Loading field map
Every mahjong player faces a recurring decision: which scoring patterns, or yaku, to pursue and when to abandon them. The answer depends on how one weighs the trade-off between hand speed and final score. Over the past two centuries, five major frameworks have offered different logics for making that choice. They did not simply replace one another; they built on, narrowed, challenged, and sometimes coexisted with earlier approaches, creating a layered strategic landscape.
The earliest approach to yaku, the Classical Chinese Scoring Paradigm (1800–1920), was not a formal theory but a culturally embedded practice. In traditional Chinese mahjong, scoring patterns were inherited through local conventions and oral transmission. Players recognized a set of auspicious or aesthetically pleasing combinations—such as the "pure hand" or "all terminals"—without a universal list or written rules. The value of a yaku was tied to its rarity and symbolic meaning, not to any mathematical calculation. This paradigm treated yaku as a fixed set of lucky patterns, and hand valuation relied on intuition and experience. It was a coherent system, but one that varied widely across regions and offered no systematic way to compare the worth of different patterns.
The Japanese Standardization School (1920–1950) transformed this landscape. When mahjong arrived in Japan in the early twentieth century, players and publishers began codifying the rules. The creation of a fixed, written list of yaku—along with standardized scoring tables—was the school's central achievement. This list, which became the foundation of modern riichi mahjong, did not replace the Classical paradigm so much as provide the infrastructure that all later theory would require. By fixing which patterns counted as yaku and assigning them explicit point values, the Japanese Standardization School made it possible to compare hands systematically. A player could now ask not just "Is this hand lucky?" but "How many points does this hand score, and is it worth the risk?" This shift from cultural intuition to rule-bound analysis was the precondition for every subsequent framework.
With a standardized yaku list in place, the next question became: which patterns should a player actively pursue? The Pattern-and-Score Maximization School (1950–1980) answered by emphasizing high-value combinations. Its practitioners memorized complex patterns—such as the thirteen-orphans hand or the nine-gates hand—and aimed to build them whenever possible. The school's method was combinatorial: it catalogued the most lucrative yaku and taught players to recognize the early signs of those patterns. This approach narrowed the field of play to a set of premium targets. It was effective in casual or tournament settings where a single big hand could decide the game, but it had a serious limitation: pursuing high-value patterns often required holding tiles that slowed the hand down, leaving the player vulnerable to opponents who won quickly with smaller yaku.
The Tile-Efficiency and Speed Play School (1980–2010) directly challenged this commitment to high score. Drawing on probability theory and computer analysis, this school argued that the most reliable path to victory was not a big hand but a fast hand. Its core insight was that a player should minimize the number of tiles needed to reach tenpai (the ready state) and maximize the number of winning tiles (ukeire). Yaku were reframed as one factor among many in a broader efficiency calculation. A cheap hand that wins early is often better than an expensive hand that never arrives. The Tile-Efficiency School replaced the Pattern-and-Score School's memorization of combinations with mathematical metrics like shanten (distance from tenpai) and ukeire. It did not reject yaku entirely—players still needed at least one yaku to win—but it demoted them from primary goal to secondary constraint. This shift was revolutionary: it transformed hand valuation from an art of pattern recognition into a science of probability.
The AI-Assisted Mahjong Analysis School (2010–Present) has taken the quantitative approach further. Using large-scale simulations and neural networks, modern AI systems can evaluate millions of possible hands and discards, generating empirical data on which yaku are actually worth pursuing under different game conditions. This school does not simply refine the Tile-Efficiency paradigm; it establishes a new, data-driven basis for all yaku valuation. In many cases, AI confirms the speed-first principle: fast, cheap hands dominate in most situations. But AI has also uncovered specific exceptions where the older Pattern-and-Score logic holds. For example, in late-game situations with high stakes, pursuing a rare yaku may be optimal even if it slows the hand, because the expected value of a big win outweighs the risk of losing. The AI school thus acts as a meta-framework, testing the boundaries of earlier models and revealing where they oversimplify.
Today, the Tile-Efficiency and AI-Assisted schools are the leading frameworks, but they coexist with the earlier traditions. The Classical Chinese paradigm survives in some casual and regional play, while the Pattern-and-Score approach remains influential in teaching beginners the most famous yaku. The Japanese Standardization School's list is still the universal vocabulary of riichi mahjong. The main agreement between the leading frameworks is that hand speed is generally more important than raw score. The main disagreement concerns how to value yaku in marginal cases: efficiency models treat yaku as a constraint to be satisfied minimally, while AI analysis suggests that the optimal choice depends on precise contextual factors such as round number, opponent tendencies, and remaining tile count. This tension between model-based heuristics and data-driven nuance defines the current frontier of yaku theory. A player today must understand both the efficiency logic that prioritizes speed and the empirical insights that sometimes justify a riskier, higher-scoring path.