CHAPTER 12

THE BUG TESTER

After Ross Smith’s promotion to team leader of the Windows Security Team at Microsoft, he scheduled one-on-one meetings with the eighty-five people under him. During these get-to-know-each-other sessions it dawned on Smith, a twenty-year software industry veteran who had worked a wide swath of jobs, that his new charges, all of them younger than he, took a distinctly different approach to work and life. They were “millennials,” those seventy-eight million members of Generation Y born between 1980 and 2000 while he was a child of the 1960s. There was, as much as it pained him to admit, a generational divide.

Smith, in his late forties, is not your typical corporate manager. When I meet him on the Microsoft campus outside Seattle on an early spring day he’s dressed in what for him constitutes office garb: shorts, ratty T-shirt, and flip-flops, despite the teeth-chattering weather—forty-two degrees and raining. That evening he plans to drive to his house in the hills and night ski, and he’ll do that in shorts and a T-shirt, too. Several years earlier he had suffered an acute case of alopecia and ever since has had no hair—and not only on his head: no eyebrows, eyelashes, nor leg or arm hair either, and never feels cold, and with his florescent white skin he bears more than a passing resemblance to that famously friendly ghost. His first job out of Rider College in New Jersey was as a data processing coordinator for the county sheriff’s office, which he did for four years until jumping to tech support for handheld devices for a couple of years. In 1991 he started at Microsoft, where he provided product support, and has spent the past two-plus decades climbing the corporate ladder.

His eighty-five highly educated programmers—a third hold a master’s degree or higher—are tasked with mind-numbing work, scrutinizing thousands of lines of code for security glitches, and the work ebbs and flows, Smith says, depending on where a product is in its development cycle. This yields substantial excess capacity. Worse, in these one-on-one meetings with him, many on his team reported feeling demoralized, underutilized, often underappreciated, with Carnegie Mellon and MIT computer graduates running basic, yawn-inducing tests such as inserting punch cards into a reader. While some of their peers were reaping fortunes exploiting Web 2.0 and building Facebook, Twitter, and a host of other glitzy start-ups, they were like Ivy League law grads assigned to traffic court.

When Smith started his career at Microsoft it was considered a “velvet sweatshop,” a characterization reported in a widely read 1989 Seattle Times company profile detailing the lives of Microsofties and the eighty-hours-a-week “techaholism” that pervades its culture. These workforce issues, the story pointed out, were typical for “achievement-oriented, high-pressure Baby Boom professionals who during the ’80s embraced work as their badge of identity but are now having to re- evaluate its impact on their personal lives, friendships and family, and life goals,” calling it “a ticklish proposition for any fast-moving company: Keeping workers motivated, creative and energized while still offering them the chance for a balanced, self-nurturing existence.”

The company prospered throughout the 1990s as Windows became the world’s operating system, and in December 1999 its stock hit an all-time high of $119.94 a share. It has been dropping ever since. Apple, the company it almost drove out of business, has supplanted it as the most valuable tech company. As Kurt Eichenwald pointed out in Vanity Fair, one relatively new Apple product, the iPhone, is worth more than all of Microsoft. The company, which had been built on innovation, had stopped innovating. Products take eons to get to market, and when they do they are either too late or often not very good. Case in point: before I booked my air ticket to the West Coast, I felt compelled to phone Smith with a confession. Microsoft had removed all landline phones and required employees to use the company’s Lync software, which runs phone calls through computers. Except every time I called Smith at Microsoft, the call would drop or the quality was miserable and I’d have to try again. And again. Microsoft personnel, I learned later, often just used their cell phones at work instead.

But I was more concerned with something else. “I use a Mac laptop and carry an iPad and iPhone,” I told Smith. “Could this be a problem?”

After all, Bill Gates had banned his own kids from owning iPods, and Microsoft is notorious for its cutthroat behavior with competitors. Then there were all those critical stories I’d written about the company over the years. I was ruing the times I’d referred to Microsoft as “the dominatrix of the desktop.”

“Nah, not at all,” Smith replied. “It might have been a problem in the past, but the place has changed over the past ten years. This isn’t the same Microsoft.” He paused. “Um, you’re not going to interview Steve Ballmer, are you?”—referring to Microsoft’s famously blustery chief executive officer. Legend has it that Ballmer heaved a chair across a room amid an obscenity-laced tantrum after an executive broke the news he was joining Google.

I told him no.

“Yeah, yeah, then you’ll be fine,” he said.

Eichenwald traces Microsoft’s demise to an obscure management practice known as “stack ranking.” Premised on Jack Welch’s vitality curve at GE, in which the lowest-performing employees (the bottom 10 percent) were driven from the company, Microsoft had each unit apportion a percentage of employees in one of five performance batches: top, good, average, below average, and poor. Instead of increasing the quality of Microsoft personnel, however, it had the exact opposite effect. “If you were on a team of ten people, you walked in the first day knowing that no matter how good everyone was, two people were going to get a great review, seven were going to get mediocre reviews, and one was going to get a terrible review,” a former software developer told Eichenwald. “It leads to employees focusing on competing with each other rather than competing with other companies.” Top programmers avoid working on the same teams, since it could harm their rankings. This in turn could affect their bonuses and promotions, because those at the bottom received no additional money or suffered the indignity of being fired. As a result, Eichenwald concluded, “Microsoft employees not only tried to do a good job but also worked hard to make sure their colleagues did not.”

Nevertheless, Microsoft, almost forty years old and an entrenched member of the software establishment, still hired a lot of young engineers, and in addition to the gang warfare that characterized employment it was apparent to Smith the company was riven by a subtle culture war. As a Gen Xer Smith could have easily dismissed these corporate newbies as spoiled, selfish, self-involved hipster wannabes afflicted with a collective case of attention deficit disorder and suffused with a poor work ethic and grandiose belief that stoplights ought to turn from red to green on their command. If he had, he wouldn’t have been alone. Eighty-five percent of hiring managers and human-resource executives polled by CareerBuilder.com described millennials as possessing a stronger sense of entitlement than older workers, expecting higher pay, flexible work schedules, promotions within a year, and more vacation or personal time. Workplace Solutions found that 68 percent of working Americans surveyed believed that Gen Y workers were less motivated to assume responsibilities and produce high-quality work than older workers. Talk to corporate recruiters and you’ll hear many grouse about members of “Gen Whine”: “They want to be CEO tomorrow,” is one common refrain. But Smith also knew that today’s underlings are tomorrow’s bosses and eventually this generation would be running the show.

To understand millennials it’s important to understand the environment from which they sprang, and while broad generalizations can be counterproductive there’s consensus on several Gen Y characteristics. For one, they seem to have inherited their parents’ quest for a work identity but without their parents’ blind adherence to the job. The roots of this shift can be traced to the dissolution of the social contract that bound workers to companies, which are part technological and part cultural. Lifetime employment was common from the end of World War II until the 1970s. Then American corporations, battered by foreign competition and an economy mired in sluggish economic growth coupled with inflation and rising interest rates, laid off legions of workers. Seeing what happened to their parents, these Microsoft millennials—and those elsewhere—aren’t likely to throw their lot in with one company (and why should they?). Yale professor Robert Hacker, author of The Great Risk Shift, calculated that income instability, which he defined as a 50 percent swing in income year to year, has tripled since the 1970s. Today’s college-educated employees have as much income instability as those without high school degrees did in the 1970s. Now the formula is: no job security equals little company loyalty.

Another driver shaping Gen Y’s relationship to life and work involves technology. These “digital natives” grew up with the Internet and embrace social networks. As Piotr Czerski, a Polish writer and commentator, declares in a popular manifesto titled “We, the Web Kids,” Gen Y does not “surf” the Internet and to them it “is not a ‘place’ or ‘virtual space.’ The Internet to us is not something external to reality but a part of it: an invisible yet constantly present layer intertwined with the physical environment. We do not use the Internet, we live on the Internet and along it.” Moreover the Web has been fused with their culture. “To us [it] is not a technology which we had to learn and which we managed to get a grip of. The Web is a process, happening continuously and continuously transforming before our eyes; with us and through us.” Instead of memorizing names, dates, addresses, street names, and phone numbers, these inveterate Web users tap keywords into Google, rely on GPS to navigate cities, access smart phones to dial or text friends and colleagues, and because they have to filter so much information and make instantaneous decisions as to its reliability, they are also perhaps more skeptical. Simultaneously the tools of creation are accessible to all so anyone can blog, post photos, create videos—in short, share her words, thoughts, art, and actions with the world without barriers.

The emergence of social networks also contributes to the digital native “webanschauung,” if you will. To Gen Yers, “society,” Czerski writes, “is a network, not a hierarchy.” With Twitter and Facebook, fans can chat with just about anyone, whether she’s a rock musician, movie star, politician, or author. This Net generation prizes access to social networks and communication with friends and family on the job. Cisco’s “Connected World Technology” report found that more than half of Gen Y employees surveyed ranked social media freedom over higher salaries when evaluating job offers, while 37 percent of those surveyed by Mom Corps said they would take a pay cut if it meant more flexibility (like the ability to set their own schedules and work from home). Meeting the demands of these younger workers could ultimately lead to the extinction of the traditional show-up-to-the-office workday.

Microsoft’s Gen Y engineers are no different. Despite a steady and generous paycheck, many are unwilling to put down permanent roots at the empire Bill Gates built; a nationwide survey found that 60 percent of millennials expect to leave their current employers (contrast with baby boomers: 84 percent said they’d stick with their current employers until they retire). What’s more, these Gen-Y Microsofties practically live online, conduct friendships and romance there, play games, work on side projects that interest them into the wee hours of the morning, and crave attention and notoriety for their accomplishments.

Multitaskers at heart and armed with smart phones, tablets, and laptops, they don’t accept the traditional barriers separating work from the rest of their existence. Ever connected to their jobs through technology, no one—not them, not their bosses—is ever truly off the clock. When they go home or out to dinner or a club, they check e-mail and respond to text messages from the office and continue to work, as self-described “personal brand guru” Dan Schawbel notes, “because who they are personally and professionally have become one and the same.”

Smith was faced with a dilemma: should his younger employees be forced to sublimate their millennial ways to fit in with Microsoft’s corporate culture or should Microsoft bend to their wills? America’s shifting demographics provided an answer. By 2016, Gen Yers will comprise about 40 percent of the American labor force and by 2025 that number should rise to 75 percent. At some companies like Ernst & Young, this group makes up 60 percent of the employee pool. A rising tide like this could put businesses underwater. It was clear to Smith that Microsoft—and by extension, he—had to address its workforce culture or face the consequences.

Smith wanted to see if game design could improve the quality, productivity, and job satisfaction of his crew, and concocted something he called the Code Review Game. It worked like this: four teams chose a section of code to attack. They received points based on the type of bugs they discovered. Different groups could come up with different strategies: one focused on code produced by error-prone developers, while another team waited until a flaw was identified and then hunted similar bugs, because the coder who contributed to the section would likely repeat the same mistake. The project went so well that the company has used such teams to vet at least six other major releases.

But this was only a small solution to the much larger problem of generational divide, and Microsoft wasn’t the only one grappling with it. It’s happening across the business landscape. Deloitte, the global accounting firm, concerned about high turnover among its youngest and newest employees, discovered that two-thirds of those who left the company took jobs doing something they could have done at Deloitte. High worker churn rates cost the firm $150,000 for each employee who left and whose position had to be refilled. In response the company created programs to help workers figure out their next career moves. Sun Microsystems embraced scheduling flexibility with a telecommuting program and now half of Sun’s employees work outside the office. With a majority of workers in their twenties preferring to work for companies that embrace volunteerism, Salesforce.com set aside 1 percent of its profits to pay employees to volunteer 1 percent of their work time to community service. Estée Lauder provides training courses on Gen Y for managers on how to deal with millennials who don’t share older colleagues’ work habits or approach to the job.

To counter this trend, some businesses aim for more fun in the workplace. The corporate culture of online shoe seller Zappos reflects the whims of its comically happy CEO Tony Hsieh. His company employs a “chief happiness officer,” and Hsieh once played an April Fool’s gag in which Zappos claimed it was suing the Walt Disney Company for claiming that it was “the happiest place on earth.” Google soothes its tightly knotted engineers with free massages and on its campus there’s a yellow brick road, giant model dinosaur, volleyball courts, and regular roller hockey games. Twitter brags that it has “worked hard to create an environment that spawns productivity and happiness.” Hyper-energy-drink maker Red Bull installed a slide in its London office. All of this emphasis on mixing play and work prompted the Economist to snark, “the cult of fun has spread like some disgusting haemorrhagic disease.”

Microsoft, too, has also been swept by change, not all of it by intelligent design. While it still reaps billions from operating system software, it confronts a more malleable customer landscape. Its core business has been under assault as software continues to move beyond individual computers and to the cloud. So you’re forgiven if “innovative” isn’t the first word that comes to mind when you think of Microsoft.

Following Smith’s manager-a-mano sessions with his team, he did what he always does when faced with a sticky work problem. He brainstormed with his brain trust, a small cadre of managers with whom Smith had climbed the ranks at Microsoft. One of them, Robert Musson, a man who reads data like others read People, recalled a research paper he’d come across that explored the relationship between trust in the workplace, job satisfaction, and pay. John Helliwell and Haifang Huang at the University of British Columbia calculated that for an employee a slight uptick in trust in management was equivalent to a 36 percent pay raise. The opposite was also true: a loss in trust led to a decline in job satisfaction equal to a 36 percent pay cut. The researchers found that those who worked on a variety of projects felt like they had earned a 21 percent pay boost while one requiring a high skill level was perceived as equivalent to a 19 percent salary hike.

All they had to do was increase trust among the workers and managers. It sounded so simple. But how do you foster it within a cutthroat organization like Microsoft? And where does trust come from?

Theory X and Theory Y

Within Microsoft’s data-driven workplace, Ross Smith, Robert Musson, and a wicked-smart, soft-spoken software engineer named Harry Emil continued their quest for trust. They identified an inherent contradiction in their workplace. The idea of trust, which encourages managers to trust their employees, to experiment, innovate, and take risks, collided with Microsoft’s need for what are called “predictable deliverables.” It was the age-old conflict between management theories: Theory X and Theory Y. With Theory X, the boss assumes the worst—that workers are inherently lazy and will at the earliest opportunity slough off their duties and responsibilities. Theory Y is based on polar opposite assumptions: workers are innately ambitious and self-motivated. While Theory X might be applied to a factory in China where workers with brain- and hand-numbing efficiency assemble iPhones, Theory Y is usually more useful in organizations that must innovate to succeed. In general, Microsoft was a hybrid but in practice it was an adherent to Theory X. Smith wanted to become more Theory Y. For that to happen management had to trust workers and workers would have to reciprocate by trusting management.

Even defining something as squishy as trust, however, wasn’t easy. It is, Smith says, a bit like freedom and air. You know when you don’t have it, but don’t give it much thought when you do. This made it hard to measure. He decided the first step would be to query his eighty-five-member team to identify specific behaviors that either enhanced or mitigated trust in the organization then create a simple online game to prioritize trust behaviors. The question was static: “Which trust factor is more important to you?” and players were faced with two choices. For example, “Tell the truth in a way people can verify” or “Do what you say you are going to do.” Or “validate expectations” or “underpromise and overdeliver.” Or “be understanding” or “establish a track record of results.” There were dozens upon dozens of trust scenarios.

A user, after completing however many questions he wanted, could rate specific characteristics like “Say what you’re going to do, then do what you say you’re going to do” as either first choice, second, third, most likely to make an impact or most likely to backfire, et cetera. Then he could view the results. Over the years some of the top choices ended up being “a role model—have integrity,” “be honest,” “respect the dignity of every person and every role,” and “listen before you speak. Understand, diagnose.”

As a manager, Smith operated with the will of the crowd in mind, and he kept the lines of communication open, instituting a weekly pizza meeting, which became a forum for each employee to raise questions, brainstorm ideas, and build relationships. It helped forge a spirit of camaraderie. At least in this small corner of Microsoft, there now existed a kindler, gentler, more trusting work environment.

Meanwhile, Musson faced a more tangible challenge: how to get more people to “dog food” a new operating system in development. Before Microsoft can create a version of a new consumer-ready product, especially one as complicated as, say, Windows, thousands of people must “stress test” it internally. (It’s called dog fooding because you are, in essence, eating your own dog food.) The more testers, the better. That’s because everybody’s computer is different and a Microsoft operating system has to support about a million different devices. More than a billion Windows users around the world means there are potentially a billion different configurations. Even if all Windows users bought the same Dell computer, they still customize them, and these combinations can cause serious trouble.

The software has to handle someone burning a CD while new e-mail messages download and he’s surfing the Web. Or maybe he needs to print a research paper while copying and pasting data from an Excel file into a PDF. How do older devices—a ten-year-old mouse with an equally aged driver—behave with the system? Each action involves multiple software programs, all of which separately or combined have the potential to induce the operating system to spit up a digital hairball.

In the early stages of operating system development, stress testers simply run the program on their PCs overnight while a program fires commands at it; if the program freezes, there’s a coding glitch, which must be identified and fixed. In later builds, users take the operating system for a test spin, using it like they would any other OS, or be asked to try specific combinations. The code is constantly updated, and there can be several builds over the course of a week, each of which must be tested.

But it’s hard to gain cooperation. “I have to do my job, so why should I do yours?” becomes a popular, if not verbally expressed, refrain. Most everyone at Microsoft is busy and gains little by helping a colleague in another group or team. And no one wants to deal with the hassle of a frozen PC that the tech staff has to drag away and fix. Sure, Musson, Smith, and Emil could order underlings to stress test the software, but they knew that was exactly the wrong approach. Instead they analyzed the behaviors they believed would motivate people to cooperate.

Harry Emil had a manager he once worked under who stuck up a piece of paper on his door listing each team member and requiring them to sign their name when they completed specific tasks. The manager couched it like he didn’t want to bother his team with a lot of unnecessary e-mail. “I’m not narcing on you guys,” he would say, but in a way he was, and it was effective. The list functioned as an analog leaderboard, since no one wanted to be viewed as someone not pulling his weight, which would lead to subtle public humiliation.

Emil also recalled a workplace game he had helped create in early 2004 that had achieved similar goals, that is gently encourage Microsofties to perform tasks outside their job descriptions. At the time 80 percent of Microsoft personnel used Google as their search engine, but the company had spent a lot of money and devoted vast resources to developing its own, MSN. Management was embarrassed that a competitor’s product was more popular in-house than the company’s. There was talk of tweaking the corporate firewall to disable Google but Emil and others believed the best product should win. Microsoft simply had to make a better search engine.

This led Emil to suggest a simple game, which he took to calling a Search Off. He created a page that offered both engines side by side and when a user typed in a search query he received both sets of results to compare. That way people who wanted Google still got their Google results, but they were also exposed to MSN. Then Emil incorporated a function that enabled users to offer feedback on both sets. He wanted to know how often Google was better, which naturally made people look at the MSN results and wonder, “Are they as good as Google?” Usually, although not always, they were. He received messages from users like, “Hey, they both solved my problem: I searched for a certain book and they both brought me that book. But Google was better because it had the book higher in its results.” Emil was pleased he had created a way to expose users to MSN without forcing anyone to use it. In the process, the game helped to clean up bugs (more than a hundred of them by the time he was done), things like when someone searched for a sports team but the top results pointed to a person with that last name.

He wondered if there were a method to further increase activity and decided to try a Search Off leaderboard listing the top twenty-five contributors. Every piece of feedback was worth a point. As soon as he turned it on the dynamics changed as people tried to leapfrog one another for the top spots. Within a week it took a minimum of a hundred clicks to get to number twenty-five on the leaderboard. After a while he noticed a falloff in new players, which he attributed to their inability to get on the leaderboard, because those who had been playing from the beginning had a head start. So he created a monthly leaderboard, which did the trick, then an interns-only leaderboard to trigger competition among them.

So when Emil and Musson’s team began receiving nightly e-mails from a senior VP of Windows, urging them to find a hundred volunteers to stress test the new operating system in development, it made sense to capture the same game dynamics that had made the Search Off successful. But they couldn’t simply use an identical comparison game. Instead the group looked at the behaviors they wanted to encourage: they needed people to install the operating system, vote, and run the stress test program. Three simple activities. They kicked around simple childhood games they could whip up: King of the Hill, Chutes and Ladders, tic-tac-toe, and memory games. They settled on Hangman, where they handed out letters to spell B-E-T-A-1 and announced to the entire Windows team that if a user installed the operating system, he’d get a “B.” Run stress and that warranted an “E.” Provide feedback, a “T.” If he ran stress for three builds in a row, that led to an “A” and a “1.” A leaderboard would report on the action.

The game proved popular. People would run into one another in hallways and brag that they got their “B” and ask others if they had gotten theirs. Vice presidents and group managers also e-mailed Emil to complain. “I don’t have a ‘T.’ I’m supposed to have one. What’s up? Where is it?” Usually it was a product of human error—the player installed the new operating system on a different machine—or there was a bottleneck in the system. Over the span of three weeks more than two thousand people participated, far more than the minimum of a hundred they needed to exit this milestone.

The Windows Beta 2 team asked Emil and his colleagues to create another game to stress test the second iteration of the software. This time he had months to plan and chose to introduce a social aspect beyond plain vanilla letters, something that would be tangible in the physical world, not just points in the form of alphabetical letters a player could claim to have won. Emil ordered ten thousand Livestrong-like rubber wristbands in four colors and designed the Beta 2 game to award prizes based on escalating participation. Players received a bronze bracelet for overcoming the first hurdle, which was to download the operating system and run the stress test program. The following week they could win a gray wristband if they matched 100 percent of their participation from the prior week. A yellow band awaited those who ramped up their participation to 120 percent of the previous week, and a blue one for those who achieved a 150 percent rate of participation.

Initially, Emil planned to run the game for six weeks, but it proved so popular it went on for four months. Few wore them down the hall as jewelry as he thought they would. Instead they collected the wristbands, which they strung in their offices. He ran out of wristbands and had to order more.

And it led to Microsoft’s most successful productivity game of all.

The Language Game

As director of tests for Windows International, James “Jim” Rodrigues was most concerned about three types of bugs in Microsoft operating software, which shipped in thirty-six languages and a hundred dialects (called “lips”). It could be that a bug caused the software to work improperly in the French version, but that’s functional and was usually caught through standard test procedures. Another afflicted the user interface. A button could be offscreen or there was truncated text. Rodrigues didn’t need to read and write French to find these obvious visual flaws. But the third kind of bug bothered him, and that had to do with the quality of the language translation. Rodrigues didn’t have the expertise on his team to fix all of these. Yet if there was a bug reported in French—or Spanish, Hindi, German, and so on—he was responsible. Microsoft could hire translators but there was no way to check their work until complaints rolled in. He could ask those who worked overseas in the French subsidiary to give the operating system a through vetting, but Windows is a humongous hunk of software and it was unlikely they would find everything. Plus, as with the American staff, it was difficult to get people already working full time to take on bug testing.

Rodrigues was well aware of Ross Smith’s game experiments at Microsoft; Smith even e-mailed an in-house newsletter around to engender interest. One day in Smith’s office, Smith was showing Rodrigues a user interface he was building, fooling around with magazine covers and asking, “Do you like it or don’t you like it?” It was a streamlined approach to a game, giving a player the absolute minimum action he would need to take to identify a bug. Rodrigues had a thought. He told Smith about the problems with the integrity of the languages within Microsoft’s operating system user interface. “What if we could do screen shots of the UI for all the languages and ask native speakers, ‘Do you like it or don’t you like it?’” he said.

Smith thought it was an excellent idea. They talked it through and agreed it should be a simple, Web-based application. A user wouldn’t have to install the latest software, which was a hassle. If a player had a spare five minutes, he could give a few opinions. On a twenty-minute break? Offer a bunch more.

They coded an automated system for grabbing screen shots of each page and set to work creating a game out of it. Instead of simple yes or no answers, he gave players markup pens in various colors. The player would review the text on a screen, use his mouse to circle it with a digital pen, and slide it into either the “Looks good” or the “Something wrong” bin. Then the next screen would pop up. But one challenge he and Smith quickly identified was the fact that language quality screen review was so easy for native speakers it wouldn’t be interesting or enticing, especially when a player was faced with multiple screens. To counter this they created game levels comprised of twenty-five dialogs or screens. Each time a player rated a screen he got a point. If he did all twenty-five, he advanced to the next level and earned a different color pen, then presented with twenty-five more images. Various in-game leaderboards kept track of the action. That way French players could see how they stacked up against their colleagues.

Over several months they released the game in all thirty-six languages and were pleased with the level of participation. Some players ran through a thousand screens. The head of the Japanese subsidiary closed the office for a day and ordered every employee to play the game. Then Rodrigues posted leaderboards pitting language against language—Japanese versus Portuguese, for example. And this drove participation even more. Rodrigues realized the game tapped into foreign workers’ sense of pride. Someone who worked for Microsoft in India or Brazil would hear it from friends and relatives if there were errors in the software. It made it seem that Microsoft didn’t care enough about their country and its users to fix them. These workers felt an urgency to fix every typo or malapropism.

In the end all 36 languages were cleaned up and more than 4,500 Microsoft employees participated, viewing a total of 500,000 Windows 7 translations. The highest participation was China, with 130 players viewing 2,600 screens. The lowest was Estonia, with 3 people looking at 96 screens. Of the dialogs that were submitted for further review, 29 percent were accompanied by player comments, which involved further explanation, even though this was optional.

The game was such a success that Microsoft extended it across other products. There was a game that gave points to engineers who responded the quickest to bug notices, and Microsoft Research released a game called Ribbon Hero that had players learn how to use Excel and other Microsoft Office software. The Windows Security team came up with an Olympic-themed game, with “runners” attempting something brand-new to innovate in a way that could help the company. Each was judged as a team, on how well they performed their regular work as well as coming up with something new and unexpected.

Meanwhile, Ross Smith has become a workplace games guru, giving talks and authoring papers, influencing others to experiment. While many of his cadres have adopted them in the workplace Smith has become the face of the movement within and outside Microsoft.

Over the past decade he has discovered that play leads to greater trust, and that encourages innovation. Collaborative play, such as his team-centric approach to bug testing, helps engender a climate of trust. Instead of his employees sitting in muffled coding silos, lost in their own little worlds, they’re working toward common goals. Stronger players assist weaker ones, and each gravitates toward what he does best to benefit the entire team. These higher levels of trust then spawn greater experimentation and risk taking, which are usually in short supply in most institutional structures. That then results in greater creativity and ultimately more innovation.

His experiences, Smith says, have taught him that players are motivated by different things. Some are in it for the glory and shame of the leaderboard or just want to best their personal high scores. Others like the game for game’s sake—to, say, crack a puzzle because they like puzzles, or participate because they enjoy the camaraderie of team play.

While productivity games can get “real work” done, like sifting through code to find bugs, adding one to an employee’s “day job” doesn’t work. It’s not good, he’s found, to have employees work toward amassing points, leapfrogging levels, climbing up leaderboards, or competing for prizes. That pits them against human resource systems and the paycheck they already receive. It’s also a bad idea to offer prizes. Then players are likely to approach the game as merely a mechanism for racking up points, which degrades the game’s ultimate purpose. Once, Bob Musson told me, they offered a mountain bike as a prize, and Musson found out a player had cheated by creating an automated program to play the game for him. But workplace games can function as a magnet to attract employees’ core skills (those that most people have, such as an ability to type, speak a language, donate their computer’s processing time, etc.) and improve “organizational citizenship behaviors.” These contribute to a healthier workplace environment.

As for Smith, he established an enviable level of trust in the bug-testing group and continues this as head of product testing in the Skype division. Wherever he goes, he reports extremely low employee attrition rates, by far the lowest of any department in the company.

“That’s what I’m most proud of,” he says.