Strategy games are frequently praised for depth and criticised for complexity, and the two are commonly confused despite being close to opposites in effect.
The distinction
Complexity is how much a player must learn to play at all — rules, systems, interfaces, exceptions.
Depth is how much meaningful decision space remains once you understand the rules.
Chess and go have minimal complexity and enormous depth. A game with hundreds of units and dozens of resources may have less depth than either, if most decisions are obvious once understood.
Which is why adding systems does not add depth, and frequently reduces it by making optimal play more discoverable through calculation.
Where depth comes from
Interaction between simple systems generally produces more depth than elaborate individual systems.
A small number of rules that combine in many ways generates a decision space no player exhausts.
Which is the design principle behind most enduring strategy games, and it is why elegant rule sets outlast elaborate ones.
Meaningful choice
A decision is meaningful when the alternatives are genuinely competitive and the correct answer depends on context.
Which means balance is not about equal power but about situational advantage — each option being best somewhere.
A dominant strategy collapses the decision space, since the choice is made once and repeated.
Identifying and removing dominant strategies is most of what balance patching actually does.
Information
Perfect information games are entirely about calculation.
Hidden information introduces inference, deception and risk assessment, which are different skills.
Which is why fog of war changes a game fundamentally rather than adding a layer, and why scouting becomes a core activity rather than a chore.
Randomness
The most contested element.
Input randomness — variation in starting conditions or available options — creates varied situations requiring adaptation.
Output randomness — chance determining whether an action succeeds — introduces variance in outcomes after decisions are made.
The first is generally regarded as adding depth. The second is more divisive, since it can make good decisions lose.
Competitive designs tend to reduce output randomness or to structure it so that expected value is what skill controls.
Snowballing
The structural problem in games where advantage compounds.
Winning a battle produces resources, which produce more strength, which produces more wins.
Which means games are decided early and the remainder is execution, and it produces long unenjoyable losses.
Catch-up mechanisms address this and risk making early play meaningless, which is the opposite failure.
Games generally handle it through comeback opportunities that require skill to exploit rather than through automatic rubber-banding.
Time pressure
Real-time strategy adds execution speed as a skill dimension, which changes what the game measures.
Turn-based removes it, allowing full consideration, which favours planning depth over reaction.
Neither is deeper. They measure different things, and the frequent argument about which is more strategic conflates the two.
The onboarding problem
Deep games with low complexity are the ideal and are hard to design.
Most successful strategy games instead manage complexity through progressive introduction, revealing systems over hours.
Which works and it means the first hours are not representative of the game, and that is a real barrier for anybody evaluating whether to continue.
Asymmetry
Different factions with genuinely different mechanics rather than statistical variations.
Which multiplies the decision space, since each matchup is its own problem, and it makes balance far harder.
Balance in asymmetric games is generally assessed statistically across large numbers of matches at each skill level, since intuition is unreliable.
Win rates differing by skill level are common and awkward — a faction weak at low skill and strong at high skill is genuinely difficult to balance for both.
The metagame
Strategy choices made in response to what other players are doing, which shifts over time independently of any change to the game.
Which means a game can feel different across months without a patch, as the population discovers and counters approaches.
Healthy metagames cycle. Stagnant ones settle into a dominant approach, which is the signal designers watch for.
Learning curves
Replay analysis, in-game statistics and post-match breakdowns all help players understand why they lost, which is the actual bottleneck on improvement.
Games providing this well retain players longer, since visible improvement is what sustains engagement in a competitive game where win rate stays near half.
Artificial opponents
Building a competent computer opponent is genuinely hard, and most rely on advantages in resources or information rather than on better play.
Which players detect, and it undermines single-player challenge in a way that is difficult to fix without substantial investment.
Interface as a skill barrier
In real-time games, a substantial portion of measured skill is interface manipulation rather than strategic thinking.
Which some designers regard as a legitimate dimension and others as noise obscuring the actual game.
Quality-of-life improvements that reduce mechanical burden shift the balance toward decision-making, and they are consistently contested by established players.