isn’t a solid wall in the way, and the mix of tiles is such that most caves resemble a series of short branching tunnels. Players will often need to backtrack, possibly because the way forward is blocked, possibly because they’re straying too far from their friends, and so on. To make this harder, the cave deteriorates over time, and passages that were once easy to traverse may suddenly find themselves flooded or blocked behind a pile of rocks, slowing down your return journey.
Knowing when it’s correct to stop pushing forward and start your return journey is the crux of the game. It depends on various factors, such as how injured you and your friends are, your relative locations, your character abilities, the current and possible future states of the path separating you, your team’s future plans, and so on. Doing this optimally is difficult. Getting close requires considerable discussion with your team. This is where the bulk of the strategic complexity in the game comes from.
However, if you’re a new player, you can just ignore all this and react to the bad things as they happen. You’ll be playing the game less well than if you were planning ahead, but the lowest difficulty setting intentionally builds in a lot of slack to give you more time to correct your mistakes.
- Characters (function)
There are eight unique characters that players can choose from at the start of the game, each with their own set of abilities, strengths and weaknesses. Each character has been designed to push you towards a certain strategic role – for example, the Scout wants to push ahead and open up constrained routes, and the Medic wants to run around after people and keep them in peak condition.
However, it’s not necessary to stick to this role. Each character also shares a set of ‘basic actions’, covering movement, exploration and healing. As the game develops and things start to go wrong, it’s often correct to abandon your ‘ideal’ role and do something less efficient for the good of the team. Knowing when to do this is a judgement call that improves with experience.
- Health and Action Points (economy, timing)
The basic resource unit in Sub Terra is the action point. Players get two action points a turn, with the option to risk a third point with a 50% chance of losing a single health. Players start with three health, and the ‘heal’ action allows them to spend two action points to gain one lost health back. This establishes a baseline exchange rate of “one health = two action points”, which is good enough for most basic reasoning.
But not all health is created equal. The penalty for losing all your health is to be knocked unconscious – this stops you from performing any actions until a teammate is able to reach your tile and heal you (a heavy time cost that could have been spent exploring). Your first health point is therefore the most valuable as it’s the only thing separating you from unconsciousness, and your third health point is the least valuable as you have two other health points to use as a buffer. Different hazards can cause you to lose different amounts of health, so if you’re in an area with high potential for cave-ins (3 damage if triggered) your third health point matters less than if you’re in an area with lots of gas (2 damage if triggered). The decision on if, when and where to spend a turn healing up is heavily contextual.
So that’s how I tried to make Sub Terra deep, in order to keep the cooperative play exciting and engaging. I hope this was interesting! In my next post, I’ll be taking a break from the heavy design theory and discussing art. See you then!
Game Designer, Sub TerraIndie gamedev robot. All opinions, fatalities and apocalypses are the responsibility of my creators, who should have worked harder on the control problem