「ヒトは乱数を生成できない」300桁の数字から個人を識別できるとの研究報告が話題に、ランダムなように見えてもヒトは無意識のうちに好みの数字やパターンを使用している

ヒトは乱数を生成できない

Human-generated pseudorandom sequences have been repeatedly found to exhibit strong regularities: cycling of response alternatives, a tendency to count in increments of one, suppression of response repetitions, and others. . . . humans apparently have a tendency to reuse certain predominant patterns and to obscure these recurring themes by implementing minor changes (i.e., variations) within the respective pattern. . . .

人間が生成する疑似乱数列には強い規則性があり、小さな変更（並べ替えや数字の挿入・削除）を交えながら無意識のうちにお気に入りのパターンを再利用する傾向がある

But tests for the individual telephone numbers, postal codes, and birthdates revealed no significant anomalies, leading us to conclude that individuality in RNG is most likely the direct result of an individual generative algorithm (“procedural cognitive fingerprint”).

個人の電話番号、郵便番号、生年月日といった数字は乱数生成に然したる影響はなく、個々の「手続き型認知指紋」に依る

Remarkably, single digit preferences alone allowed for identification of participants based on their random sequences. Individual digit preferences shine through in the random sequences, even though participants were instructed to generate an equal distribution of digits. Still, increasing the analyzed pattern length significantly improved the identification performance.

1桁の数字から個人を識別することも可能、パターン幅が長いほど識別のパフォーマンスも向上する

Reference
Schulz, MA., Baier, S., Timmermann, B. et al. A cognitive fingerprint in human random number generation. Sci Rep 11, 20217 (2021). https://doi.org/10.1038/s41598-021-98315-y