Vol. 26 No. 15 · 5 August 2004
pages 19-21 | 3960 words
How many jellybeans?
David Runciman
- Profiles, Probabilities and Stereotypes by Frederick Schauer
Harvard, 359 pp, £19.95, February 2004, ISBN 0 674 01186 4
- The Wisdom of Crowds: Why the Many are Smarter than the Few by James Surowiecki
Little, Brown, 295 pp, £16.99, June 2004, ISBN 0 316 86173 1
Most of us, most of the time, are deeply prejudiced in favour of individual over collective judgments. This is hardly surprising, since we are all biased. First, we are biased in favour of our own opinions, which we tend to prefer to those of anyone else. Second, we are biased in favour of individuals generally, because we are all individuals ourselves, and so are broadly sympathetic to the individual point of view. We like to think of people exercising their personal judgment, and not just blindly following the rules. For example, who wouldn’t prefer, when appearing before a judge, to learn that the judge was willing to hear each case on its merits, and exercise some discretion if necessary? General rules, we think, are likely to be discriminatory, because they cannot take account of special circumstances. Individuals, by contrast, can use their own judgment, and make exceptions.
You are not Logged In
- If you have already registered login here
- If you are a print subscriber using the site for the first time please register here
- If you are not yet a subscriber you can subscribe here
- If you are a member of a subscribing institution or University library please login here
- If you have an Institutional print subscription and online access is not included, find out about our Institutional online subscriptions
This article is also available for purchase from the London Review Bookshop. Contact us for rights and issues enquiries.
- print this article
Letters
Vol. 26 No. 17 · 2 September 2004
From Jeffrey McGowan
In his review of James Surowiecki's The Wisdom of Crowds (LRB, 5 August), David Runciman misses an important point about the use of experts. The obvious requirements for choice of sample, independence and so on are only necessary conditions, not sufficient ones. A properly random sample of people can be completely wrong about something if they are all working on the same incorrect assumption. Consider the jellybean example: if there is a large glass sphere hidden among the jellybeans, then you will get a normal distribution of guesses centred on the wrong mean. An expert – someone who knows about the glass sphere – will be able to make a much more accurate guess.
If you want to determine the relative usefulness of experts and crowds in a particular situation, you need to consider the relationship between them, and the ability of the crowd to make reasonable guesses. There are situations in which there are no real experts, such as an unfixed jellybean contest; situations in which everyone is an expert; and situations in which there are real experts. A good example of the second type of scenario is betting on horses. Ask a thousand random people to look at the form and then bet on a race and most professional gamblers will be happy to bet according to the crowd's decision.
A simple example of the third type of situation is the well-known doubling problem: ask a thousand randomly chosen people how much money you'll have in a month's time if someone gives you a penny today, two tomorrow and so on. The group mean will almost certainly be an answer that's much too low; but ask an expert (anyone who understands basic exponential functions) and you'll get the right answer. The run-up to the war in Iraq was probably one of these situations. The point as regards Iraq is not that the crowd somehow knew better than the experts, but that the experts lied.
Jeffrey McGowan
Glastonbury, Connecticut
From Clifford Story
It is a fundamental principle of experiment that, if there's no bias, experimental error will be randomly distributed. That's why an experimenter will repeat his experiment, and average the results. The average, with experimental error roughly cancelled out, will be more accurate than a single measurement. It should be no surprise that this applies to guessing the number of jellybeans in a jar. If, however, we asked people to guess the number of jellybeans in a jar labelled '1000 jellybeans', the average guess would be in the neighbourhood of 1000. The label introduces a bias into the experiment.
Runciman proposes that matters of public policy be submitted to 'a large group of people … to give it their best guess'. But while we can easily find unbiased jellybean counters, I would not know where to look for unbiased policy-makers. Had such a process been applied to the decision to invade Iraq last year, the American people would have approved it (Runciman is wrong to suggest the opposite). Public opinion last year was biased (in the experimental sense) by the overwhelmingly one-sided propaganda favouring invasion. We see through this now, but all the averaging in the world could not have saved us then.
Clifford Story
Murfreesboro, Tennessee
From Michael Scott
'Taken individually,' Rachmaninov once said, 'the people in an audience may be poor critics of music, but as a complete body, the audience never errs.'
Michael Scott
Mauritius
