Smith Journal – January 2019

035 SMITH JOURNAL

THERE’S A FAMOUS STORY ABOUT
THE WISDOM OF CROWDS. IT GOES
LIKE THIS.

.........................................

In 1907, the great English polymath
Francis Galton attended a county fair.
The locals were running a competition to
guess the weight of a prize ox (no-one said
it was an exciting county fair). Galton, ever
the numbers nerd, started crunching the
data, and found that while people’s individual
guesses were all wrong, their average (1,197
pounds) was improbably close to the ox’s
actual weight (1,198 pounds). This stumped
him. A group of strangers, none of whom
knew the ox personally, had more or less
arrived at the correct answer. What the
hell was going on?

He didn’t know it at the time, but what
was going on was ‘collective intelligence’, a
phenomenon by which solutions to complicated
problems can be arrived at by ordinary folk in
great numbers. The idea lurked on the fringes
of scientific thinking during the 20th century,

but it has recently gained traction in the

so-called Information Age, when group-based

projects like Wikipedia abound. In fact, it is

now seen as such a compelling idea that the

U.S. intelligence community’s research wing,

the Intelligence Advanced Research Projects

Activity (IARPA), is currently running a

$100 million contest to see if scientists can

unlock the cognitive power of crowds.

Four teams have since been recruited by

IARPA to fill in a set of intelligence gaps

using collective intelligence. Among those

chosen is a small research division at the

University of Melbourne known as the

SWARM Project, an acronym that stands for

‘Smartly-assembled Wiki-style Argument

Marshalling’. SWARM’s is a relatively

small team, with just a handful of principal

investigators, but U.S. intelligence clearly

thinks they’re on to something. In fact,

IARPA recently granted SWARM $ 19

million to attempt something radical:

crowdsource antiterrorism solutions.

SWARM’s spokespeople employ slightly

less direct language when explaining their

goals. “We’re trying to improve general

human thinking,” says the team’s co-leader,

cognitive scientist and philosopher Dr Tim

van Gelder. “There’s a long list of areas

of human activity where good thinking

is required,” he says, and antiterrorism

is just one of them. But he more or less

agrees with the crowdsourcing angle. “Our

approach is collaboration; we’re interested

in getting accurate and well-reasoned

results from large groups.”

To do this, van Gelder and his colleagues

built something called an ‘arguwiki’ – a cloud-

based program that allows groups of people

to chat, exchange ideas, debate, argue and

propose solutions to various problems. Teams

of 20 to 40 laypeople are drawn from the

general public, presented with hypotheticals,

then asked to write intelligence reports and

recommendations. Librarians, baristas,

lawyers, tradies and hairdressers were

suddenly working together to analyse

complex national security scenarios.

>>

what the bees know

THEY’VE LONG BEEN DISMISSED AS PITCHFORK-WIELDING MOBS.

BUT CROWDS HARBOUR HIDDEN TALENTS – AND SOME AUSTRALIAN

RESEARCHERS ARE JUST BEGINNING TO TEST THEIR POTENTIAL.

Writer James Shackell