In a study published today in the Journal of Human-Animal Interaction, Kane & Eines (2002) report
on research that suggests a new algorithm for identifying points of maximal betweenness centrality
in systems of human foot traffic flow: dogs.
“Identifying points of high betweenness centrality in human pedestrian traffic is an important consideration
in designing the layout of offices, homes and dormitories, and public spaces,” co-author Robert Eines points
out, “Information about pedestrian flow is crucial to design hallways that are sufficiently wide, floor plans that
disperse traffic to avoid concentrating people in congested spaces, reduce wear on carpet and floors, and
Any dog owner has probably noticed how often man’s best friend is underfoot. It turns out this is not just
coincidence or perception. Somehow dogs are capable of identifying the points of maximum betweenness
centrality in a wide range of human pedestrian traffic flows. Kane & Eines realized that what dog owners saw
as a minor nuisance could be a valuable diagnostic tool in the design of building layout. “I was watching my
Beagles one day,” says Barbara Kane, “and it occurred to me that if the dogs could do it in my home, why
not in offices?” Her first step was to confirm that the places dogs occupied in her home were, in fact, the
crucial points in terms of traffic flow. Intrigued, she measured the amount of traffic between rooms in her
home for two weeks and analyzed the data using UCINET, a computer program for the analysis of social
networks. “The results were more striking even than I imagined,” she says. The places her dogs preferred
to stay, measuring according to the amount of time either of her dogs spent there, “had an almost perfect
rank-order correlation with the betweenness centrality scores. They love the landing on the stairs!”
Would what worked in her home work in other settings? It took some convincing to find offices that would
allow her to bring in dogs for several days to test her hypothesis. Her break came when she found a nursing
home that thought the dogs might be of therapeutic value. That became her first test. “I rounded up the dogs
from all my friends,” she recalls, “we had twenty-three dogs in two Dodge Caravans! We brought them to the
nursing home. What a scene. But the results were very supportive, and the Seniors really enjoyed ‘adopting’
the dogs for a week.” The authors also learned a key lesson. The general pattern was always disrupted at
mealtimes because the dogs tended to congregate in the cafeteria whenever food was being served.
That initial experience attracted the interest of Robert Eines, an architect who specializes in human traffic
Flow. He heard about the study from his mother, a resident in the nursing home. Together, and with the
data from the nursing home to their credit, they persuaded a few companies to invite the dogs into their offices.
As reported in today’s article, they found an incredibly strong rank order correlation (.94) between the amount
of time spent by dogs in any spot and its betweenness centrality score in the human traffic flow as computed
mathematically. This has been true in each of the several offices studied, other dormitory facilities, individual
homes, and wherever the dogs have been brought in while there was human foot traffic going on. They also
report a slight tendency for larger dogs occupy the most central locations for the longer periods of time than
smaller dogs do.
Why does this work? “We are not sure,” the authors admit, “if it is a response to human traffic, then it
should not work in vacant settings.” Kane & Eines are in the process of testing that possibility, and so
far the results suggest the tendency, while still present, is considerably weaker in the absence of people.
“We have noticed it takes the dogs a day or two to find the absolute best positions, even when people are
present,” Eines says, “Together with the much weaker results when people are not present, this leads us
to believe that the dogs’ ability is a reaction to human traffic patterns, rather than an innate sense based
on the building structure.”
If people must be present for the dogs to work, their use for diagnostic purposes may be slightly less
attractive, but Kane & Eines are now exploring the possibility that the dogs could be used commercially
to assist in the design process or in the reconfiguration of existing office layouts. Currently the dogs can
only “measure” on a rank-order basis, a limitation which the authors’ are concerned with trying to overcome.
Even with that limitation and the slight inconvenience of having the dogs around for a few days, floorplan
designers and building managers may be interested in the dogs as a faster and cheaper way of measuring
human traffic patterns and identifying points of high betweenness centrality. “Its easier and cheaper than
hiring a social network analyst,” one office executive noted in the article.
“We are familiar with police and security dogs, guide dogs for the blind -- this could be a whole new line of
work for our furry friends,” Kane says, “perhaps in the future it will be commonplace to see dogs ‘working’
in offices helping us make traffic flow more efficient.”
April F. Laffer