Telecommuting, Serendipity, and Innovation
Does proximity spur collaboration? A new study finds it does
Steven Cherry: Hi, this is Steven Cherry for IEEE Spectrum’s “Techwise Conversations.”
Yahoo’s CEO, Marissa Mayer, ignited a firestorm earlier this year when she ordered the company’s entire workforce back into the office, banning telecommuting throughout the organization. Other CEOs rushed to condemn her action, and only a handful of management experts defended it in the interests of spurring innovation.
One was John Sullivan, who on this show pointed out that Google—the company Mayer worked at for many years—doesn’t allow telecommuting and works hard to ensure workers come together, in the hallways, at the coffee bar, and in the cafeteria, to spur their innovative collaborations. Sullivan said,
And the Google data shows, incidentally,…its interaction between engineers and finance, engineers and design, engineers and production or marketing. Those are where the interactions—they’re called “serendipitous interactions” or “casual interactions”—that’s where the collaboration and the innovation comes from.
It turns out, though, we don’t know the first thing about such interactions, because at best we study them in terms of the physical distance of two workers who might productively interact. But physical distance isn’t the right thing to measure. You can work in an office next to the most popular coffee bar in the corporate campus and never interact with anyone because you keep your office door closed all day.
That’s an example given in a new working paper from the University of Michigan’s Ross School of Business, entitled “Zone Overlap and Collaboration in Academic Biomedicine.” Its authors developed a notion they call the "functional zone” and found that pairs of workers are more likely to collaborate the more their paths and areas overlap.
And so, here to talk about casual interactions, is a man who’s actually studied them. Jason Owen-Smith is an associate professor of sociology and organizational studies at the University of Michigan and one of the four coauthors of the “Zone Overlap and Collaboration” paper. He joins us by phone.
Jason, welcome to the podcast.
Jason Owen-Smith: Thanks for having me.
Steven Cherry: A recent article in The Wall Street Journal mentioned that Google is designing a new headquarters and quoted a spokesperson as saying, “We want it to be easy for googlers to collaborate and bump into each other.” Is it as simple as that? Make sure people bump into one another?
Jason Owen-Smith: I don’t think so. There are many things that happen when people do bump into one another. I think of these kinds of serendipitous or passive contacts, when you bump into someone enough that you come to recognize them and eventually nod at them or talk to them and converse with them. Sometimes information passes between you that you didn’t expect to find that can be valuable. Those kinds of information transfers are unlikely to happen if people aren’t bumping into each other, but the step from people bumping into each other to potentially valuable products and services for a company is a long one, and we’re not entirely sure what the process is that goes from one to the other. So, maybe necessary, but I don’t know that it’s efficient.
Steven Cherry: So let’s look at the research. You studied two different buildings in a medical center. What did you measure?
Jason Owen-Smith: So we’re interested in the question of how physical proximity leads people who are in interdisciplinary translational biomedicine, the kinds of people who are actively trying to make fundamental science discoveries that will have clinical applications at the bedside. We’re interested in knowing how folks in those fields, who have not collaborated before, come to collaborate, and when those collaborations are successful.
As a first step to that, we became interested in the question of how physical space and proximity works. And we made several assumptions about people. We were frustrated by what is a very mixed set of results about sheer linear distance. There’s some findings going back to the 1970s that suggest that collaboration is less likely if you’re more than 30 yards separated from each other, but there are a lot of different configurations that one can be separated by 30 yards, and some seem more challenging than others. So we set out to try to figure out how people actually occupy their buildings. We made a set of basic assumptions. The first was that people really don’t occupy the full building. Most of us really occupy a fairly small component of our buildings. We then assumed that the zones, the territories people occupy, are anchored, if you will, by a set of basic spaces that they traverse.
In the life sciences for researchers, we defined this as their office, their laboratory, the nearest relevant bathroom, and the nearest elevator. We’re doing some empirical work now to figure out what other spaces would be important: food, coffee bars, shared equipment and services, things of that nature. We then made the assumption that people are basically, fundamentally, fairly lazy. And so we assume that people traverse the paths among those four spaces in the course of their day using the shortest possible route. And so for every person in each of these buildings, we identified these four spaces and identified the shortest path among them, and we called that their “territory.” For every pair of people, we then identified how much of that territory overlapped.
So your territory could overlap completely, like every path you walk could be the same as someone else. It could overlap not at all. You might never see someone who works in your building because your territories don’t overlap. And we tried to use that distance to explain statistically when a new collaboration formed, which we defined using administrative data, as observing for the first time a joint application to use human subjects, animal subjects, or a proposal sent to an external granting agency, and conditional on having started a new collaboration, whether the collaboration got funded, got a funded grant, within a period of three years, I believe.
Steven Cherry: So what did you find?
Jason Owen-Smith: What we found was, I think, pretty compelling. In terms of forming new collaborations, we found that an increase of 100 feet in path overlap—so if the paths I walk every day overlap with yours by 200 feet rather than 100—all other things being equal, you and I are somewhere around 17 percent more likely in our models to form a new collaboration.
Now, this happens on a low base rate. So this is sort of like thinking about incidents of disease. There are relatively few new incidents of collaborations formed, but the likelihood of this relatively rare thing increases with this small increment in overlap. We also found that if you form a new collaboration, the same 100 feet of overlap in space makes it about 20 percent more likely that you’ll get a grant funded. And so our preliminary findings suggest, we think quite strongly, but we’re working to nail them down, that this type of overlap that enables these kinds of passive contacts results both in a greater likelihood of collaboration and a greater likelihood of early success in these fields.
Steven Cherry: After World War II, MIT’s Building 20 became a hotbed of innovation, in part because of a hodgepodge of academic disciplines [that] were housed there serendipitously. High-speed photography, Chomskyan linguistics, the hacker culture that came out of the Tech Model Railroad Club, and the Bose Corporation are just a few things that came out of it. Have we studied much the who? Who’s having these serendipitous encounters?
Jason Owen-Smith: We have not. We’re beginning to work toward that. We’re doing some survey work now. The University of Michigan just purchased and is populating what used to be Pfizer’s global headquarters. It’s now called the North Campus Research Center. [Editor’s note: The correct name is North Campus Research Complex] And this is an effort to mindfully colocate a bunch of once-dispersed interdisciplinary groups in a single research campus, north of the main campus of the University of Michigan. And there we will know much more about where people came from, what their research is, what their interests are, those kinds of things.
This is actually a remarkably difficult thing to study, in part because if you think about it, there are many different registers in which one can understand distance. There’s physical distance. There’s social distance. You and I may have offices separate from one another by 100 miles, but if our best friend is someone we share in common, we’re actually fairly socially proximate. There’s a sort of cultural or, if you will, institutional difference. So it may be that it’s simply harder from—I make it up—sociology and high-energy physics to talk with each other, than it is for people from high-energy physics and electrical engineering to talk with each other. There’s organizational distance. We know, for instance, that folks lower down the totem pole in organizations are less likely to initiate conversations with folks higher up, much higher up, the totem pole than them in many organizations.
And so the question of who, and of the idiosyncratic patterns that underpin the way space is occupied, are going to prove very interesting and very challenging for researchers in our fields for a long time.
Steven Cherry: The sociologist and inventor Hugh Loebner once told me that if you want to have a successful party, you put all the food at one end of the room and all the drinks at the other. What are some of the things that you think companies can do to encourage serendipitous interactions?
Jason Owen-Smith: Well, there, this is, again, one of the things where I have ideas, but we’re just beginning to do the research. So, the short form is, thinking systematically about it, what we think those passive contacts that are created by overlap do is actually serve two purposes. One purpose is what we call “prospecting.” It’s the idea that these passive contacts bumping into people enables you to come into contact with information or resources you wouldn’t otherwise think to seek out. So that’s the sense of serendipity. The people you bump into may know stuff that is useful to you that you don’t know would be useful to you, and you may learn it in the course of these interactions.
So they’re creating an organizational arrangement, either through the design or through the allocation of space, that facilitates interactions among disparate groups that otherwise wouldn’t have a reason to seek each other out, may seed these kinds of innovations. That’s the sense of creating a space where the food’s at one end and the drinks are at the other, because what that does is enforce mixing. It actually, if you will, creates a bit of friction. It makes it less efficient to get what you need at the party, but the benefit of that lack of efficiency in gaining food or drink is increased contact and conversation among guests.
That’s different than what we think the finding that leads to greater success in collaborations is. There we call the idea “mobilization.” There what we’re interested in is when you know you have a project and you know you have to attend to each other, and you’re trying to do something difficult and uncertain in an early stage of innovation. Bumping into each other a lot allows for a lot of quick, soft touches during the day: “Oh, I just remembered I have this little technical issue.” Or “Could you take a minute and help me troubleshoot this assay.” Those kinds of interactions are facilitated by proximity, because proximity makes it more likely that you will bump into people that you already know.
So I think the real challenge for an organization is to decide whether they want to create a space that will facilitate, that will raise, the transactional frictions within the established groups in order to enable greater contact between established groups, or whether they will try to create a space that will allow established groups to more efficiently go about their collaborative work. Those two things are likely to be trade-offs that come through the same physical arrangement of space.
And there the key is to figure out what draws people to move around the building at a given moment in a given day habitually, and then to position the things that make people move to them in a fashion that will draw together in a routine, habitual way the people that you think should be talking to one another. Now, that might be food. It might be drink. It might be a shared microscope or a group printer. Like I said, we assumed it was bathrooms, offices, labs, and elevators. We’re doing some work now with motion-tracking data for a group of 47 engineers to try to actually understand through observation over a six-week period what the paths people actually take through their building are.
Steven Cherry: So if I understand you, you’re sort of distinguishing between two types of innovation. There’s sort of the initial spark that gets struck, and then there’s the sort of ongoing fanning the flames of an existing project.
Jason Owen-Smith: Yes. There’s the aha moment that may lead to a collaboration. And then there’s the process of making a new collaboration—to do something that hasn’t been done before work.
Steven Cherry: And I’m glad you brought up friction. It seems there really is, at times at least, a tension between productivity on the one hand and innovation on the other. And that was sort of the issue, getting back to Marissa Mayer. She wanted more innovation, even if, I guess, it came at the cost of some productivity. What’s your opinion? Do you think she was right to ban telecommuting?
Jason Owen-Smith: My native sense then—this is more an intuition than anything else—is that to spur innovation that is unexpected, you need some form of physical proximity. Because virtual work, to the best of my knowledge, does not yet have mechanisms that allow you to come to realize that someone you don’t have a reason to talk to knows something you need to know.
So if you know and can articulate your needs, you might be able to use distance-enabled technology to search for potential collaborators who know what you need to know and might work with you. But if what the company is trying to bet on is unexpected, serendipitous discoveries that lead them in new directions, then, yes, it seems like a probably, a very smart decision.
Steven Cherry: Well, Jason, it’s an important area that we’re just starting to give the attention it deserves. Thanks for studying it, and thanks for joining us today.
Jason Owen-Smith: Thanks for your time. It’s been fun.
Steven Cherry: We’ve been speaking with University of Michigan sociologist Jason Owen-Smith about the ways that physical proximity do, and don’t, encourage innovation.
For IEEE Spectrum’s “Techwise Conversations," I’m Steven Cherry.
This interview was recorded Wednesday, 5 June 2013.
Segment producer: Barbara Finkelstein; audio engineers: Celia Gorman and Francesco Ferorelli
Read more “Techwise Conversations,” find us in iTunes, or follow us on Twitter.
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.