Need to Solve a Problem? Take a Break From Collaborating

Organizations spend a lot of money enabling employees to solve problems collectively. But inducing more collaboration may actually hinder the most important part of problem-solving: actually solving the problem. Research by Jesse Shore, Ethan Bernstein, and David Lazer.
by Carmen Nobel

Got a problem? Throw some collaboration software at it.

It's a common strategy among today's managers: Organizations spend a lot of money on technology that enables employees to tackle problems collectively. Hence, the market is rife with connectivity tools and services such as's Chatter, Microsoft's Yammer, and Jive Software's suite of namesake products. The global enterprise social software market is forecast to grow to $8.14 billion in 2019, according to MarketWatch, from $4.77 billion in 2014.

“When it comes to solving problems, connectedness is a double-edged sword”

Unfortunately, it turns out that inducing more collaboration may hinder the most important part of problem-solving: actually solving the problem. While connecting employees does increase the ability to gather facts during the early stages of tackling a problem, it also inhibits the ability to analyze those facts and find a solution, a team of Boston researchers reports.

"When it comes to solving problems, connectedness is a double-edged sword," says Harvard Business School Assistant Professor Ethan Bernstein, co-author of Facts and Figuring: An Experimental Investigation of Network Structure and Performance in Information and Solution Spaces, with Jesse Shore, an assistant professor at Boston University's Questrom School of Business, and David Lazer, a political science professor at Northeastern University. The paper appears in the journal Organization Science.

A High-stakes Whodunit Game

Previous academic research generally has focused separately on two aspects of problem-solving—either gathering facts or figuring out solutions. And the findings often have contradicted each other.

Collaboration may hinder the
solution-finding stage of problem-solving.© iStockPhoto/Rawpixel

"Substantial recent research implies that clustering—the degree to which people with whom a person is connected are themselves connected to each other—can improve problem-solving performance by increasing coordination," the authors write in the "Facts and Figuring" paper. "By contrast, equally powerful research suggests that clustering can undermine performance by fostering an unproductive imbalance between exploration and exploitation, even for simple tasks."

Thus, Bernstein, Shore, and Lazer set out to figure out how collaborating could both help and hurt the problem-solving process. To pursue the efficacy of collaboration the research team developed a straightforward experiment that mirrored real-world problem-solving work.

Rather than start from scratch, they customized a platform called ELICIT (Experimental Laboratory for Investigating Collaboration, Information-sharing, and Trust), developed by the US Department of Defense's Command and Control Research Program. In the DoD's high-stakes "whodunit" game, players try to solve several aspects of an imaginary pending terrorist attack: the identity of the terrorists, the target of the attack, and where and when the attack would happen.

In the experiment, players tried to solve several aspects of an imaginary pending terrorist attack: the identity of the terrorists, the target of the attack, and where and when the attack would happen.

For the study, the researchers hired 417 participants to play the game. Players received two clues at the start of each round and were allowed to search for more clues once per minute; they had 25 minutes to solve the problem. (The experiment took place at Harvard Business School's Computer Lab for Experimental Research.)

Participants were randomly assigned to one of 70 16-person networks, some of which were more interconnected—or "clustered," in academic parlance—in terms of who could share information with whom during the game. "In the most-clustered conditions, people were connected in a clear team structure," Shore explains. "In the least clustered, nobody's partners were also partners with each other."

Participants received 60¢ for each minute (equivalent to $36 per hour) that they had the correct answer registered before the clock ran out—a considerable incentive, considering the majority of the participants were undergraduates who needed pocket money. "Generally speaking, these were students from northeastern universities, all of whom are relatively smart, so we're not talking about the average population," Bernstein says. "But actually, that's probably good for us, because the average population is not who would be hired to solve intelligence problems for the Department of Defense."

In terms of seeking out unique facts, the results showed that the most-clustered groups gathered 5 percent more information than the least-clustered groups, because clustering prevented network members from unknowingly conducting duplicative searches. "By being in a cluster, individuals tended to contribute more to the collective exploration through information space—not from more search but rather by being more coordinated in their search," the researchers write.

However, clustering also seemed to inhibit the breadth and number of answers that the players proposed. The least-connected networks came up with 17.5 percent more theories and solutions than did the most-connected networks. Less clustering also increased the likelihood of correct solutions in that those in very clustered positions were more likely to copy an incorrect theory from a neighbor than their less-clustered counterparts.

"We realized that the network structure seemed to have opposite effects for searching for information and searching for solutions," Shore says. "That was sort of the 'aha' moment."

Implications For The Workplace

The team's findings highlight the pros and cons of ever-present connectivity. On the one hand, tools like collaboration software can facilitate coordinated information gathering at both a local and a global scale. On the other hand, too much connectivity can lead to a premature consensus, to the detriment of organizations that aim to transform information into knowledge. Fostering effective problem-solving requires flexibility—flexible technology, flexible employees, and a flexible workspace.

"The optimal communication and collaboration structures change over time," Shore says. "To complicate matters, in longer-term projects you're going to be iterating between information gathering and solution generating. So the organizational structure needs to switch back and forth to facilitate that."

Bernstein notes that some firms do that already. "These results seem to make a lot of sense to organizations like The Boston Consulting Group or McKinsey & Company, which deliberately use very different structures for different functional stages of a project," he says. "Problem-solving looks different from other stages of a project in which the team is defining the problem, gathering data, or synthesizing results."

In terms of how to design corporate offices, the findings in "Facts and Figuring" provide rationale for the hot architectural trend of "agile" workspaces, including a mix of open floors, private spaces, and movable surfaces. "Architects have designed our workspaces flexibly so that we can change them and use them in different ways," Bernstein says. "But nobody has trained us on how to use them. We need to think about how to use our spaces wisely."

For knowledge-intensive companies, the findings highlight the need for enterprise software that addresses the disparate aspects of problem-solving.

"Another organizational response would be to design communications infrastructures that could somehow separate facts from figuring and adopt differently-structured communication networks for each category," the authors write. "In other words, rather than allow the march of technology to dictate organizational performance, it is possible to imagine technology being harnessed to achieve different performance goals. Even without the separation of facts and figuring, the results of this study are likely to be especially relevant for computer-mediated problem-solving because of the ease of manipulating the structure in which participants communicate."

For managers, the findings highlight the need to determine from the get-go whether a problem-solving task requires a search for facts or a search for answers, and then, if possible, tackle the problem accordingly—enforcing collaboration only where it makes sense. "That also means that their collaboration tools, and the policies that others set around them, need to have an off switch—to be used very selectively, but at the discretion of those at the front line of problem-solving," Bernstein says.

And for academics, the research provides a good jumping-off point for future studies about physical and virtual networking. In their next stage of research, the team plans to conduct several field studies in real-world office environments as well as further experimental studies on digital collaboration. In the meantime, "Facts and Figuring" helps to explain how knowledge works, both in the workplace and in the classroom.

"Students sometimes want you just to tell them the answer—like 'just give me the answer to leadership!'" says Bernstein, who co-teaches the required Leadership and Organizational Behavior course to first-year MBA students at HBS. "We have never done it that way. We want to give them the facts, and we do, but we want them to explore their own individual solutions based on those facts. Because ultimately, that's going to get the field of study further along—and get them further along—over the course of time. So that's how we teach. But it's just not the way we've designed organizations, historically."

About the Author

Carmen Nobel is the senior editor of Harvard Business School Working Knowledge.

Post A Comment

    • Scott Good
    • Research Director, Senior Consultant, Crescendo Consulting Group
    Fascinating study. I love the methodology (especially the $0.60 per minute to the undergrads). The results are consistent with Crescendo's approach to conducting strategic planning initiatives and Community Health Needs Assessments (CHNAs). For ex., in CHNAs, hospital leadership teams are required to prioritize a long list of community needs. We have found that using a modified Delphi Technique often strikes a good balance between casting a broad net while avoiding premature consensus. Follow up with me if you'd like details.
    • Bruno R Borghi
    • Director, Akeirou
    Teams are not efficient immediately. It is well known that there are stages before a team performs (form, storm, norm, perform...). This study would be more interesting and meaningful if the evolution of performance across several successive rounds was measured. Even though the team does not perform well the first round, when the team uses feedback loops such as retrospective at the end of each round (? la PDCA) the team's process improves quite quickly.
    Also, smart young people do not necessarily know smart problem-solving methods, such as De Bono's 6 thinking hats. Another experiment comparing teams trained to a problem solving method with teams not trained would be interesting.
    • Marija Liudvika Drazdauskiene
    • Professor, Wszechnica Polska, a higher school in Warsaw
    I have appreciated the article and the idea. The conclusion reminded me somewhat of a wise and very brave professor who fixed the rule in the central university in the former Soviet Union where collective work was the law. While assigning a task, she once said: "You will be personally responsible. When all are responsible, nobody does anything with responsibility." It sounded like a general's order then and it worked. Thank you.
    • Richard Gardiner
    • Semi retired urban planner, Self Employed
    I have been involved in creative problem solving for over 45 years and I can tell you one thing about your approach: you are suffocating the problem solvers. You are so invested in systems thinking you are short circuiting creative problem solving. That's fine for vertical or linear thinking such as in process or product refinement but it won't work in original innovative problem solving. In other words the really break through discovery problem solving.
    • Kapil Kumar Sopory
    • Company Secretary, SMEC(India) Private Limited
    This is an interesting research which, however, is praiseworthy at laboratory (classroom) level but I doubt it is unlikely to produce effective results in the field. Projects these days are a mixture of a variety of activities in all of which a single individual does not generally possess the required expertise. Hence, we have to create teams for accomplishing the task by collaborating with each other focusing on the goal. Yes, the team leader in particular is responsible and accountable for the finalization of the project by the targeted deadline and in the manner expected. Yes, in some situations clustering could undermine performance if there is unproductive balance between exploration and exploitation. To counter such a situation, a watchful management has to effect timely correctives if it observes that collaboration is leading to chaos which is not at all acceptable. It may be necessary then to shuffle/remove/replace the unwanted t
    eam members.
    From my own experience clustering has been seen to improve problem solving by all well-knit teams creation of which has got to receive due thought ab initio.
    • Jiten
    • Lead Quality Consultant & Program Manager
    I have been waiting for this kind of research in this specific concept. It is very happy to know that HBR knowledge is having up to date knowledge on what is going on in organizations with respect to the Teamwork, Team management etc. Yes, the word team work has actually downgraded recent days, because of so many factors. The top factors are 1)There are my talkers and no one really wants to work on the problem 2)Roles and responsibilities are very chaotic
    To achieve a perfect team work, there are many pre-requisites or stages to be completed such as Forming, Storming, Norming and Performing. So, the cost of achieving a "team work" is very high,compare to the cost spent on an intelligent individual contributor. its similar to the analogy - Too many cooks spoil the disk. So, if you want to achieve a less conflict greater results within short span of time with less time,effort,money..then give the task to "individual contributor" roles...instead of assigning to group or so called team members. After all, at the end of the year, companies measure the performance of each individual and not at a team level. So, there is no real team work in this world for service industry. But, it is not true for non-service or manufacturing world. Team work is meaningful for a manufacturing every member in a team work on his/her own piece of material to build a final product. but for service industry the product is invisible.
    • Don Gilman
    • Executive Director, Institute for the Advancement of Critical Thinking
    Very interesting article that furthers the research into collective problem solving. As Mr. Borghi writes in his comment, I also would be interested in seeing if the results were similar with groups trained in collective solution generation. De Bono's thinking hats is a good example that, if applied, could potentially lead to an increase in the breadth and number of potential answers. Perhaps the solution includes training our teams on group-level critical thinking skills.
    • Doug Elliott
    • Founder, D. Elliott & Associates
    This article is interesting to me in what it does not address which is the problem itself. In my experience getting to the 'right' problem is the real challenge to problem solving. The emphasis on enabling any form of solving processes is an absolute waste of resources when these processes are directed to illusory problems. All too often we assume that when a problem is posed, it's a real problem. Our natural, self-interested instincts are to solve it anew because this is how most career advancement is measured, collaboratively or otherwise.

    Problem solving involves getting answers to what we don't know. In my experience with R&D, invention and innovation activities, most of what we don't know is nothing more than ignorance of a solution that is already known. Just not by us. Thus most solutions can be found by going to data archives or a librarian, but this isn't much of a career advancement. I would be much more interested in the research on problem solving processes that are concerned with figuring out whether we have a real problem or not.
    • Dmitry Sokolov
    • Founder, noaSphere, Internet of Sense
    Thank you for the paper and discussion. Very interesting!
    I believe ability to "switch" between clustered/disconnected network structures is a part of the culture of decision making and can be learnt as mentioned before. I also agree with Doug Elliott regarding the "reinventing wheels". Frequently, the solution(s) of a problem is known. From this prospective, the problem becomes "not real" as it is immediately transferred to the category of tasks.
    I am wondering whether the environment or platform supporting this kind of adaptive decision making culture and real-time access to known solutions is already developed. Your suggestions would be appreciated.
    • Ben
    • Founder, TwoSix
    This is a fantastic experiment. It shows a really good example of end to end collaboration.

    It's important to know the right people and to get the right resources in place to start with. Creating the right environment to work together is how to make things happen.
    • Cynthia Silva Parker
    • Senior Associate, Interaction Institute for Social Change
    Interesting research, but I see an important gap. Collaboration involves a lot more than just being connected. It requires deliberate methods to help people think together, build understanding, and take action--both a framework for organizing the thinking and skillful facilitation. Without that, collaboration often does not work. With it, groups are smarter, more strategic, and more aligned around implementing the decisions they make. My colleagues at Interaction Associates have written about this recently at and my experience with colleagues at Interaction Institute for Social Change also bears this out.