Is Reversi (Othello) a Solved Game?

No, 8×8 Reversi (Othello) has not been solved. While computers have reached superhuman strength — solving the endgame perfectly and playing at a level no human can consistently match — the complete game tree (~10^28 possible positions) has never been exhaustively analysed. Smaller variants have been solved: 6×6 Reversi is a proven first-player win under perfect play. The 8×8 game remains open. For how this affects the programs players use today, see how Reversi AI works and Reversi software tools.

What Does It Mean to “Solve” a Game?

In game theory, a game is solved when the outcome of perfect play from the starting position is proven — win, loss, or draw — regardless of what either player does. There are two levels of solution:

Weakly solved: The outcome from the starting position under perfect play is known. You know that the first player wins (or draws, or loses), but you don’t necessarily have a complete strategy map.

Strongly solved: Perfect play is computed from every possible position in the game. A complete strategy that leads to the optimal outcome is known regardless of where in the game you are.

Most “solved” games are only weakly solved. Strong solutions require enormous computational resources.

Notable Solved Games

Why Hasn’t 8×8 Reversi Been Solved?

The Scale of the Problem

Reversi’s game tree is large, though not astronomical by the standards of complex board games:

• Possible positions: Estimated at approximately 10^28
• Average branching factor: Around 10 legal moves per position
• Game length: 60 moves total (one per empty square, from the 4 starting discs)

For comparison:

• Chess has roughly 10^44 possible positions and 10^123 game tree nodes
• Go (19×19) has approximately 10^170 possible positions

Reversi is tractable by the standards of these games — but 10^28 positions is still far beyond exhaustive enumeration with current hardware. Solving checkers (10^21 positions) took decades of distributed computing — Jonathan Schaeffer and colleagues at the University of Alberta required nearly two decades of work before publishing their proof in Science in 2007. Reversi’s game tree is roughly 10 million times larger.

Why Endgames Are Solved But Full Games Aren’t

The key asymmetry in Reversi computation: endgames are tractable, full games are not.

With 20 empty squares remaining, there are at most 10^20 positions to consider from that state — and strong programs can solve these in seconds. With 25 empty squares, still feasible with good hardware. But solving from move 1 requires propagating perfect play backward through the entire game tree, which remains computationally out of reach.

This is why computer Reversi programs play perfectly in the endgame but use heuristic evaluation (scoring functions) for the early and middle game. The heuristics are very strong — strong enough to beat any human — but they are not provably perfect.

The 6×6 Solution

In 1993, researchers Victor Allis and Michael Buro confirmed that 6×6 Reversi is a first-player win under perfect play. The analysis was feasible because 6×6 has only 36 squares and a far smaller game tree than the standard 8×8 version. Allis had previously applied similar retrograde analysis techniques to solve Connect Four (1988).

This result matters for understanding the 8×8 game: it suggests the outcome of perfect 8×8 Reversi isn’t obvious from first principles. The 6×6 result also confirmed that the first-player advantage (Black moves first) can be decisive at smaller board sizes. For more on Reversi variants and smaller board sizes, see Reversi variants.

How Strong Is Computer Reversi Today?

While the game isn’t solved, computers have achieved a level of play far beyond any human. Key milestones:

Logistello (1990s)

Developed by Michael Buro of the University of Alberta, Logistello was one of the first programs to dominate human world champions. In 1997, Logistello defeated World Othello Champion Takeshi Murakami 6–0 in a match. The program combined deep alpha-beta search with a sophisticated learned evaluation function, as described in Buro’s 1997 paper in the ICCA Journal.

This was a watershed moment — similar to when IBM’s Deep Blue defeated Kasparov in chess the same year. Reversi computers had, for practical purposes, surpassed human ability.

Modern Engines

Contemporary Reversi engines use:

• Endgame solvers: Perfect play for the final 20–25 moves
• Pattern-based evaluation: Heuristics trained on millions of games
• Neural network evaluation: More recent engines incorporate deep learning approaches similar to AlphaZero
• Opening books: Databases of analysed opening positions

These programs are essentially unbeatable by humans in serious play, though they are not executing provably perfect play from move one.

What Would Solving 8×8 Reversi Require?

Purely hypothetically, a complete weak solution would require:

1. Hardware: Either a quantum computer with sufficient qubits, or a massively distributed classical computing effort many orders of magnitude beyond what solved checkers
2. Algorithm: Retrograde analysis (solving backward from all terminal positions) combined with aggressive symmetry reduction
3. Time: Estimated at potentially centuries with current classical hardware; far less with future technologies

The practical impact of such a solution would be interesting but somewhat anticlimactic for players — since computers already play at a superhuman level, knowing the theoretical outcome under perfect play wouldn’t change the human experience of the game.

What the “Near-Solved” Status Means for Players

The fact that computers dominate Reversi has real practical implications:

For learning: Computer analysis can show you the best move in nearly any position. Playing against and analysing games with strong AI is one of the most effective ways to improve.

For competitive play: Humans still compete at the highest levels because human psychology, stamina, time management, and opening preparation all matter in tournament conditions — even against other humans. The WOC remains a meaningful championship.

For game appreciation: Unlike some games where computer dominance has dampened enthusiasm, Reversi’s community has embraced computer analysis as a teaching tool. Opening theory, endgame technique, and positional understanding have all deepened because of AI analysis.

Comparison With Chess and Go

Reversi is the most likely of the three to eventually be solved, given its smaller state space — but it remains an open problem for the foreseeable future.