Cass R. Sunstein on Collective Intelligence - Dictionary of Arguments
Collective intelligence/Sunstein: when questioning randomly selected groups, one usually learns very precisely what people think, e. g. election prognoses, popularity of TV shows, etc.
It is something else if you want to know what is true, not what people believe. Here are some famous examples:
Hazel Knight asked students many years ago how warm it was in the room. Estimates varied greatly, the mean hit the correct value but quite precisely. (1)
When people need to estimate the number of beans in a container, it is similar. (2)
The British scientist Francis Galton estimated the weight of an ox at an auction. The result was close to a pound. (3)
Sunstein: Question: should a company rely on the judgement of previous employees when recruiting new employees? And should people question others when making decisions about their lives and obey the average result? What about environmental measures?
Average/Sunstein: in which situations is an average made up of many opinions meaningful? See also...
Randomness: the probability that groups are correct is higher if these groups are randomly composed. (See Decision Theory/Condorcet).
Problem/Sunstein: unfortunately, one must presuppose expertise from those involved in order for the outcome of collective decisions to turn out in the sense of actual circumstances. In two types of cases, the judgement of a statically selected group will be wrong:
(a) where there is a tendency towards a particular result among the members
(b) if the answers are worse than random answers.
Test persons are often misguided by so-called anchors, e. g. numbers that are scattered into an explanation. Likewise, judges are influenced. The larger groups become, the greater the risk that such an anchor will have an effect ((s) as the same anchor is effective for each member). (4)
Statistics: should more account be taken of statistical knowledge? This depends on whether the interviewed experts were in a position to provide good answers that can then be evaluated statistically.
Community/Aristotle: when several come together (...) everyone can contribute his share of virtue and moral wisdom (...) and some will understand something, others will understand something else and all together will understand everything. (5)
Sunstein: here the whole is the sum of its parts and that is what was aimed for. This is a reading of Aristotle's suggestion that a group works better than a few of the best. But there is also the view that a group discussion delivers more than the sum of its parts. A form of information gathering in which the exchange of views provides creative solutions.
However, there may be other ways in which synergy effects and learning can lead to a group's performance that exceeds that of the best members. (6)
However, since uniformity is achieved during consultations and confidence in a result is generated, this can be favoured in the end, even if it contains errors.
Group Pressure/Solomon Asch: In a famous experiment, Ash showed how group members swung back to a clearly incorrect assessment of the group after making correct estimates (in terms of line length). (7)
Investment clubs sometimes make bad decisions when members are tied by tight social ties and dissenting opinions are censored (8), (9).
Representatives of minorities in groups often behave more reservedly and develop less weight (10). In concrete terms, they speak less and exert less influence. (11)
1. Lorge et al., “A Survey of Studies Contrasting the Quality of Group Performance and Individual Performance, 1920–1957,” 344.
2. See Surowiecki, The Wisdom of Crowds, p. 5
3. Surowiecki, pp. xi–xiii.
4. Lorge et al. P. 346.
5. Aristotle, Politics, trans. E. Barker (London: Oxford University Press, 1972), 123.
6. See David J. Cooper and John H. Kagel, “Are Two Heads Better Than One? Team versus Individual Play in Signalling Games,” American Economic Review 95 (2005): 477; Gigone and Hastie, “Proper Analysis,” 143–53 (offering some examples of group success, while showing that such success is not typical).
7. See the overview in Solomon Asch, “Opinions and Social Pressure,” in Readings about the Social Animal, ed. Elliott Aronson (New York: W. H. Freeman, 1995), 13.
8. See Brooke Harrington, Pop Finance: Investment Clubs and the New Ownership Society (Princeton, NJ: Princeton University Press, forthcoming).
9. See José M. Marques et al., “Social Categorization, Social Identification, and Rejection of Deviant Group Members,” in Hogg and Tindale, Group Processes, pp. 400, 403.
10. See Glenn C. Loury, Self-Censorship in Public Discourse: A Theory of “Political Correctness” and Related Phenomena, Boston University, Ruth Pollak Working Paper Series on Economics (1993), p 3.
11. See Caryn Christensen and Ann S. Abbott, “Team Medical Decision Making,” in Decision Making in Health Care, ed. Gretchen B. Chapman and Frank A. Sonnenberg (Cambridge, UK: Cambridge University Press, 2000), 272–76 (discussing effects of status on exchange of information in group interactions)._____________Explanation of symbols: Roman numerals indicate the source, arabic numerals indicate the page number. The corresponding books are indicated on the right hand side. ((s)…): Comment by the sender of the contribution. Translations: Dictionary of Arguments The note [Author1]Vs[Author2] or [Author]Vs[term] is an addition from the Dictionary of Arguments. If a German edition is specified, the page numbers refer to this edition.
Cass R. Sunstein
Infotopia: How Many Minds Produce Knowledge Oxford 2008
Cass R. Sunstein
#Republic: Divided Democracy in the Age of Social Media Princeton 2017