Rationality doesn’t equal efficiency: Cellphone data shows how we navigate cities

by

The paths peo­ple take are record­ed by their cell­phones. Anony­mous data from thou­sands of phones shows the paths peo­ple take in Boston (above) and San Fran­cis­co (below). Car­lo Rat­ti, CC BY-ND

Think of your morn­ing walk to work, school or your favorite cof­fee shop. Are you tak­ing the short­est pos­si­ble route to your des­ti­na­tion? Accord­ing to big data research that my col­leagues and I con­duct­ed, the answer is no: People’s brains are not wired for opti­mal navigation.

Instead of cal­cu­lat­ing the short­est path, peo­ple try to point straight toward their des­ti­na­tions – we call it the “pointi­est path” – even if it is not the most effi­cient way to walk.

As a researcher who stud­ies urban envi­ron­ments and human behav­ior, I have always been inter­est­ed in how peo­ple expe­ri­ence cities, and how study­ing this can tell researchers some­thing about human nature and how we’ve evolved.

Chasing down a hunch

Long before I could run an exper­i­ment, I had a hunch. Twen­ty years ago, I was a stu­dent at the Uni­ver­si­ty of Cam­bridge, and I real­ized that the path I fol­lowed between my bed­room at Dar­win Col­lege and my depart­ment on Chaucer Road was, in fact, two dif­fer­ent paths. On the way to Chaucer, I would take one set of turns. On the way back home, another.

Sure­ly one route was more effi­cient than the oth­er, but I had drift­ed into adapt­ing two, one for each direc­tion. I was con­sis­tent­ly incon­sis­tent, a small but frus­trat­ing real­iza­tion for a stu­dent devot­ing his life to ratio­nal think­ing. Was it just me or were my fel­low class­mates – and my fel­low humans – doing the same?

Around 10 years ago, I found tools that could help answer my ques­tion. At the Senseable City Lab at the Mass­a­chu­setts Insti­tute of Tech­nol­o­gy, we were pio­neer­ing the sci­ence of under­stand­ing cities by ana­lyz­ing big data, and in par­tic­u­lar dig­i­tal traces from cell­phones. Study­ing human mobil­i­ty, we noticed that, on the whole, people’s routes were not con­ser­v­a­tive, mean­ing they did not pre­serve the same path from A to B as the oppo­site direc­tion, from B to A.

How­ev­er, the tech­nol­o­gy and ana­lyt­i­cal meth­ods of that time pre­vent­ed us from learn­ing more – in 2011, we could not reli­ably tell a pedes­tri­an apart from a car. We were close, but still a few tech­no­log­i­cal steps short of tack­ling the enig­ma of human nav­i­ga­tion in cities.

Big cities, big data

Today, thanks to access to data sets of unpar­al­leled size and accu­ra­cy, we can go fur­ther. Every day, everyone’s smart­phones and apps col­lect thou­sands of data points. Col­lab­o­rat­ing with col­leagues at the MIT Depart­ment of Brain and Cog­ni­tive Sci­ences and oth­er inter­na­tion­al schol­ars, we ana­lyzed a mas­sive data­base of anonymized pedes­tri­an move­ment pat­terns in San Fran­cis­co and Boston. Our results con­sid­er ques­tions that my young self at Cam­bridge didn’t know to ask.

After we ana­lyzed pedes­tri­an move­ment, it became clear that I am not the only one who nav­i­gates this way: Human beings are not opti­mal nav­i­ga­tors. After account­ing for pos­si­ble inter­fer­ence from peo­ple let­ting Google Maps choose their path for them, our analy­sis of our big data sets fueled sev­er­al inter­con­nect­ed discoveries.

First, human beings con­sis­tent­ly devi­ate from the short­est pos­si­ble path, and our devi­a­tions increase over longer dis­tances. This find­ing prob­a­bly seems intu­itive. Pre­vi­ous research has already shown how peo­ple rely on land­marks and mis­cal­cu­late the lengths of streets.

Our study was able to go a step fur­ther: devel­op­ing a mod­el with the capa­bil­i­ty to accu­rate­ly pre­dict the slight­ly irra­tional paths that we found in our data. We dis­cov­ered that the most pre­dic­tive mod­el – rep­re­sent­ing the most com­mon mode of city nav­i­ga­tion – was not the quick­est path, but instead one that tried to min­i­mize the angle between the direc­tion a per­son is mov­ing and the line from the per­son to their destination.

This find­ing appears to be con­sis­tent across dif­fer­ent cities. We found evi­dence of walk­ers attempt­ing to min­i­mize this angle in both the famous­ly con­vo­lut­ed streets of Boston and the order­ly grid of San Fran­cis­co. Sci­en­tists have record­ed sim­i­lar behav­iors in ani­mals, which are described in the research lit­er­a­ture as vec­tor-based nav­i­ga­tion. Per­haps the entire ani­mal king­dom shares the idio­syn­crat­ic ten­den­cies that con­fused me on my walk to work.

Evolution: From savannas to smartphones

Why might every­one trav­el this way? It’s pos­si­ble that the desire to point in the right direc­tion is a lega­cy of evo­lu­tion. In the savan­na, cal­cu­lat­ing the short­est route and point­ing straight at the tar­get would have led to very sim­i­lar out­comes. It is only today that the stric­tures of urban life – traf­fic, crowds and loop­ing streets – have made it more obvi­ous that people’s short­hand is not quite optimal.

Still, vec­tor-based nav­i­ga­tion may have its charms. Evo­lu­tion is a sto­ry of trade-offs, not opti­miza­tions, and the cog­ni­tive load of cal­cu­lat­ing a per­fect path rather than rely­ing on the sim­pler point­ing method might not be worth a few saved min­utes. After all, ear­ly humans had to pre­serve brain pow­er for dodg­ing stam­ped­ing ele­phants, just like peo­ple today might need to focus on avoid­ing aggres­sive SUVs. This imper­fect sys­tem has been good enough for untold generations.

How­ev­er, peo­ple are no longer walk­ing, or even think­ing, alone. They are increas­ing­ly wed­ded to dig­i­tal tech­nolo­gies, to the point that phones rep­re­sent exten­sions of their bod­ies. Some have argued that humans are becom­ing cyborgs.

This exper­i­ment reminds us of the catch: Tech­no­log­i­cal pros­the­ses do not think like their cre­ators. Com­put­ers are per­fect­ly ratio­nal. They do exact­ly what code tells them to do. Brains, on the oth­er hand, achieve a “bound­ed ratio­nal­i­ty” of “good enoughs” and nec­es­sary com­pro­mis­es. As these two dis­tinct enti­ties become increas­ing­ly entan­gled and col­lide – on Google Maps, Face­book or a self-dri­ving car – it’s impor­tant to remem­ber how they are dif­fer­ent from each other.

Look­ing back on my uni­ver­si­ty days, it is a sober­ing thought that humanity’s bio­log­i­cal source code remains much more sim­i­lar to that of a rat in the street than that of the com­put­ers in our pock­ets. The more peo­ple become wed­ded to tech­nol­o­gy, the more impor­tant it becomes to make tech­nolo­gies that accom­mo­date human irra­tional­i­ties and idiosyncrasies.

– An archi­tect and engi­neer by train­ing, Pro­fes­sor Car­lo Rat­ti teach­es at the Mass­a­chu­setts Insti­tute of Tech­nol­o­gy, where he directs the Senseable City Lab. This arti­cle was orig­i­nal­ly pub­lished on The Con­ver­sa­tion.

News in Context: