正解: BCorrect: B
集合知が機能するには、多様性・独立性・分散性・集約の四条件が必要で、どれか一つが欠けると集団は個人より愚かにもなりえます。Collective intelligence requires diversity, independence, decentralization, and aggregation; when any is missing, a group can become less accurate than an individual.
Surowiecki は、Francis Galton が1906年の品評会で雄牛の体重当てを分析し、約800人の推定の中央値が実際の体重にきわめて近かったという逸話から出発し、集団の誤差が互いに打ち消し合う条件を整理しました。多様性は異なる誤りを生み、独立性は誤りの相関を防ぎ、分散性は局所的な知識を活かし、集約の仕組み（市場、投票、平均）が個々の判断を一つにまとめます。選択肢Aは多様性を破壊し、同じ偏りを共有する集団を作ります。Cは独立性を失わせ、集団思考の典型的な原因です。Dは規模だけでは不十分で、SNS上の情報カスケードのように多数が互いに影響し合う場合、規模が大きいほど誤りが増幅されます。Surowiecki begins with Francis Galton's 1906 analysis of a county fair contest in which the median of roughly 800 guesses of an ox's weight came remarkably close to the true figure, and he distills the conditions under which group errors cancel out. Diversity produces different errors, independence prevents errors from correlating, decentralization draws on local knowledge, and an aggregation mechanism such as a market, vote, or average combines individual judgments. Option A destroys diversity and creates a group sharing the same biases. Option C eliminates independence and is a classic cause of groupthink. Option D is insufficient; in information cascades on social media, larger groups amplify errors when members influence one another. In practice, simply having everyone write their estimate before seeing others' numbers greatly improves the conditions for collective wisdom.
現実の組織では、四条件を同時に満たすのは難しく、特に独立性は上下関係や会議の場の雰囲気によって容易に失われます。また集合知は「推定」や「予測」のような答えのある問いに強い一方、創造的な戦略の選択のように評価軸自体が争点となる問いには単純に適用できません。Real organizations rarely satisfy all four conditions at once; independence in particular is easily lost to hierarchy and meeting dynamics. Collective wisdom also works best for estimation and forecasting questions with a knowable answer, and does not transfer simply to creative strategic choices where the evaluation criteria themselves are contested.
予測や見積もりでは、匿名で個別に数値を集めてから議論する「見積もりポーカー」や社内予測市場のような手続きを標準化します。戦略選択のような問いでは、集合知を「選択肢の評価」段階に限定して使い、選択肢の創出や評価軸の設定は別の方法（デザイン思考や統合思考）と組み合わせます。For forecasts and estimates, standardize procedures such as planning poker or internal prediction markets that collect numbers anonymously and individually before discussion. For strategic choices, confine collective judgment to the evaluation stage, and combine it with other methods such as design thinking or integrative thinking for generating options and setting criteria.