Communications - February 2011

formation? What does it mean for you
to send me information?”
Many differences exist between
classical and quantum information,
Sudan says, notably that classical in-
formation can be copied and quan-
tum information, by definition, can-
not. Sudan is not merely interested in
discovering fundamental principles;
he wants to figure out which ones have
practical importance. “The next level
is going to focus on ‘How do I under-
stand this information? Where did it
come from? And how can I manipu-
late it in ways that are convenient?’ ”
he says.

Researchers are hoping to apply information theory to fields beyond communications and computing. Christopher Sims, an economist at Princeton, applies Shannon’s insights about channel capacity to how consumers respond to economic information. In theory, if the central bank alters interest rates or the money supply, prices should change quickly, but in reality they don’t. Sims says that’s because people have limited physical capacity to process all the data, so they usually don’t act on the information until it crosses some threshold, such as appearing on the Yahoo! homepage. They’re striking a balance between the cost of processing information and the

“We cannot even
understand how
much information
is transmitted
on the internet,”
says Wojciech
szpankowski,
“because we don’t
understand the
temporal aspect
of information.”

reduction in uncertainty—the payoff
in tracking small interest rate fluctua-
tions isn’t worth the effort it takes to
react to them. Sims has dubbed this
strategy “rational inattention.”
Sims is hoping to combine informa-
tion theory with control theory, which
looks at the behavior of dynamic sys-
tems, and come up with new economic
insights. Perhaps, he says, some of
those insights will feed back into engi-
neering. “There’s a long way to go here,
and we’ve only just begun to get infor-
mation theorists interested in what
we’re doing here,” Sims says. “We’re
still using information theory in a rela-
tively primitive way.”
“I’m hoping in the first five years
we’ll make some advances,” Sz-
pankowski says. “We will at least for-
mulate the right questions.”

Algorithmic Entomology

network software engineers have long found inspiration in ant colonies, whose collective wayfinding strategies have shed light on the problems of routing data across a busy network. in recent years, ant colony optimization has emerged as a proven algorithm for network optimization, one of several swarm intelligence models based loosely on the behavior of nature’s social animals.

now, a recent study from the University of sydney suggests that ant colonies may possess even greater problem-solving abilities than previously thought. The study, “optimization in a natural system: argentine ants solve the Towers of hanoi,” published in The Journal of Experimental

Biology, demonstrates an ant colony’s ability to adapt its navigational “algorithm” in response to changing environmental conditions. The findings may help open up new research avenues for optimizing the flow of traffic across data networks.

The international team of researchers enlisted a colony of argentine ants to solve the famous Towers of hanoi problem in which participants must find an optimal solution for arranging disks of varying sizes onto a set of three rods.

Translating the logic of
the puzzle into a maze with
32,768 possible pathways to a
food source, the researchers
turned the ant colony loose
for one hour, then abruptly
blocked some of the pathways.
in response, the ants swiftly
recalibrated their methods,
and soon found an optimal new
route to the food supply.