In December 1992 at JFK Airport, Lori Rosenkopf took her first step into a flight simulator used by TWA to train commercial pilots. It was an opportunity to experience firsthand the multi-million dollar machines that she would eventually feature in an article entitled, “The Coevolution of Community Networks and Technology: Lessons from the Flight Simulation Industry.”
But on that particular day six years ago, Rosenkopf’s attention was focused on the cockpit. “The instructors handed me the controls and said to go ahead and give it a try. They didn’t tell me that you are supposed to turn the wheel away from the direction you want to go in. I taxied along thinking I was turning left, but instead was turning right. I basically drove the plane off the runway.”
While Rosenkopf travels on airplanes these days strictly as a passenger, she is no stranger to the intricacies of technology. She earned bachelor’s and master’s degrees in engineering from Cornell and Stanford, and spent six years in the 1980s working at Eastman Kodak Co. as an industrial engineer and at AT&T Bell Labs as a systems engineer.
“When I was in school I thought I understood how companies selected one particular technology over another,” she says. “Engineers would do problem sets where they compared alternative approaches according to very strict performance measures, such as stress level or throughput. The technology that had the best results won. But once I started working I saw that engineers aren’t trained to consider the social, organizational and/or political ramifications of a particular approach and how these factors can affect the technology’s ultimate success.
“Engineers like to think that it’s obvious which technology is the best, but it’s important for managers/engineers to understand how many other harder-to-measure factors enter into that decision.”
Rosenkopf, the Douglas Vickers Term Assistant Professor of Management, went on to get her PhD from Columbia in 1994 in the management of organizations. Since then, her research has focused primarily on the kinds of communications networks that exist within and between firms in a particular industry and the ways in which these networks influence the transmission of knowledge, the creation of innovation and the development of technology. She has studied four high-tech industries — semiconductors, optical disc, cellular and flight simulation — looking specifically at the organizational factors that affect how people choose, measure and compare one technology over others.
To illustrate the role that networks play in these processes, Rosenkopf sketches out a diagram showing two types of workgroups. In the first example, person A’s network consists of links to five individuals within the company. Each of these five also has networking links, but they are all to each other. The result is a clique, in which each person shares information with members of his or her team, but typically doesn’t look outside for additional data.
Person B is also connected up with five people, but these five individuals have links to others outside the group, perhaps in another department, an industry trade group or a community organization. So instead of receiving redundant information from five people, as A does, B has 20 different streams and flows of information. “Managers in situations like these tend to be the ones who come up with ideas for more innovative products and services,” says Rosenkopf.
Most work groups, of course, tend to fall somewhere between these two extremes. And setting up these networks takes time, intention and the ability to create new ways to link up with information sources both inside and outside the firm.
“Everyone knows this is what managers are supposed to do,” says Rosenkopf, “and in some venues, such as sales, this approach is institutionalized as part of the job. But if you are an engineer, or an R&D manager or even a top executive, there are so many issues competing for your attention that when push comes to shove you become very involved in your own products and processes.”
For engineers and technical managers, network links can be established through participation in cooperative groups like standards bodies, task forces, industry-wide committees and government organizations. On a broader basis, companies can create whole new networks of learning through alliances, through encouraging mobility in the engineering and executive ranks, through common directors and through social networks that emerge informally, such as the afterhours fraternizing that is an integral part of the work culture in places like Silicon Valley.
Firms that have these kinds of networks — bridging relationships as opposed to cliques — not only tend to conceive of new product ideas more often, says Rosenkopf, “but just as important, they are more successful in getting others to buy into their ideas.” Consider flight simulation.
Navigating New Technology
In the flight simulation industry, Rosenkopf says, a very small number of firms produce full flight simulators (FFSs) and flight training devices (FTDs). Full flight simulators typically cost $15 to $20 million each, replicate the flight experience by integrating cockpit instrumentation with full motion and visual capability, and during the 1980s were supported by commercial airlines, regulatory bodies and aircraft manufacturers. The flight training devices cost $1 to $3 million each, have no motion or visual capabilities and were supported by academic and military researchers, regional and general airlines and flight schools.
“Because these are very expensive, customized machines that are closely overseen by federal regulators, it’s an arena where you don’t have a bunch of engineers going off to some skunkworks R&D lab figuring out what to do, building it and then hoping it sells,” Rosenkopf says. “Rather you have a whole community of people involved in the process — the simulator manufacturers, the airplane manufacturers (McDonnell Douglas, Boeing, Airbus), government regulators and the users/customers (commercial airlines). They all care very deeply about what the simulators should do, and how they should be used to demonstrate that the pilot is truly proficient. Consequently there are many networks and linkages even among players who are competitors. Everyone ultimately has to agree on the product because otherwise you are investing millions of dollars building something for which there might not be a market.”
Rosenkopf’s goal in studying the flight simulation industry was to show how interorganizational linkages led to the dominance of full flight simulation approaches (FFS) over modular approaches for more than 15 years.
She identified 27 different groups involved in the process, including trade associations, professional societies, suppliers, standards bodies, regulators, governmental bodies and manufacturers. They ranged from the International Air Transport Association Flight Simulation Technical Committee and the American Institute of Aeronautics and Astronautics Working Group on Simulation Facilities to the Wind Shear Working Group and the Royal Aeronautical Society International Simulation Standards Working Group.
“My research showed that it was the same people over and over again — a core of about 15-20 individuals — who were on many of the committees and therefore were defining what the industry would do, what the product would look like, what the rules would be,” Rosenkopf says.
“By drawing lines between all the different groups I was able to define this dominant coalition which, it turns out, was fighting for the far more expensive FFS approach over the less expensive FTD. Members of this coalition were very organized, always pushing the same agenda, meeting at the same forums in different cities, and so forth. Their collaboration really propelled the choice of FFSs. It meant that the $15 to $20 million machines were the ones built for pilot training back in the 1980s, even though the benefits of using FTDs may have been greater, both in terms of improved pilot training as well as lower training costs which would ultimately translate into lower-cost airplane tickets for the public.
“You could conceive of better approaches for training pilots that used both FTDs and FFSs but that didn’t get accepted, not because they weren’t good but because they didn’t have a coalition of broad network support.”
Interestingly, by the early ‘90s, Rosenkopf says, some of the groups that hadn’t been part of the main coalition or clique, and that typically didn’t communicate outside their own small network — such as flight schools in the middle of Nebraska that trained pilots for the regional airlines rather than the big commercial players — began to talk together and slowly build new groups. Eventually they forced the other coalition to recognize that there might be circumstances where the $1 million devices are preferable to the $20 million ones. Today, hybrid systems that use both FTDs and FFSs are being explored by the industry.
“Networks affect technology and technology affects networks,” says Rosenkopf. “Even after there is closure on a dominant design, firms must still consider how network dynamics may trigger or hinder the next technological change.”
Adhesives and Laundry Detergents
If you look at any technology, Rosenkopf says, you see that ultimately, over long periods of incremental improvements, there is some dramatic advancement, something that fundamentally transforms the industry. Vacuum tubes, for example, gave way to semiconductors, mechanical watches to quartz watches, x-rays to catscans to magnetic resonance imaging.
The big question, of course, is which ideas to pursue. “In hindsight there is always a right answer. For me, that’s when I fall back on these ideas of networks. Your job as a manager is to establish the bridge networks that give you more multifaceted views of your technology options which then allows you to make wiser choices.”
Rosenkopf recalls a recent meeting with the technology director of the adhesives technologies center at 3M. “His job, he told our group, wasn’t to figure out what particular variant to pursue with respect to adhesives, but rather to encourage people in the organization who normally don’t converse with each other to begin interacting on a regular basis. Much of 3M’s philosophy is focused around setting up task forces, councils, different sorts of environments where people from domains that usually don’t intersect are now crossing paths and sharing information. Managers are looking to create these unique combinations of knowledge that don’t naturally occur.”
Rosenkopf cites Procter & Gamble as another example of a company that is continually looking to expand its frontiers. “P&G’s Tide laundry detergent comes in powder and liquid form and both are very successful,” she says. “But P&G didn’t just sit back and say they are going to make increasingly whiter and brighter liquid and powder. One of their experiments is with a product called Tide sheet. It’s all-in-one laundry care that goes into a wash, releases detergent and fabric softener, and then goes into the dryer and prevents static cling. The only thing it doesn’t do is remove your laundry and fold it.
“It sounds like a great idea. But the company has to think about a lot of side issues, such as the effect of such a product on related businesses like fabric softener as well as the potential cannibalization of Tide powder and liquid.
“What has happened so far, however, is that the company can’t get the product to work. It keeps burning up in the dryer. But P&G isn’t giving up. They are still experimenting with this and with all sorts of other approaches such as capsules, pellets and sponges. They are also looking into technology that would allow consumers to essentially dry clean garments at home, and they are researching fabrics that would repel dirt more effectively than current materials.
“P&G knows that 19 out of 20 experiments are going to be total failures, but they also know that the five percent that aren’t will be the seeds of their future business. And better that they are cannibalizing their own products than have Unilever figure out new technology ahead of them.”
The Microsoft Factor
Two years ago Rosenkopf began a study of inter-firm learning among start-ups in the semiconductor industry, analyzing how both formal mechanisms (such as alliances) and informal mechanisms (such as the inter-firm mobility of engineers, geographic proximity and technological similarity) help process knowledge that other firms are also generating. Patent citations are a major component of her research.
“Patent citations give you a view into the kinds of knowledge that a firm has built upon in order to create new knowledge, similar to the way that academic citations show all the different influences on a scholar’s intellectual development,” Rosenkopf says. “If I’m looking at patent no. 5,200,536, and that cites back to patent no. 4,307,232, I can look up that old patent, see which firm produced it, where they were, what inventors were involved and so forth. In addition, patents can give you a little window on what a particular company might be doing five or ten years down the road rather than now, which can be helpful in determining its future product strategies.”
Her research shows, among other things, that in the semiconductor industry, firms that enter into alliances tend to do it with firms in the same geographic region that are technologically similar. “It gets back to networking. Everyone is looking at the same bits of knowledge and recirculating them. You would do better to look at firms in different areas with dissimilar technologies in order to create bridges to more unique information,” she says.
Start-ups, Rosenkopf points out, typically tend to be as innovative or even more innovative than incumbent firms. In the semiconductor industry, “many of the landmark innovations — including the memory and microprocessor chips — originated from start-ups. How then do firms with limited R&D spending maintain their innovativeness?”
The answer, she says, gets back to the whole issue of the learning behavior of small firms and their willingness to look for external sources of knowledge.
“Firms that take advantage of technological developments at other organizations and don’t just focus on core competence or their own particular competitive advantage seem to be the ones that over the longer haul will have more influence on technological development,” says Rosenkopf. “Other firms can end up in what is sometimes called a competency trap: They are very good at doing what they do but they start to drift away from the path that the majority of players are focusing on.
“The idea of developing core competencies is very hot,” notes Rosenkopf, “but the issue is you can only continue to develop this competence by making sure you are integrating all the developments of other firms. If a company’s attitude is that we are the best and we can do no wrong and everyone else will follow us, it falls off the mountain sooner or later … Technology is so big and broad and moving so quickly that a company has to be monitoring outside developments, figuring how to co-opt those developments into their own business.”
The best example of this, Rosenkopf says, is Microsoft.
“There are very few things that Microsoft has invented,” she notes. “Virtually all its successes have had their roots outside the company, but by now Microsoft has an array of established platforms and so much money that it can basically pursue whatever it wants to. Its tentacles are everywhere. No matter which technology is up and coming, Microsoft has started to copy it, created an alliance or bought the firm. Most companies don’t have that level of resources, and consequently have to pick and choose their networks. In many cases they don’t do enough of that exploration. Microsoft has the ultimate B network.”
Firms like Netscape, Novell, IBM and Sun Microsystems are all trying to create the same sorts of networks, the same bridges, Rosenkopf says. “They are either much more resource limited or they are playing catch-up in establishing the networks they need.”
All companies say they want to be innovative, Rosenkopf notes. “They come up with ideas to accomplish that goal, ranging from employee involvement schemes, suggestion programs, entrepreneurial seed ventures and so forth. They generate lots of unique, interesting ideas that could be pursued. But ultimately they have to choose among different options. What frequently happens is they choose things that fit very closely with their established platforms. The choices they make squeeze out all the interesting variations they considered. Extensions of all the platforms and technologies are important, but firms also have to emphasize more variations away from current operating procedure.”
Teamwork: Diminishing Returns
“When you have an executive team — people in positions to make strategy — that has been together too long, studies show that they get stuck in the same way of solving problems,” says Rosenkopf. “That means that companies should think about bringing in a new executive, or a new team, in order to change things around. You shouldn’t be waiting for people to retire or leave the organization.
“Of course some companies don’t have this problem because their industry is so fast moving that it’s hard to keep an executive team together. But in more stable and mature industries, executive teams tend to go on and on forever. These are the firms that get set in their ways.
“Research on group problem-solving ability shows that when you create a group, it has to learn first how to work together. Initially, its performance is poor, then goes up and then goes down again because of this staleness. The peak time together is about two years. So you can imagine how an executive team that has been together 10 or 15 years may be well past the point of peak performance. That’s not to say you throw everyone, or even anyone, out. You might want to add new resources or create several new positions and change the dynamic that way. Your basic goal is to jumpstart the team.”