The report said the biggest intelligence problem was "group think" but nothing in any of the reform measures I've seen even tried to deal with this problem -- if there really is a way to deal with it.

Dan, I have a question about the Surowiecki book: I keep hearing people talk about "collective wisdom," along with methods of "aggregating data," and how this is a good way of 'predicting' the future, or some such...

But my question is: is this really any different than just, basically, the Law of Large Numbers?

(Where the words 'methods of aggregating data' do the work of 'independent samples from an unknown distribution', and the whole thing is equivalent to the mathematical statement that, in the appropriate space, the distribution of finite samples converges in the limit to the underlying distribution from which the samples are drawn).

Law of Large Numbers, Glivenko-Cantelli, or something along those lines...

I'm SO relieved to see someone openly questioning the idiotic idea of giving one person (or a small cadre or wise people) total control over all US intelligence. I don't really care what the evidence is, I intuitively buy the argument (how's that for an easy sale?).

I would strongly support the creation of a commissioner/committee to oversee all the operations of our various intelligence agencies. Obviously, someone should have a handle on the whole enchilada, and that someone should have the president's ear pretty regularly. However, to put all budgetary authority AND all planning AND all hiring and firing authority into one guy's hands--have these people lost their minds? W spoke the truth Monday when he said the President and the president alone bears the ultimate responsibility for these matters. Thank God someone has the Constution's protection at heart.

Of course, the 9/11 commission could never contemplate that the person/persons chosen for this omnipotent position could be anything other than wise, impartial and more than a bit clairvoyant. No J. Edgar Hoover type could ever capture more than his share of power, much less misuse it!

And it bears asking WHY they believe this to be true? Why? Well, ask yourselves what pool of talent such a Tsar could possibly be chosen from. Answer: former Congressmen! They NEVER make mistakes, do they?!

Finally, this morning, an NPR reporter hinted (too subtly I fear, but I was grateful for it nevertheless) that much the squabbling about to take place over intelligence reforms amounts to good old-fashioned legislative/executive branch cat fighting.

My sense is that until Congress shines a light on its own culpability and offers significant self-reform, it can forget about overhauling Presidential power. But I may be in a small minority here (and in so much else these days).

aha. You Chicago types always want to go too far ...

"Surowiecki's argument is simple -- when left to their own devices, large numbers of people who have diverse talents and perspectives will be consistently better than all individuals at problem-solving, decision-making, and future predictions."

... under certain conditions that Surowiecki lays out.

IIRC, Surowiecki has also commented on Jane Galt's site as well. I mention it only because I too think it's pretty cool that he gets in there and mixes it up with the hoi polloi (though I'm not sure how hoi the CT crowd is - all those Ph.Ds, etc.). He and Gladwell (and I guess you'd have to include Hersh) are the best things about the New Yorker, and sometimes on blogs you get to have your McLuhan-Woody Allen moment with him. Wierd and cool.

> Surowiecki's argument is simple -- when left to
> their own devices, large numbers of people who
> have diverse talents and perspectives will be
> consistently better than all individuals at
> problem-solving, decision-making, and future
> predictions.

That is interesting - I will definately have to read this book.

FWIW, every time I have done the "Wilderness Survival" or similar management training exercises, the individual results have exceeded the group results by at least 25%. And my personal experience in both corporate and community settings is that crowds rapidly converge to the level of the LEAST capable member.

But perhaps I don't have enough experience, or am missing some other key factor.

Cranky

Cranky,

Ever stood in a very long line (say, in an airport), only to peak around the corner, see ANOTHER line with almost no one in it? Wisdom of the crowd or herd mentality? You decide.

I havent read the book, but if the crowds are wiser than individuals, how does the author explain the irrational and often violent mob mentality? Is it merely a matter of the right conditions that determine whether a group is intelligent or idiotic mob? And certainly a mob of children will not make as good of decisions as a single educated adult.

Surowiecki's argument is simple -- when left to their own devices, large numbers of people who have diverse talents and perspectives will be consistently better than all individuals at problem-solving, decision-making, and future predictions.

I knew there was a reason why China and India are superpowers and the US isn't!

Wisdom of the crowd or herd mentality? You decide.

Cranky

For Nick:

Surowiecki discusses rioting mobs as being similar to participants in a financial asset bubble. It's partially convincing.

As someone else posted, Surowiecki doesn't claim crowds are universally wise. I agree that it's a very good book.

> Surowiecki discusses rioting mobs as being
> similar to participants in a financial asset
> bubble. It's partially convincing.

Um, could you please point me to the convincing explanation of what finanacial bubbles are and how they occur? Because otherwise there is an aspect of "and then a miracle occurs" to the argument.

Cranky

This is the first time I read your blogsite.

I was absolutely flabergasted to read your apparent boost of a book which claims that large numbers of people make better decisions than small groups. Everything in my 30-plus year professional career in consulting engineering, working with many different constituents and stakeholders, proves that exactly the opposite is true. Large groups can invariably be counted on to arrive at conclusions that none of the participants can individually justify after the fact.

What is important is to recognize the relationships involved in a given decision process and to insure that all of the stakeholders are deliberately involved and participate in the decision. This is done by representative participation of each of the stakeholders, using facilitation if necessary.

The 'crowd' as you call it does just the opposite by selectively discounting and terminating the participation of many stakeholders through interpersonal dynamics (ie - human behavior).

Do you make this stuff up or what?

Frank

Frank, Surowiecki's argument is more nuanced than your making it out to be. You're pointing out that a weighted aggregation mechanism works better than an unweighted one, and I'm pretty sure Surowiecki would agree -- one of his examples is google and its page rank algorithm (which weights). His point is, many times, in uncertain domains, we trust the opinion of one person, when it would be better to get the opinion of multiple people and aggregate (think CEOs, etc. etc. etc.)

> Do you make this stuff up or what?

Frank,
Your analysis is cogent, but I would suggest that as this is your first post you might want to hold off on comments of this nature for a while. Mr. Drezner is usually pretty thorough in his analysis.

Cranky

"A few Iraqis were trained for postwar security but "not in the volume that many had hoped," Rumsfeld said."

How come Rumsfeld likes to reinvent the wheel so often?

We had trained troops, yet he wanted to use resources to give training, eqipment, etc., to Chalabi's group.

We had a State Dept and USAID with plans, expertise, trained personnel, for the occupation/democratization phase of the war but Rummy gave it to the already overtaxed military and wierdly underqualified CPA.

We had millions of unemployed Iraqis who could drive trucks yet Rummy farms the whole transpo deal out to Halliburton and Betchtel, who then employed Pakistani, Phillipino, Indian and other low wage people to drive. (And now, some of these people are being held ransom by irate unemployed Iraqis. Though some are held by other groups.)

We have highly trained US soldiers performing tasks like building schools, and Rumsfeld funds over 20,000 private contractors to perform soldier type tasks like guarding dignitaries and providing protection for the Pakistani and Phillipino drivers. Compare the going rate of a minimum of $500 per day for the merc types to what might be the per day rate for construction people. Not to mention the frustration of our soldiers who are risking thier lives for less than 1/10th the pay of the private guys.

Cranky asks:

"could you please point me to the convincing explanation of what finanacial bubbles are and how they occur?"

Well, you could try reading the book. Failing that, the author makes the standard arguments about participants either believing the hype or thinking they are smarter than the other chumps and therefore will know when to bail out.

Brian "Also Cranky"

Won't comment on the book since I haven't read it, but JS' biz columns in the New Yorker are generally excellent: insightful, educational, engaging and -- my favorite part -- just one page long.

> Well, you could try reading the book.

Um, haven't had time to do that since 11 AM this morning, no ;-)

But the reality of financial panics challenges many widely-used economic theories. It has been studied for 300 years (at least) and to the best of my knowledge there is no satisfactory explanation. If this author has developed one in the course of his research on "group minds", then he has not 1 but 2 Nobel prizes on the way!

Cranky

Just a couple of things: all the questions that people have raised here about my thesis are good ones, and I think that in the book I try to answer them. As more than a few people have already said, a lot of the book is devoted to figuring out when collective decisionmaking works well and when it doesn't, and I hope I've avoided the "and then a miracle happens" solution (though I'm hardly an impartial judge in this case). The key things that make "crowds" intelligent are diversity (of information and opinion) and independence. I don't think those are always easy to come by, but I think the evidence is pretty overwhelming that when they're present, groups are remarkably smart.

I also think Dan's right to emphasize the implications of all this for the intelligence community. (I'm trying to write a piece about this right now.) Having a National Intelligence Commissioner or whatever may make sense from a bureaucratic/organizational point of view. But from an intelligence point of view, it seems a lot more important to make sure that the analyses that the intelligence community produces reflect the collective wisdom of all the people in it, rather than being tailored (if only because of the nature of hierarchical bureaucracies) to reflect the judgment of the people at the top of the hierarchy.

Not to depress all the folks clamoring for intelligence reform, but I'm fairly certain that the only purpose of the National Intelligence Figurehead is to have someone to decapitate when something goes wrong. A single figure won't be making any fraction of decisions about what intelligence to percolate up. He's just there to fire.

I haven't read the book yet, but I've been wondering about the wisdom of crowds vs. the wisdom of experts. I've noticed that the consensus of experts in many areas seems to be remarkably undiverse when compared to, say, market behavior. What is the mechanism by which the uninformed add value?

Dan is right in indicating that the JCS and Joint Staff have limited authority in determining the budget and allocating funds. But they do review and comment upon the budgetary proposals of the Combatant Commanders (formerly known to many as "CINCs") and the Vice Chairman leads a board that reviews programs and proposed programs to recommend where R&D and procurement money should go.

There needs to be a lot of thought given to structure and functions, but if a "Joint Chiefs of Intelligence" were formed, comprised of the Directors of the significant Intelligence Agencies, the Chairman (basically the National Intelligence Director role) could be granted similar powers over budgetary submissions by the various agencies and consolidated vetting of all new programs to assure they can meet test for utility, interoperability, and compatibility. Doing so very well might provide the foundation for the sort of coordination (sharing of diverse ideas) with independence that Mr. Surowiecki advocates.

In re: "But from an intelligence point of view, it seems a lot more important to make sure that the analyses that the intelligence community produces reflect the collective wisdom of all the people in it, rather than being tailored (if only because of the nature of hierarchical bureaucracies) to reflect the judgment of the people at the top of the hierarchy."

Wasn't that what Poindexter's futures market in policy was supposed to do?

I haven't read the book but is the argument that taking each individual's observations as (weighted) evidence, we update our own assessments in some Bayesian fashion I suppose, and thereby get at a more accurate answer?