（最近，关于修订的《治安管理处罚法》中”吸毒记录封存”的这一新规，在社会上引起了巨大的争议。也许下面这篇文章能从另一个视角，引起我们的深刻反思：如何去看待“精英”认知与群体智慧之间的差异，尽力找到避免社会撕裂和内耗的平衡点——笔者注）
1907年3月7日，英国统计学家弗朗西斯·高尔顿发表了一篇论文，阐释了后来被称为“群体智慧”效应的现象。他进行的估计实验表明，在某些情况下，大量独立估计值的平均值可以相当准确。
这种效应利用了这样一个事实：当人们犯错时，这些错误并不总是一致的。有些人倾向于高估，而有些人则会低估。当足够多的这类错误被取平均值时，它们会相互抵消，从而产生更准确的估计。如果人们相似且倾向于犯同样的错误，那么他们的错误就不会相互抵消。用更专业的术语来说，群体智慧要求人们的估计是独立的。如果由于任何原因，人们的错误变得相关或相互依赖，估计的准确性就会下降。
然而，华金·纳瓦哈斯领导的一项新研究为这一经典现象提出了一个有趣的转折。该研究的关键发现是，当人群被进一步分成允许进行讨论的小组时，这些小组的平均值比同等数量的独立个体的平均值更准确。例如，从四个五人讨论小组的估计中得到的平均值，明显比从20个独立个体那里得到的平均值更准确。
在一项针对100名大学生的后续研究中，研究人员试图更好地了解小组成员在讨论中实际做了什么。他们是否倾向于听从那些对自己的估计最有信心的人？他们是否跟随那些最不愿意改变想法的人？这种情况有时会发生，但这并不是主导反应。最常见的情况是，小组报告称他们 “分享论点并一起推理”。不知何故，这些论点和推理导致了整体错误的减少。尽管纳瓦哈斯领导的研究存在局限性，许多问题仍有待解决，但它对小组讨论和决策的潜在影响是巨大的。
附：英文原文 On March 7,1907, the English statistician Francis Galton published a paper which illustrated what has come to be known as the “wisdom of crowds” effect. The experiment of estimation he conducted showed that in some cases, the average of a large number of independent estimates could be quite accurate.
This effect capitalizes on the fact that when people make errors, those errors aren’t always the same. Some people will tend to overestimate, and some to underestimate. When enough of these errors are averaged together, they cancel each other out, resulting in a more accurate estimate. If people are similar and tend to make the same errors, then their errors won’t cancel each other out. In more technical terms, the wisdom of crowds requires that people’s estimates be independent. If for whatever reasons, people’s errors become correlated or dependent,the accuracy of the estimate will go down.
But a new study led by Joaquin Navajas offered an interesting twist(转折) on this classic phenomenon.The key finding of the study was that when crowds were further divided into smaller groups that were allowed to have a discussion, the averages from these groups were more accurate than those from an equal number of independent individuals.For instance, the average obtained from the estimates of four discussion groups of five was significantly more accurate than the average obtained from 20 independent individuals.
In a follow-up study with 100 university students, the researchers tried to get a better sense of what the group members actually did in their discussion. Did they tend to go with those most confident about their estimates? Did they follow those least willing to change their minds? This happened some of the time, but it wasn’t the dominant response. Most frequently, the groups reported that they “shared arguments and reasoned together.” Somehow, these arguments and reasoning resulted in a global reduction in error. Although the studies led by Navajas have limitations and many questions remain, the potential implications for group discussion and decision-making are enormous.
