"Online and In Synch," Feature Article, June 2007

online and in synch

Systems and negotiation theory has a lot to tell us about collaborative design.

by Jean Thilmany, Associate Editor

a toolbox that attaches to the top of a folding ladder keeps all needed tools within easy reach. No more climbing back down to retrieve a stray screwdriver. A three-in-one tool combines safety knife, marker, and pen.

Maybe they're not revolutionary, but these products sure are handy. And invention company Wagic of Los Gatos, Calif., thought of them first. The name stands for "What a Good Idea." To see the toolbox draped over the top of a ladder and crammed with every conceivable screwdriver ever needed is to immediately envy the inventors of such an eminently useful—yet such an obvious—idea.

That these tools hadn't existed before Wagic engineers dreamed them up speaks to issues with collaborative design, said Ron Johnson, the company's chief executive officer. When Wagic reinvented itself from a standard product development company to one that created its own innovative brands, Johnson knew he'd have to take a look at the engineering design process itself.

"Many companies fail in product development when they get locked into a path and can't deviate," Johnson said. "We can be pretty far into the design process and realize we need to make significant changes."

Wagic comes up with its unique products thanks to a well-considered collaboration process. Johnson has placed great emphasis on the way his engineers work together, even as they're mostly working at their own, separate computers.

Engineers are encouraged to pick up each other's three-dimensional models at any time and make their own changes. The company's CAD system—OneSpace Modeling, from CoCreate of Fort Collins, Colo.—supports that kind of passing back and forth.

But Johnson may be unique in his focus on the collaborative process. According to a researcher at MIT, many engineering executives haven't given much thought to the way engineers work best together. Collaborative engineering design—mostly done by engineers working online at separate computer terminals—may be the norm today, but the seemingly spontaneous process still holds a lot of room for improvement, said Mark Klein, principal research scientist at MIT's Sloan School of Management in Cambridge, Mass.

By applying complex systems and negotiation research to collaborative design, he and his colleagues have come up with several sound ideas to help engineers work together toward the best product possible.

Human Behavior

Go back a few decades, and you'd simply walk across the office or drive a few miles to speak with your engineering-design collaborators. That's all changed now, of course. Today, in addition to working with designers around the globe, you likely work with more team members than you did in the past. You might not even know what your coworkers look like (although you can hazard a guess) as you meet via the Internet or a product lifecycle management system to brainstorm and make changes to CAD models.

In this age of disparate design, Klein and several of his colleagues have turned their attention to collaboration. We're all human, of course. And clearly working together has pros and cons. Researchers wanted to study the types of human behaviors that inhibit true collaboration and suggest ways to improve collaborative design.

The engineering community hasn't been clamoring for a better way to work together because the way engineers collaborate now is still so new it seems, on the face of it, to work just fine, Klein said. Typically, no one at a company gives much thought to the way engineers exchange ideas online and in person. It just happens.

That won't be true for very long, Klein said.

"While many important advances have been made in such areas as engines, materials, and avionics, the basic design concept hasn't changed much," Klein said. "Radically innovative design challenges, such as high-performance commercial transport, will probably require substantial changes in design processes."

The marketplace is making increasing demands for innovative, sustainable design. Many engineered products will have to be different from anything conceptualized today. To get there, engineers are going to have to work together in the smoothest possible manner.

In fact, Klein became interested in collaboration issues when he saw them played out firsthand during a stint at Boeing, where online and collaborative design is the norm. Theoretically, putting all these engineering heads together should produce the best design possible. But that's not exactly true in practice. Klein observed that working together in this way can be expensive and time-consuming because engineers have a difficult time settling on what everyone feels is the single best design.

At companies that assemble airplanes, automobiles, or huge machines, the design process can go around and around in a series of changes.

The design process can go round and round amid a continued series of updates and changes, particularly at a place where thousands of engineers are at work separately on parts that will come together to form a huge, complex airplane.

Before the whole comes together, each individual team focuses up close and personal on its own part of the assembly. And therein lies one problem, Klein said.

"Everyone is doing their own little piece. Basically, they're all trying to make the best carburetor, the best engine block," Klein said. "It's a network of people making sure their piece of the pie is the best without looking at how it will fit within the overall whole. But all those pieces will need to fit together."

Adding to the problem is size of the overall product and the sheer number of players involved.

"When you have a really complex product, it's really hard to keep track of everything, even if there were only three people working on it," Klein said. "No one has a big enough brain to keep track of everything."

In a distributed design process with myriad players all working on their own parts, no one could possibly be charged with overseeing the entire assembly. Individual managers oversee their own little bit of the design. So the overall assembly could be moving toward a great or terrible outcome. Who would know?

"It's not obvious when you're in the midst of it, because you can't get the big picture, if things are going south or not," Klein said.

He likens a collaborative design process to the up-and-down swings of the stock market. No one invested in the market has any interest in seeing it tank. So why does it sometimes trend downward? Because investors buy and sell based on their personal interests.

Behavior resulting from those individual personal reasons can come together for an outcome no one wants to see. Say that a few investors decide to sell based on the day's news. Others get word of a sell-off and call their brokers to get out, too. The ball keeps rolling toward the outcome no one wants because everyone acts individually. That can happen with design.

The nature of engineering design itself also keeps online collaborative design from moving inevitably toward the best, most innovative outcome. Engineers well know that, when it comes to design, there's no hard and fast answer to the question of which idea is better. The correct answer is: "It depends." A decision will lead to a bunch of different results that, in turn, can lead to another bunch of different results, an ever-branching tree.

With many engineers making multiple decisions and each of those decisions affecting future decisions, you can see why everyone involved in aircraft or vehicle design might not always be on the same path toward perfection.

Klein has identified a number of potential problems inherent in collaborative design via technology. But how can companies address them? That starts by taking a look at how individual engineers behave within their design teams.

Prisoner's Dilemma

One idea for better design that Klein and his colleagues have put forth might seem counterintuitive at first.

Using what they know about systems and negotiation theory, Klein and his colleagues turned their attention to the way that engineers negotiate as they work together on a single design. Researchers found that in order to get the best digital model possible, everyone involved must give a little. This is called the prisoner's dilemma. By giving in at first, engineers can expect to get to a better design several steps later, Klein said.

The thing to remember here is that negotiation works differently in the collaborative design world and at the church rummage sale. Engineers can't negotiate at work the way they'd haggle over the price of a Polaroid camera at a garage sale, Klein said. That type of back-and-forth negotiating doesn't work in a field full of variables and what-ifs.

"If we're talking about negotiating over the price of something, we both give a little to meet in the middle," Klein said. "With design, we may find the design halfway between yours and mine is not the best at all. There may be a third direction.

"There may be hundreds of other directions, but you're not going to find them if you're focused on meeting in the middle," he said.

Instead, if an engineer is willing to consider proposals from team members at first—even if they don't seem like the best ideas—but then turns into a tougher and tougher negotiator over time, all team members end up with a better design than if no one had given in at first, Klein said.

"If you start off saying, 'This is bad, but let's follow it for a while and see what happens,' you get the better design," he said.

Of course, all engineers must be on board with this negotiating tactic. If even one engineer remains steadfast and incapable of giving in, that hardest nut to crack will always win. The person who concedes in that type of arrangement—the prisoner with the dilemma—will be frustrated every time, Klein said.

Cycle of Influence

Klein and his colleagues also looked at how to stop the endless loop of design changes and revisions. When each change affects another, changes can go around and around in a circle that has no end.

So why are some design cycles more cyclical than others?

The answer lies in the influence the players have on one another, Klein said. If each of them is roughly even in job title, then the influence they have on each other is even as well, and everyone settles into a design. If one engineer's influence is greater than the others', design can circle around and around as the players without influence go with their bosses' requests and then turn around to suggest changes themselves in order to impress their bosses.

After design is done, extra steps are sometimes needed to ensure that the product will be fully recyclable or to create accompanying training material. The influence of these end players is often least, Klein said, so they might subconsciously choose to institute another round of changes to exert their influence.

Finally, in their paper Klein and his colleagues identified a third problem with collaborative design. Researchers who've looked at team design have long been puzzled by why design problems all crop up together in clusters.

"If you look at design problems as the design is being worked on, sometimes a bunch of bugs would be reported. Then everything would be good. Then a bunch of bugs would be reported again," Klein said. "And it happens like that every time."

So what is going on?

Klein contends it's likely a phenomenon known in complex systems research circles as the liars' club. The club is founded when engineers work in parallel on separate parts of a design. Let's say one of them finds a problem with a part of the design. Chances are, the engineer will hold off on reporting it to the manager.

Collaboration was the norm when Boeing and Marshall Space Flight Center engineers worked on the lab module for the International Space Station.

"In most engineering organizations, people get mad if you have a problem because they think it could create scheduling delays," Klein said. "So they think, 'I'll just wait and report my problem because someone else could go first and then the fault will be theirs.' Then, when someone blinks, everyone reveals their problem."

These flurries of design problems slow down design as development stops to focus on them. Problems that are worked on as they crop up take less time to deal with.

So how to lessen the cycle of influence? How to ensure there are no prisoners within a team? How to break up the liars' club?

According to Klein, design managers must be aware of the potential for these kinds of problems and work from the beginning of the design process to mitigate them.

"They aren't things you'd learn when you get an M.E. degree," Klein said.

He suggests that managers spend as much time designing the design process as they would designing the part.

"In particular, they need to pay close attention to the nature of incentives in reporting or in not reporting a problem, and to the influence people have with pieces of the design," he said. "If you get punished for having problems and rewarded for being a tough negotiator—even though that makes sense locally—it doesn't make sense of the best kind of design.

"If you go to any airplane company, they won't have an explicit map of relationships. You just sort of learn it. I think that kind of social knowledge needs to be made explicit, and studied and reproduced when you have another similar project."

And then all of the systems will be go for a truly innovative design.