Researching Multiplayer Competitive Play

Otherchess is not just a game — it’s an unexplored design space. Thousands of board configurations are possible, most have never been played, and the question of which ones produce fair, fun, competitively viable games is wide open. The community is doing the work of answering that question through the collective labor of play and board design.

We are actively researching the theory of advanced multiplayer competitive play and want your input as we experiment with new ideas. This page lays out the problems we’re thinking about and the solutions we’re prototyping.

Fairness & Skew

In a two-player game, fairness is straightforward: neither starting position should confer a systematic advantage. We measure skew — how much one seat wins more than its fair share across many games between equally-skilled players. Traditional chess has a well-documented White advantage of ~54–56%. In Otherchess, skew varies by board and is discovered empirically, game by game.

Every rated game you play contributes a data point to the fairness estimate for that board. Over time, the community will know which boards are balanced and which impose a starting-position tax on one seat. We’re building infrastructure to surface this data: per-board win rates, per-seat breakdowns, and aggregate fairness scores — all bucketed by player-count configuration, because the same board played 2-player is a different competitive entity than the same board played 4-player.

What we need from players: Play rated games on a variety of boards. The more data per board, the faster we converge on accurate fairness scores. If a board feels lopsided, that signal is valuable.

Fun as a Measurable Property

Fairness alone doesn’t make a board good. A board where every game draws in ten moves is perfectly fair and completely boring. Fun — strategic depth, meaningful choice, sustained engagement — is a second independent axis.

We’re exploring measurable proxies for fun:

• Average game length — longer games suggest more decision space
• Player return rate — you replay boards you enjoyed
• Completion rate — games abandoned early suggest low engagement
• Low early-resignation rate — early resignation suggests the position felt hopeless, not fun

These proxies are imperfect. Player preferences vary — some find tactical fights fun, others prefer slow strategic maneuvering. The fun axis may ultimately be multi-dimensional. But even a rough aggregate score helps surface the best boards from the thousands available.

Open question: What makes a board fun to you? Do you prefer wide-open tactical positions, claustrophobic wall mazes, asymmetric starts? We’re interested in whether distinct player-type preferences emerge.

The Cabal Problem in Multiplayer Games

The jump from two players to three is not a quantitative change — it is a qualitative one. In a two-player zero-sum game, every gain by one player is a loss by the other. There is no ambiguity about motivation. In a three-player game, two players can cooperate against the third — and one player may be willing to sacrifice their own outcome to guarantee a win for their partner.

This is the cabal problem: a coalition where one member deliberately loses so another wins. It cannot be deterred by board geometry — the sacrifice is the strategy, so making it costly is incoherent. A related problem is the kingmaker effect: a player who cannot win but can still determine who does.

In any tournament or competitive structure, match-fixing is already a known problem — a player can always throw a match to manipulate standings. But in a 3+ player game, the cabal operates within a single game rather than across separate matches. It’s easier to execute, harder to detect, more immediately felt by the victim, and more decisive in outcome. Two colluders in a 3-player game can guarantee the third player loses regardless of skill.

The Shifting Balance — and Why It’s Fragile

There is an appealing vision of three-player dynamics: the shifting balance of power. The two trailing players cooperate against the leader. The leader falls behind. A new leader emerges, becomes the target. No lead is safe. The game becomes a dynamic equilibrium of constantly shifting alliances between three genuinely selfish agents.

This can produce exciting and strategically rich play. But it only works if all three players are genuinely and independently selfish — each wanting to win, willing to betray a temporary ally the moment it becomes advantageous, and never willing to sacrifice their own outcome to influence someone else’s. If any player is willing to lose on purpose — for a grudge, a side-deal, a rematch favor, or because they control multiple accounts — the equilibrium collapses. The game shifts from pure strategy to a trust exercise.

Chess culture values the board determining the outcome, not off-board psychology. So while the shifting balance is theoretically interesting, it may be inherently unstable as a competitive format.

Cabal-Free Configurations

The structural rule is simple: a configuration is cabal-free if and only if there are exactly two players.

A 6-side board played by 2 players (3 sides each) is zero-sum. There is no third party to sacrifice against. The cabal problem does not exist. That same board played by 3 players (2 sides each) has the full cabal problem.

This means the space of cabal-free Otherchess is enormous: any board, any number of sides, assigned to exactly two players. A 6-side or 8-side board played 2-player can have multi-front warfare, rich asymmetric dynamics, and strategic complexity far beyond standard chess — without any coalition ambiguity. Explicitly asymmetric boards (e.g. 1 player with 3 sides vs. 1 player with 1 stronger side) are a particularly interesting subspace.

Proposed: The Draw Pact Mechanism

We’re exploring a draw pact mechanism that could extend competitive viability to 3+ player configurations.

A draw pact is a conditional agreement between a subset of players, made at a time when they are not the only players remaining. It stipulates: if at any later point the pact members are the only players left, a draw is automatically declared among all members.

The pact is not a social agreement — it is a game mechanism enforced by the platform. Once declared, it fires automatically when its condition is met.

How this neutralizes cabals:

• If players A and B form a draw pact and C is eliminated, the game immediately draws between A and B. There is no incentive for A to sacrifice for B — A cannot gain more than a draw from the partnership.
• Coalition structure becomes explicit and public rather than hidden and psychological.
• Players who refuse draw pacts signal that they intend to compete independently — also useful information.

Open design questions we’re thinking about:

• Should draw pacts be irrevocable or cancellable by mutual consent?
• Should a player be limited to one draw pact at a time?
• Should pacts be visible to all players? (Probably yes, for competitive integrity.)
• Should rated games weight draw-pact draws differently from positional draws?
• Does this actually make 3-player rated play viable, or do higher-order strategies emerge?

We want your input. Does the draw pact mechanism sound like it would make multiplayer games more fair? Would you use it? Do you see failure modes we haven’t considered? This is active research and we’re genuinely uncertain about the right design.

The Bigger Picture: Mapping the Design Space

Traditional chess has been played for ~1,500 years. Its single starting position has been analyzed to exhaustion. The question “what is good chess?” has been answered — thoroughly, if not finally.

Otherchess has thousands of possible starting positions, most never played. Their fairness is unknown. Their fun potential is unknown. Which ones are competitively viable is entirely open. This is not a gap to be embarrassed about. It is the opportunity.

The Otherchess community is the first group of humans to systematically explore this space — asking which configurations are balanced, which produce satisfying games, which sustain competitive depth. Every rated game is a data point. Every board design is a hypothesis. Every player who engages seriously is doing the work of discovery.

The deeper ambition is a function from configurations to quality: given a board and a player count, what is the expected fairness, fun, and competitive stability? That function does not yet exist. Building it — through play, data, and theory — is one of the most exciting things about this project.

Get involved: Play rated games on boards you haven’t tried. Design new boards and watch how they play. Tell us what felt fair and what felt broken. The boards being played today are the raw data for the theory of Otherchess.