Femtoreview: The Wisdom of Crowds, by James Surowiecki
The
Wisdom of Crowds, by James
Surowiecki, is a look at what must be many years of research into the
way collective decision making may actually be more efficient and accurate than
individual—even expert—reflection. I think this might be a very
useful concept in scientific research and in running a research
group.
This femtoreview is sort about the book: The
Wisdom of Crowds, by James
Surowiecki. However, in the grand book reviewing tradition, you will
find only cursory references to the actual
book.
"Groupthink" has been much maligned of late. We'll leave that just hanging out there—fortune-cookie-style—for you to adapt to your particular world view and political perspective. The plexus is not about politics.
What is it, and why is it so bad? First read the wikipedia's definition:
"Groupthink" has been much maligned of late. We'll leave that just hanging out there—fortune-cookie-style—for you to adapt to your particular world view and political perspective. The plexus is not about politics.
What is it, and why is it so bad? First read the wikipedia's definition:
Groupthink is a term coined by
psychologist
Irving
Janis in
1972
to describe a process by which a group can make bad or
irrational
decisions. In a groupthink situation, each member of the group attempts to
conform his or her opinions to what they believe to be the
consensus
of the group. In a general sense this seems to be a very rationalistic way to
approach the situation. However this results in a situation in which the group
ultimately agrees upon an action which each member might individually consider
to be unwise (the risky
shift).
It kind of reminds me of being a teenager. It seems to me that one of the main principles of groupthink is that the members of the group must in some way not be selfish. That is to say, they are willing to sacrifice their own outcome/reward in order to gain something from the group (or simply to retain their membership in the group). Were each member to insist on maximizing his or her immediate outcome, his or her acquiescence to the group would presumably be harder to tolerate.
Of course, James Surowiecki's book is not really about groupthink, since he would likely view it as a pathology—a local minimum that needs to be surpassed in order to restore the robust and efficient decision-making powers of a group.
This book is all about the counterintuitive observation that groups of otherwise non-experts can nevertheless arrive at "expert" decisions as long as their information "gene pool" is deep enough and that they are not trying to impress each other (i.e. they avoid groupthink). There are many fascinating examples in the book, and I really don't want to give them away. If you do start to read this book and find the idea of prediction markets intriguing, you might enjoy playing around at the NewsFutures site. It's totally free, and you cannot even spend money if you want to (there are other sites that allow this, and I'm sure you will not have difficulty finding them on your own).
Since the point of these reviews is to give the reader a little taste and then for us to give our general opinion, let us just say that there are some very interesting, and perhaps eerie parallels that we see as we try to view this book through the lens of the practicing scientist. For example, how are scientific facts assimilated into the body of knowledge? Is there a panel or an "académie"? Some people might point to the Science Citation Index (or Web of Science as it seems to be called these days) as a sort of disinterested arbiter of consent. After all, if my papers get thousands of citations, they are probably right, no?
One thing I see as I peer ever so gingerly into biology is that there seem to be ideas, theories, mechanisms, etc. that just "are" and are sort of accepted, and one might wonder whether it's all been proven or just taken to be the most plausible explanation. I certainly don't know enough to argue, and I imagine that physical scientists do pretty much the same thing, but it's still hard to know how much weight to give to all those lovely cartoons in the textbooks.
The point is simply that there is no question that scientific debates are carried out among people—social beings who for all their data and error analysis and controls studies, are just as susceptible to coercion and a desire to fit in as anyone else. We might be even worse, since unlike the non-academic science world, where one may derive a certain measure of success from financial reward, we are more likely to judge ourselves exclusively through the recognition of our peers. Perhaps this is an inevitable consequence of anything that is driven largely by fashion. I don't know.
What can we do about it? One thing that really resonated was a discussion in the book of just how little dissent is really necessary in order to force a group out of a badly taken decision. In other words, most of the member may have been in agreement, but only one—or a very small number—can actually be effective in moving the group to another option. This was actually portrayed on film a long time ago as 12 Angry Men. The idea I have for my group will be to discuss papers in that style. The exercise will be called "Reviewer #8" in homage to that great film (Henry Fonda played Juror #8: he who cast the sole not-guilty vote). The idea is that everyone reads the paper—as in a normal group paper discussion—but one member is instructed to disagree with some aspect of the paper. It might have to be a completely bogus or farfetched criticism (some papers are beyond reproach), but maybe not. Either way, in debating the complaints, we should all be forced to appreciate the experiment/theory/conclusions better. It could also be called "devil's advocate", but depending on one's views regarding the devil, this implicitly takes sides whereas "Reviewer #8" is only mostly anti-reviewer, and most of us should be able to bear that burden!
It kind of reminds me of being a teenager. It seems to me that one of the main principles of groupthink is that the members of the group must in some way not be selfish. That is to say, they are willing to sacrifice their own outcome/reward in order to gain something from the group (or simply to retain their membership in the group). Were each member to insist on maximizing his or her immediate outcome, his or her acquiescence to the group would presumably be harder to tolerate.
Of course, James Surowiecki's book is not really about groupthink, since he would likely view it as a pathology—a local minimum that needs to be surpassed in order to restore the robust and efficient decision-making powers of a group.
This book is all about the counterintuitive observation that groups of otherwise non-experts can nevertheless arrive at "expert" decisions as long as their information "gene pool" is deep enough and that they are not trying to impress each other (i.e. they avoid groupthink). There are many fascinating examples in the book, and I really don't want to give them away. If you do start to read this book and find the idea of prediction markets intriguing, you might enjoy playing around at the NewsFutures site. It's totally free, and you cannot even spend money if you want to (there are other sites that allow this, and I'm sure you will not have difficulty finding them on your own).
Since the point of these reviews is to give the reader a little taste and then for us to give our general opinion, let us just say that there are some very interesting, and perhaps eerie parallels that we see as we try to view this book through the lens of the practicing scientist. For example, how are scientific facts assimilated into the body of knowledge? Is there a panel or an "académie"? Some people might point to the Science Citation Index (or Web of Science as it seems to be called these days) as a sort of disinterested arbiter of consent. After all, if my papers get thousands of citations, they are probably right, no?
One thing I see as I peer ever so gingerly into biology is that there seem to be ideas, theories, mechanisms, etc. that just "are" and are sort of accepted, and one might wonder whether it's all been proven or just taken to be the most plausible explanation. I certainly don't know enough to argue, and I imagine that physical scientists do pretty much the same thing, but it's still hard to know how much weight to give to all those lovely cartoons in the textbooks.
The point is simply that there is no question that scientific debates are carried out among people—social beings who for all their data and error analysis and controls studies, are just as susceptible to coercion and a desire to fit in as anyone else. We might be even worse, since unlike the non-academic science world, where one may derive a certain measure of success from financial reward, we are more likely to judge ourselves exclusively through the recognition of our peers. Perhaps this is an inevitable consequence of anything that is driven largely by fashion. I don't know.
What can we do about it? One thing that really resonated was a discussion in the book of just how little dissent is really necessary in order to force a group out of a badly taken decision. In other words, most of the member may have been in agreement, but only one—or a very small number—can actually be effective in moving the group to another option. This was actually portrayed on film a long time ago as 12 Angry Men. The idea I have for my group will be to discuss papers in that style. The exercise will be called "Reviewer #8" in homage to that great film (Henry Fonda played Juror #8: he who cast the sole not-guilty vote). The idea is that everyone reads the paper—as in a normal group paper discussion—but one member is instructed to disagree with some aspect of the paper. It might have to be a completely bogus or farfetched criticism (some papers are beyond reproach), but maybe not. Either way, in debating the complaints, we should all be forced to appreciate the experiment/theory/conclusions better. It could also be called "devil's advocate", but depending on one's views regarding the devil, this implicitly takes sides whereas "Reviewer #8" is only mostly anti-reviewer, and most of us should be able to bear that burden!
Posted: Mon - July 18, 2005 at 04:03 PM |
Quick Links
Calendar
| Sun | Mon | Tue | Wed | Thu | Fri | Sat |
Categories
Archives
XML/RSS Feed
Comments powered by
Statistics
Total entries in this blog:
Total entries in this category:
Published On: Jul 18, 2005 04:17 PM
Total entries in this category:
Published On: Jul 18, 2005 04:17 PM
