I'll give some somewhat random thoughts after skimming one chapter: 1) The Democrats acted less like a crowd than a controlled partisan group. They did less to express individualized opinions, and instead reacted to politics on a national level. Assuming as one might that Florida and Ohio were stolen through black box voting, etc., and that Kerry actually won, the Democrats as a whole still did poorly simply because they failed to reflect localized priorities, which would have given a more robust opinion/platform/strategy. Dean's strategy of building up the party in 50 states rather than keep trying to pick the hot contests seems to fit well in the "good decisions of crowds" mold. E Pluribus Unum, not E Unus Unum. (or E Anus Anum?)
2) I saw an economic analysis of the tulip bulb crisis quoted in Charles Mackay (Popular Delusions and The Madness of Crowds). This analysis noted that the type of tulip bulb hording that went on made economic sense under the conditions. A cartel had just been broken, and there was good reason to speculate, similar to a hot IPO. That doesn't mean everyone comes home a winner, only that the odds are better than the local lottery. 3) I think the thesis' 4 points leaves out at least two points, which are the transparency of the issue/information and the evaluation of the result. If the deck is stacked (WMD's in Iraq, say), the resulting decision is likely to be flawed. Unfortunately, most of our decisions have skewed information in numerous ways. The second aspect, evaluation of the result, is non-trivial with non-trivial problems. Short-term? Long-term? General good? Field specific? Under whose terms are we evaluating the fitness of the solution, and are those the best, most sensible? An example of the latter is the Betamax vs. VHS contest. The market made a wise decision - Betamax didn't offer compelling advantage to the consumer for the extra price, and it locked in manufacturers, but for the professional recorder it turned out poor. A somewhat similar tradeoff was made between analog cassette and DAT audio, and as consumers we got stuck with pretty crappy sounding home recording for about 20 years (and it wasn't much better for professionals thanks to high costs). A short-term "good" decision can have long-term consequences. 4) The book does seem to point to a view many of us probably share, "going it alone is not the best policy". Advice and input are important for more robust solutions, though if the input is not diverse, independent and decentralized, it merely reflects the thinking of Great Leader. But that is a different issue and model from the idea that a crowd of non-coordinated agents shows decision-making ability as good as an individual. Group Think => Extended Sampling => Mob Think When testing a GUI or other system, one rule-of-thumb has been that of marginal returns after 4 or 5 unique testers, i.e. you will find 95% or so of the problems with 4 or 5 sets of eyes, so instead of getting another 100 sets of eyes, fix the problems and then run a new test with 4 or 5 unique testers. 5) The use of the Imo monkey stories is a bad sign. While Surowiecki at least doesn't give us the Hundredth Monkey myth, he seems to exaggerate the actual takeup and transmission of potato washing and dunking-wheat-in-the-sea among monkeys. One piece of recent research shows that kids favor imitation far more than monkeys, even where imitation is a poor strategy for the circumstance. http://rocketjam.gnn.tv/headlines/6663/Children_Learn_by_Monkey_See_Monkey_Do_Chimps_Don_t Is there something compelling about human allegiance? Would human crowds decide better than monkey crowds, and would that relate to "autonomous" agents still wanting to imitate or be like others in the group, whereas monkeys might be more autonomous? Or is desire to please unimportant in this evaluation process. 6) In my study of offshoring, I came across several economists - John Dunning, Rugman, Verbeke - that used the imagery of flagships steaming into foreign port with their consorts surrounding them - basically that offshoring was much more of a symbiotic relationship involving competitors who helped each other as well, shared resources, advantages in numbers, and various alpha, beta, delta roles I wonder if a similar analogy works for opinion "clusters" - where you have opinion groups and communities that form around alphas - some as beta supporters, some as delta "lurkers", competitors that often prop each other up through thesis-antithesis healthy debate and that if you're trying to aggregate opinion, getting a cross-cut of these clusters could make a robust system. Deltas would probably be better for idea transmission between groups, Alphas would be better in coalescing solid opinion. Unfortunately, the Internet seems to be removing diversity, independence, and ironically decentralization from Web/blog political thought. Even as it excels at methods for aggregating the results. Hmmmm....methinks that's enough before my first morning coffee. Owen Densmore wrote: > I've been reading this critter: > http://tinyurl.com/hexhe > .. and am interested in its application to social modeling, and > possibly business/organizational modeling. > > The thesis is that good decisions can be made by crowds if they are: > - Diverse > - Independent > - Decentralized > - Good method for aggregating the results. > > I started on the book a while back while discouraged after the > democrats shot themselves in the foot the last election. Thinking > crowds were stupid, I was surprised a bit by the author's thesis. > > Anyone read it? Have opinions? Got ideas how to apply it to > community modeling? > > -- Owen > > Owen Densmore > http://backspaces.net - http://redfish.com - http://friam.org > > > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org > > > ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org
