This post is based on the paper: “Harnessing the Wisdom of the Inner Crowd,” written by Stefan M. Herzog and Ralph Hertwig that appeared in Trends in Cognitive Sciences, October 2014, Vol. 18, No. 10. This is a slightly different take on a subject addressed in the posts: Dialectical Bootstrapping and Bootstrapping. Herzog and Hertwig seem to be the go to guys on bootstrapping. In the title they obviously refer to James Surowiecki’s The Wisdom of Crowds. (see post Smart Mobs and Diverse Problem Solvers). They explain that a lone individual can enlist the wisdom of crowds by averaging self-generated, nonredundant estimates. They review evi-
dence for this ‘wisdom of the inner crowd’, and consider how it can be produced, how its accuracy can be improved, and whether people use it to their advantage. Frankly, Figure 1, above puts the advice in one spot.
Reducing error redundancy will increase averaging gains within a person, but what reduces error redundancy within a person? The key insight is that error redundancy reflects, among other things, the amount of redundant information contained in multiple estimates – either by the same person or by different people. Consistent with this, averaging gains are larger for people with lower (vs higher) working memory spans. Error redundancy also decreases when previous estimates are not permitted to hold strong sway over subsequent ones. Specifically, techniques that reduce the ‘control’ of previous estimates can reduce error redundancy and boost averaging gains. There are two ways to do this. One approach uses the power of forgetting. For example, increasing the time delay between subsequent estimates – and thus freeing latter estimates from the traction forces of former estimates – can increase averaging
gains. A second approach takes advantage of the mind’s ability to construct alternative realities. People can be actively prompted to base their second estimate on different assumptions, pieces of information, inference methods, and elicitation procedures. One way to achieve this process of ‘dialectical bootstrapping’ is to ask people to ‘consider the opposite’ when making a new estimate.
According to Herzog and Hertwig, people underestimate the benefit to be gained from averaging their own opinions with those of advisors. People are more likely to combine their estimates if they actively challenge the premises of their first answer in the process of generating a second one. Considering the opposite may make us aware of conflicting, yet legitimate, assumptions and reasons. The authors suggest that combining the resultant estimates is an elegant tool for trading off these conflicting realities.
Another factor conducive to combination is the magnitude of disagreement between two estimates. The larger the numerical difference between the two estimates, the more inclined people are to combine them. People seem unable to outperform the inner crowd and would be better off always strictly averaging their estimates in most environments. Herzog and Hertwig also wonder if it would be possible to identify people who could benefit more than others from
enlisting the inner crowd. For instance, do elderly people with declining cognitive resources benefit more (akin to people with lower working memory span)?