Banner and Crosspost

Home    Overlay   Discord   Mutators   Maps   Integration   Links   About

June 9, 2021

Transmission 09 - Gameplay variety

Gameplay variety
Transmission 09
  Gameplay variety

This post will focus on gameplay variety as it’s particularly important for keeping the game interesting and fresh. I will also touch on related topics including player interactions, issues with the early game, cheese and all-in rushes.

Sources of gameplay variety

For the game to stay interesting, we want game-to-game variety. It would be no fun if every game looked the same. We also want moment-to-moment variety, where there is a potential for something exciting happening at any moment. Games like soccer or basketball accomplish that very well – scoring a goal is always just a few passes away. Gameplay variety has to come from somewhere:

  • Player execution – in games with a high skill ceiling for execution, variety (and randomness) is introduced through imperfections. Mistakes will inevitably happen, and engagements will resolve slightly differently each time. A strategy has to be constantly adjusted around these outcomes.
  • Player interactions and strategies – rich strategic space and interactions between players can lead to a great number of distinct game states. Players can't be familiar with all of them – leading to each game feeling more unique. This will amplify any differences between games and moments. Chess and Go make use of this. Distinct game phases can improve moment-to-moment variety.
  • Player selection – variety introduced through getting different teammates and enemies. In 1v1 this leads to different matchups. In team games, you will often get a unique combination of allies and enemies. When amplified by team bonuses (Age of Empires II) or varied specializations (MOBAs), this can lead to each game being unlike another.
  • Built-in randomness:
    • Randomly selected map from a map pool (StarCraft)
    • Randomly generated maps (Age of Empires II)
    • Random drops (Warcraft III)
    • Random combat (miss chance in Brood War or Company of Heroes II)
    • Random map patterns, enemy compositions, mutators (StarCraft II Co-op)
    • Random card draws, effects (Hearthstone)
    • Random units (blind pick mode in monobattles in StarCraft II)
Engagements are full of mistakes and that makes each of them unique (StarCraft: Remastered)

Built-in randomness

Built-in randomness is a great source of variety for casual games, however, it can be tricky for competitive modes as it shouldn't significantly affect the outcome of matches. It's there to introduce gameplay variety or limit the information horizon, not to take away the player's control over outcomes. To prevent these issues, randomness can be either reduced or take the form of input randomness – that players can react to and manage better.

A mirror matchup on a randomly generated but symmetric map is an example of input randomness. There is randomness, but players are informed of it and then adjust their builds and strategy around it. (This is similar to Prismata's game start). If it was not a mirror matchup, and one faction was heavily favored by the map, then it would be output randomness. The random element (map generation) would come after a decision (choosing a faction). Players would have no influence on which player would be randomly rewarded for choosing their faction.

Input randomness feels better because players keep more control over the game. Though it wouldn't be hard to come up with unfair and frustrating examples as well. The example with a mirror match on a randomly generated and symmetric map is an outlier, typically it's hard to introduce such an impactful random element while keeping it fair. This post on Prismata and many others have gone in-depth about the role of randomness in games.

Discover mechanic in Hearthstone – cards are randomly chosen from a pool, but the player retains some control by then choosing one card and forming a strategy around it.

Player interaction

Interactions between players can be a great source of variety in a game, and are typically the most exciting part of the game as well. Too few interactions can lead to boring or uneventful game phases, as is often seen in the early game.

More interactions can also give players better feedback on how they are doing during a game. If two players just macro-up, fight one battle and the winner takes the game, was the issue with macro, build, composition, or micro? Providing this feedback is very hard for RTS games, but more interactions between players can help with that. A game where both players macro up and end it with a single engagement is a bit like racing where contestants only compare times in the end. Racing becomes a lot more interesting when you can compare how the contestants are doing throughout the match, and that's without direct interactions. Although racing has the advantage of having a simple game state where it's clear who is ahead and by how much.

It might be a good idea to target a minimum number of interactions between players in a game. It doesn't feel great when one side lost because they didn't scout one thing or just after one engagement. This is often unsatisfying and can feel like a waste of time.

The mission structure in Company of Heroes II leads to constant player interaction and provides good feedback to players on who is ahead

Skill shots are widely used in MOBAs because they facilitate good interactions between players – allowing both players to express themselves through skill. It's rewarding to successfully both hit and dodge skill shots, and less frustrating to fail to do so. The feedback is immediate, it's clear where the mistake was and how to improve. The skill shot usage can be exciting, easily comprehensible for both players and viewers, and introduce variety through imperfect player execution and enemy prediction. Skill shots might not always be the best fit for an RTS, but we can look for similar mechanics.

StarCraft II Co-op is a two-player cooperative PvE mode. Because it lacks back-and-forth interactions with a human opponent, it has to rely more on built-in randomness to improve gameplay variety. Though to a lesser degree, variety can be introduced through an ally. This works a lot better when the presence of an ally has a meaningful impact on your gameplay. This can happen through passive bonuses (Guardian Shell, Vorazun invisible bonuses) or active cooperation (shared Nydus Networks, Stetmann’s satellites, Dehaka’s P1 bonuses).

Early game

The early game is where RTS games often struggle with long stretches of nothing interesting happening. Legacy of the Void approached these issues by increasing the number of starting workers, Age of Empires II introduced Empire Wars mode where players start with more villagers and some pre-built structures, and Grey Goo has its quickscript mod giving players more resources and accelerating the early game. In all cases, the goal is to skip at least a part of the uneventful early game.

Cheeses and all-in rushes can make the early game more interesting, but they are not a great solution as they often have binary outcomes and facilitate only limited interactions between players before the game ends (e.g., did you scout it or not, or a game being decided by one fight). And by their nature cheeses and all-in rushes cannot, or better to say should not cover all of the games played – still leaving many matches with uneventful early games.

In the previous post, I wrote about macro, which could make limited interactions in the early game less of an issue for players if macro was sufficiently enjoyable by itself. However, that won’t solve it for spectators. If there are five minutes of just macroing up, viewers will naturally switch their attention to something else, and possibly miss the action when it happens. Viewers are not excited by enjoyable macro, and even if there are strategic decisions introduced through it, the complexity of those decisions won't carry five or more minutes with very limited player interactions.

★ ★ ★

I think the lack of player interactions in the early game is a consequence of its steep economic growth and high opportunity costs. If a player sends units over the map, the chances are that the enemy will have significantly more by the time the units get there and crushes the attack. That leaves the first player with two choices:

  • Make those units earlier – and with that sacrifice the economic growth and go all-in
  • Don't attack in the first place – which leaves us with the uneventful early game

Another way to look at this problem is that units have to be balanced around all-ins first, which can leave them too weak for aggression in the early game when not going all-in.

Few things that can introduce more interaction into the early game:

  1. For the attacking player, early units that you can't easily all-in with – for example units that don't scale with numbers, units focused on dealing economic damage, or units with number limits (like heroes).
  2. Avenues to deal limited damage to the defending player. The limit is the important part, it can't be easily pushed into an all-in. You can steal sheep and boars from the enemy in Age of Empires II, but the player will be only slowed and not dead. You can harass villagers outside of the protection of Town Centers, but early units have very low damage against buildings. The enemy capturing your region in Company of Heroes II is bad but not game-ending. This ability to deal limited damage puts limits on the effectiveness of static defenses as they shouldn’t make a player completely safe, at least not without investing in them excessively.

    Having avenues for limited damage doesn’t mean something big cannot happen. For example, a player might overextend and lose a hero which can be a game-ending moment. However, punishing an opponent's mistake is different than going all-in from the start. There is at least one additional interaction that's required and preceding the action.
  3. Good ability to retreat in the early game. It prevents fewer units from being overwhelmed, reduces risks of noncommittal aggression, encourages early action, and reduces skill floor (not having to know when moving out equals losing the game).
  4. Points of interest on the map can encourage early interactions between players (for example Immortal's Pyre camps, WC3's creep camps, C&C3's Tiberium spikes, or the whole map in Company of Heroes). These points of interest provide something to fight for and build strategies around. However, if it’s crucial to control these points, defensive strategies might become unviable.

The goal of all of these is to make early interactions more viable, encourage players to gain advantages through them, but without them deteriorating into all-ins and thus being constrained by that balance. Immortal: Gates of Pyre is covering multiple points with Pyre camps, safety towers, and making sure units can retreat in the early game.

Age of Empires II has prolonged early game, but the free scout unit can provide some action, and reacting to an unknown randomly generated map makes things more interesting.

The focus on worker harassment in StarCraft II is a form of limited damage. Hellions killing workers deal damage to the economy, but they likely won't kill your buildings and army as well. However, this alone didn't make the early game that exciting, Hellions come relatively late, and diving on workers is a big commitment. The gameplay around denying creep spread is better as it facilitates more interactions, but it’s still limited and the stakes are quite low. Ideally, there would be more avenues for early interactions between players.

★ ★ ★

Introducing gameplay variety early into the game – whether it's through player interaction or randomly generated map – can also break static build orders. That could make the early game more interesting, increase the skill ceiling (reacting to new situations), and lower the skill floor (reducing the burden of knowledge of learning build orders). Completely breaking static build orders might be unrealistic due to various practical reasons, however, I don't believe it would be a bad thing if that happened. Strategies would form around convergent points which are the basis of build orders but without the exact path how to get there.

Mid & late game

Gameplay variety is important past early game as well, otherwise the game would become unexciting and the meta stale. This affects the longevity of the game and can reduce the need for patches to shake up the meta.

Typically a variety is introduced through player execution that has kept games like StarCraft I/II, Warcraft III, or Age of Empires II exciting to this day with only some strategic variations. However, we can't forget that games like Chess or Go did manage to stay alive for centuries based on the variety through sheer strategic complexity and some player differences.

It's also interesting to note that the perceived variety can be different between players and spectators. For players, even small things can be interesting: changing timings or build-orders. However, spectators likely won't notice these things and require more visible changes – different strategies or army compositions.

Cheeses and all-in rushes

I have touched on cheeses and all-in rushes already as they introduce some variety to the early game. But it might be interesting to look at them more closely even if it’s not further connected to the main topic of gameplay variety.

  • Cheese – an aggressive strategy that relies on not being scouted. If it’s scouted, its chance of winning is minimal. It usually hits early due to limited scouting opportunities and other factors.
  • All-in rush – early attack (rush) that has to kill or severely cripple the opponent (all-in)

The reliance of cheese on not being scouted leads to luck having a stronger impact on the game’s outcome. Also, in a standard game, randomness introduced through decisions and execution will be mostly averaged, and the better player will usually win. However, in a short game and with only a few interactions, these factors won’t get averaged, and so the chances of them going favorably for the less skilled player are higher. For these reasons, it’s a good choice for a less skilled player to cheese a better player, as luck gives him better chances than a standard game which tests players’ skills more thoroughly. If I want to win, I would rather play rock-paper-scissors with Serral than a game of StarCraft.

An all-in rush doesn’t necessarily rely on being unscouted. But it’s still a good choice for less skilled players as there will likely be fewer player interactions.

It’s not to say that good players won’t cheese, in fact they should. In game theory, mixed strategies are the optimal solution for certain situations. And so even the best macro players should mix in cheese or aggressive builds to prevent opponents from being too greedy and cutting corners, even if that means losing few games they would have won playing normally.

Cannon rushes usually fall under cheese, all-in rush, or both depending on the variant (StarCraft II)

An upside of cheeses is they provide gameplay variety in the early game, and there is often a good diversity between cheese strategies. The more cheeses there are and the newer they are, the more effective they become. It's also less difficult to come up with a cheese than to innovate in midgame and lategame. That makes it easier and rewarding to come up with new ones. Still, cheese is often perceived as something bad and frustrating to play against. Why is that?

One reason might be its perceived “cheapness” as it’s much easier to learn a single aggressive strategy than to properly defend all of them and know how to play a normal game on top. I imagine this is even worse in lower leagues. Less experienced players might gravitate to them more as cheeses or rushes are the simplest working strategies and the easiest paths to get a few wins. Because of this, it might be harder for players in lower leagues to play a “standard game”, as they will have to learn how to defend all the aggressive builds first. This significantly increases the skill floor for new players if they don’t want to cheese themselves.

Cheeses increasing the role of luck in games is another reason. Losing on ladder to a coin flip always stings. In tournaments, a clear underdog is more likely to cheese, but there are very few players like Has who can cheese all the way to the top. The more likely scenario is that the clear underdog will get eliminated in the next round, and we might not see any more games of the better player. But even a cheese between two good players can feel like the game was decided by a coin flip and we have missed out on a better game.

Hiding a Dark Shrine to rush the enemy with Dark Templar (StarCraft II)

And finally, cheeses change how the game is played. A player might have an image of a game he wants to play – some macro, micro, multitasking, and armies clashing against each other, and queues up looking for this experience. However, the opponent cheesing can force the player to play a different game, for instance a game of defending a cannon rush. While scouting a cheese, having a solid build, and learning how to defend and execute these strategies are things that definitely fall in RTS gameplay, the player might feel that the gameplay is different enough from the expected experience that he has no desire to engage with it – it can feel like a wasted time whether it's a win or a loss. Of course, you could extend this to any strategy that the opponent is doing. ”I want to play against bio and the opponent went mech, oh no!” We want to have a diverse set of strategies viable, and so the question is then how different gameplay can be forced on players?

Individual players will have their own preferences – some like certain cheeses, some don’t, some prefer one matchup over another. With matchups, unranked players could have the option to queue for a certain matchup, but with cheese you have to balance these different experiences living together while some of them will be forced onto players. And if some cheeses become unviable for various reasons, the early game might become uneventful – as it usually happens once meta stabilizes. Because of this, I think it’s better to design for rich early interactions that don’t rely on the existence of cheeses and all-in rushes. They can still exist, but the game doesn’t have to rely on them.

When not depending on all-ins to keep the early game interesting, a game can be more stable – players have a higher defender’s advantage. I believe a more stable early game is better for both team games and players in lower leagues. Conversely, when depending on all-ins to keep the early game interesting, the game necessarily has to be balanced around the possibility of a player dying at any moment, which puts limits on the defender's advantage and can feel overly punishing.


Thank you for checking out this post. These ideas have been on my mind for some time, and getting to write about them has been fun. The topic of gameplay variety, early game, and player interactions is very interesting and important for the game to be fun and exciting on game-to-game and moment-to-moment scales. I don’t believe this post brought up something radically new. I see it as an overview that could improve further discussions on related topics.

For discussion check this thread on r/FrostGiant.

Previous posts:

Recent posts



Powered by Blogger

Main post