From the archive, originally posted by: [ spectre ]


“…a book written by James Surowiecki about the aggregation of
information in groups, resulting in decisions that, he argues, are
often better than could have been made by any single member of the

Types of crowd wisdom

Surowiecki breaks down the advantages he sees in disorganized
decisions into three main types, which he classifies as:

Cognition   –   Market judgment, which he argues can be much faster,
more reliable, and less subject to political forces than the
deliberations of experts, or expert committees

Coordination   –   Coordination of behavior includes optimizing the
utilization of a popular restaurant and not colliding in moving
traffic flows. The book is replete with examples from experimental
economics, but this section relies more on naturally occurring
experiments such as pedestrians optimizing the pavement flow, or the
extent of crowding in popular restaurants. He examines how common
understanding within a culture allows remarkably accurate judgments
about specific reactions of other members of the culture.

Cooperation   –   How groups of people can form networks of trust
without a central system controlling their behavior or directly
enforcing their compliance. This section is especially pro-free


Four elements required to form a wise crowd

Not all crowds (groups) are wise. Consider, for example, mobs or
crazed investors in a stock market bubble. Refer to Failures of crowd
intelligence (below) for more examples of unwise crowds.

According to Surowiecki, these key criteria separate wise crowds from
irrational ones:

Diversity of opinion   –   Each person should have private information
even if it’s just an eccentric interpretation of the known facts.

Independence   –   People’s opinions aren’t determined by the opinions
of those around them.

Decentralization   –   People are able to specialize and draw on local

Aggregation   –   Some mechanism exists for turning private judgments
into a collective decision.


Failures of crowd intelligence

Surowiecki studies situations (such as rational bubbles) in which the
crowd produces very bad judgment, and argues that in these types of
situations their cognition or cooperation failed because (in one way
or another) the members of the crowd were too conscious of the
opinions of others and began to emulate each other and conform rather
than think differently. Although he gives experimental details of
crowds collectively swayed by a persuasive speaker, he says that the
main reason that groups of people intellectually conform is that the
system for making decisions has a systematic flaw.

Surowiecki asserts that what happens when the decision making
environment is not set up to accept the crowd, is that the benefits of
individual judgments and private information are lost, and that the
crowd can only do as well as its smartest member, rather than perform
better (as he shows is otherwise possible). Detailed case histories of
such failures include:

Too homogenous   –   Surowiecki stresses the need for diversity within
a crowd to ensure enough variance in approach, thought process, and
private information.

Too centralized   –   The Columbia shuttle disaster, which he blames
on a hierarchical NASA management bureaucracy that was totally closed
to the wisdom of low-level engineers.

Too divided   –   The U.S. Intelligence community failed to prevent
the September 11, 2001 attacks partly because information held by one
subdivision was not accessible by another. Surowiecki’s argument is
that crowds (of intelligence analysts in this case) work best when
they choose for themselves what to work on and what information they
need. (He cites the SARS-virus isolation as an example in which the
free flow of data enabled laboratories around the world to coordinate
research without a central point of control.) Recent reports indicate
that the CIA is now planning a Wikipedia style information sharing
network that will help the free flow of information to prevent such
failures again.

Too imitative   –   Where choices are visible and made in sequence, an
“information cascade” can form in which only the first few decision
makers gain anything by contemplating the choices available: once this
has happened it is more efficient for everyone else to simply copy
those around them.

Too emotional   –   Emotional factors, such as a feeling of belonging,
can lead to peer pressure, herd instinct, and in extreme cases
collective hysteria.


“We analyze the extent to which simple markets can be used to
aggregate disperse information into efficient forecasts of uncertain
future events. Drawing together data from a range of prediction
contexts, we show that market-generated forecasts are typically fairly
accurate, and that they outperform most moderately sophisticated
benchmarks. Carefully designed contracts can yield insight into the
market’s expectations about probabilities, means and medians, and also
uncertainty about these parameters. Moreover, conditional markets can
effectively reveal the market’s beliefs about regression coefficients,
although we still have the usual problem of disentangling correlation
from causation. We discuss a number of market design issues and
highlight domains in which prediction markets are most likely to be



“In 2000 Michael Foster, who ran (and still runs) the National Science
Foundation quantum computing research program, convinced DARPA
(Defense Advanced Research Projects Agency, the blue-sky research arm
of the U.S. Defense Department) to fund research on prediction markets
starting in 2001. Prediction markets are speculative markets created
for the purpose of aggregating information on topics of interest.
Previous field studies had found that such markets out-predict co-
existing institutions regarding the weather, printer sales, movie
sales, elections, and much more.

This research program was eventually named “FutureMAP”, but the first
DARPA call for proposals went out under the name “Electronic Market-
Based Decision Support.” This call basically said “We’ve heard this
works elsewhere; show us it works for problems we care about.” The
call went out in May 2001, for proposals due in August, and by
December two firms had won SBIR (Small business independent research)
grants. The winners were Neoteric Technologies, subcontracting to
Martek and professors at the University of Iowa, and Net Exchange,
founded by a Caltech professor (John Ledyard) and subcontracting to
professors at George Mason University (myself and David Porter), and
later to the Economist Intelligence Unit. The Net Exchange project
came to be called the “Policy Analysis Market” (PAM).

The plan was for two firms to get $100K for a six month Phase I, and
after which one of them would be awarded $750,000 to continue Phase II
over two more years. There was also the possibility of applying to get
$100,000 of funding for the six months between these phases. More
money became available than initially planned, so in fall 2002 both
firms were funded to continue to Phase II, and Net Exchange applied
for and won interim funding. Also during 2002, the infamous John
Poindexter (who I have never met) became a DARPA executive, and
Foster’s FutureMAP program was placed within Poindexter’s
organization, the Information Awareness Office (IAO). In December
2002, DARPA called for proposals for related research, at this point
using the name FutureMAP. In summer 2003 a half dozen teams, at Penn
State, Metron, ICT, GMU (including me), Sparta, and BBN, were awarded
$100,000 each.

Neotek sponsored an end of phase I conference in June 2002, and showed
a few demonstration markets, using their pre-existing software, on
SARS and the color security threat level. When FutureMAP was
cancelled, Neotek had still not identified their market topics, and
had surely spent less than half of their Phase II funding. Net
Exchange spent about two thirds of their Phase II funding, and the new
small projects had spent little of their funding. Michael Foster had
asked for, but not received, $8,000,000 more in FutureMAP funding over
the next few years.

From the very start, the Net Exchange team began laboratory
experiments to study the issue of price manipulatoin, as this was a
widely expressed concern. Also from the start, we planned to focus on
forecasting military and political instability around the world, how
US policies would effect such instability, and how such instability
would impact US and global aggregates of interest. The Net Exchange
president, Charles Polk, named this the Policy Analysis Market (PAM).
We later had to narrow our focus to a smaller region, the Mideast,
because the Economist Intelligence Unit charged a high price to judge
after the fact what instability had actually occurred in each nation.

We planned to cover eight nations. For each nation in each quarter of
a year, we planned to have traders predict its military activity,
political instability, economic growth, US military activity, and US
financial involvement. In addition traders would predict US GDP, world
trade, US military casualties, and western terrorist casualties, and a
few to-be-determined miscellaneous items. This would require a hundred
or so base markets. Most important, we wanted to let our traders
predict combinations of these, such has how moving US troops out of
Saudi Arabia would effect political stability there, how that would
effect stability in neighboring nations, and how all that might change
oil prices.

For many years before PAM, Net Exchange had specialized in
combinatorial markets, where buyers and sellers can exchange complex
packages of items. So from the start of PAM, we planned to see how far
we could go in developing combinatorial prediction markets. In Phase I
Net Exchange put together a combinatorial market similar to their
previous markets, and at the end of Phase I we ran a complex
simulation where a dozen students traded over a few days for real
money. Unfortunately, only about a dozen trades occurred, a serious

In the interim phase, the Net Exchange team prepared for and ran lab
experiments comparing two new combinatorial trading mechanisms with
traditional mechanism. These experiments, where six traders set 255
independent prices in five minutes, found that a combinatorial market
maker was the most accurate. Phase II was mostly being spent
implementing a scaleable production version of this market maker. It
requires a net subsidy to traders, and so because we had budgeted
$50,000 for this subsidy, individual bets were limited to a few tens
of dollars.

We were concerned that we might not attract enough traders to achieve
a meaningful test. While we had considered running markets within
government agencies, we choose public markets due to legal problems
with conditional transfers of money between agencies and the absence
of a single agency strongly interested in collaborating. On May 20,
200, DARPA reported to congress on the IAO, and described FutureMAP in
terms of predicting a bioweapons attack against Israel. In June 2003
we began to tell people about our webpage, and to give talks drum up
interest. Charles Polk created the PAM website, wherein in the faint
background sample screen, he included as colorful examples of
miscellaneous items the assassination of Arafat, and a missile attack
from North Korea.

In the summer of 2003, the Senate but not the House had cancelled IAO
funding, which included all FutureMAP funding, because of privacy
concerns with another IAO project, “Total Information Awareness.” Due
to this funding uncertainty, when the media storm hit our plans were
to start on September 1 with one hundred testers to which we had each
given $100. Registration to be one of those testers was to open August
1, with public trading to being January 1, 2004.

The media storm hit on July 28, 2003, when two senators complained
that we were planning to let people bet on terrorist attacks. The next
morning the secretary of defense announced that FutureMAP was
cancelled. In the intervening day, no one from Congress asked us if
the accusations were correct, or if the more offending aspects could
be cut from the project. DARPA said nothing. The next day, John
Poindexter resigned, and two months later all IAO research was


Futarchy: Vote Values, But Bet Beliefs
by Robin Hanson, August 2000

This short “manifesto” describes a new form of government. In
“futarchy,” we would vote on values, but bet on beliefs. Elected
representatives would formally define and manage an after-the-fact
measurement of national welfare, while market speculators would say
which policies they expect to raise national welfare.

Democracy seems better than autocracy (i.e., kings and dictators), but
it still has problems. There are today vast differences in wealth
among nations, and we can not attribute most of these differences to
either natural resources or human abilities. Instead, much of the
difference seems to be that the poor nations (many of which are
democracies) are those that more often adopted dumb policies, policies
which hurt most everyone in the nation. And even rich nations
frequently adopt such policies.

These policies are not just dumb in retrospect; typically there were
people who understood a lot about such policies and who had good
reasons to disapprove of them beforehand. It seems hard to imagine
such policies being adopted nearly as often if everyone knew what such
“experts” knew about their consequences. Thus familiar forms of
government seem to frequently fail by ignoring the advice of relevant
experts (i.e., people who know relevant things).

Would some other form of government more consistently listen to
relevant experts? Even if we could identify the current experts, we
could not just put them in charge. They might then do what is good for
them rather than what is good for the rest of us, and soon after they
came to power they would no longer be the relevant experts. Similar
problems result from giving them an official advisory role.

“Futarchy” is an as yet untried form of government intended to address
such problems. In futarchy, democracy would continue to say what we
want, but betting markets would now say how to get it. That is,
elected representatives would formally define and manage an after-the-
fact measurement of national welfare, while market speculators would
say which policies they expect to raise national welfare. The basic
rule of government would be:

When a betting market clearly estimates that a proposed policy would
increase expected national welfare, that proposal becomes law.

Futarchy is intended to be ideologically neutral; it could result in
anything from an extreme socialism to an extreme minarchy, depending
on what voters say they want, and on what speculators think would get
it for them.

Futarchy seems promising if we accept the following three assumptions:

* Democracies fail largely by not aggregating available
* It is not that hard to tell rich happy nations from poor
miserable ones.
* Betting markets are our best known institution for aggregating

GDP is today the most common measure of national wealth. It seems hard
for frequent travelers to escape the impression that people in high
GDP nations tend to be richer and better off than those in low GDP
nations. Economists thus tend to be willing to recommend policies that
macroeconomic data suggest are causally related to increasing GDP. It
seems that it is not that hard to, after the fact, tell rich satisfied
nations from poor miserable ones. GDP may be good enough, and with the
full attention of our elected representatives, we should be able to do
even better, such as by including happiness, inequality, health,
leisure, and environment measures.

If we can measure how rich nations are, we can use such measurements
to settle bets. This is good because betting markets, and speculative
markets more generally, seem to do very well at aggregating
information. To have a say in a speculative market, you have to “put
your money where your mouth is.” Those who know they are not relevant
experts shut up, and those who do not know this eventually lose their
money, and then shut up. Speculative markets in essence offer to pay
anyone who sees a bias in current market prices to come and correct
that bias.

Speculative market estimates are not perfect. There seems to be a long-
shot bias when there are high transaction costs, and perhaps also
excess volatility in long term aggregate price movements. But such
markets seem to do very well when compared to other institutions. For
example, racetrack market odds improve on the predictions of racetrack
experts, Florida orange juice commodity futures improve on government
weather forecasts, betting markets beat opinion polls at predicting
U.S. election results, and betting markets consistently beat Hewlett
Packard official forecasts at predicting Hewlett Packard printer
sales. In general, it is hard to find information that is not embodied
in market prices.

A betting market can estimate whether a proposed policy would increase
national welfare by comparing two conditional estimates: national
welfare conditional on adopting the proposed policy, and national
welfare conditional on not adopting the proposed policy. Betting
markets can produce conditional estimates several ways, such as via
“called-off bets,” i.e., bets that are called off if a condition is
not met.