Save this storySave this storySave this storySave this story
On Saturday afternoon, Elon Musk posted on X that “all federal employees will shortly receive an email requesting to understand what they got done last week.” He added that noncompliance would result in termination. Within hours, an e-mail from the Office of Personnel Management, with the subject line “What did you do last week?,” appeared in the in-boxes of millions of federal employees. “Please reply to this email with approx. 5 bullets of what you accomplished last week and cc your manager,” it began. The deadline for this task was Monday, 11:59 P.M. Eastern Time. Right before Musk took over Twitter, in 2022, he had texted this same question to Parag Agrawal, the company’s C.E.O. at the time. Soon after, Agrawal was fired. It seemed, for the moment, that the entire federal government would now be subjected to a similar dismissive scrutiny. But then the plan began to fall apart.
Later that same night, Musk clarified the O.P.M. request in a series of backtracking posts on X. “The passing grade is literally just ‘Can you send an email with words that make any sense at all?’ ” he wrote at 10:36 P.M. “To be clear, the bar is very low here,” he added, ten minutes later. “An email with some bullet points that make any sense at all is acceptable!” The next morning, he implied that maybe the real purpose of the e-mail request was actually just to uncover fraud: “We believe non-existent people or the identities of dead people are being used to collect paychecks.”
At the same time, various Trump-appointed agency and department heads began publicly advising their employees to ignore the request. “The F.B.I., through the Office of the Director, is in charge of all of our review processes,” wrote Kash Patel, the newly confirmed F.B.I. director, in an e-mail to his staff. “For now, please pause any responses.” Tulsi Gabbard, the new director of National Intelligence, issued similar instructions to her charges, citing the “inherently sensitive and classified nature of our work.” By Monday morning, employees at many agencies, including the Departments of State, Defense, and Homeland Security, had also been told not to respond pending further guidance. Later that afternoon, O.P.M. announced that answering the e-mail was voluntary.
From Musk’s perspective, the chaos surrounding this hasty order isn’t that important. As with many of the recent actions of his Department of Government Efficiency, what matters is the signal it sends: Musk wants to be seen as a productivity Prometheus, bringing Silicon Valley’s move-fast-and-break-things effectiveness to the lumbering operations of the federal government. If these efforts falter, it’s just further evidence of the entrenched nature of the bureaucracy he’s battling.
The problem with this heroic mythology, however, is that it’s based on a faulty premise. Musk wants the world to believe that the nimble tech sector has already figured out the keys to knowledge-worker productivity. But, if this was the case, why did Twitter devolve into chaos soon after Musk’s takeover, as he introduced and then cancelled multiple employee-evaluation schemes before suddenly firing half his workforce without further explanation? As it turns out, the core question of that O.P.M. e-mail from Saturday—What are employees actually doing?—is one that Silicon Valley itself has been struggling with since its early days.
In the nineteen-forties, a young scholar named Peter Drucker was invited to study the operations of General Motors, then the world’s largest corporation. It was opportune timing for Drucker because G.M. was tackling a productivity challenge. The company knew how to manage its automobile factories. Following the approach perfected by Henry Ford earlier in the century, it divided the process of building a car into small steps—say, attaching a steering wheel or winding the wire on a magneto—that could then be assigned to assembly-line workers. Productivity in this context was easy to manage: measure the pace at which individuals completed their well-defined tasks.
Managing G.M.’s offices proved trickier. The new class of deskbound workers who populated G.M.’s expanding administrative apparatus was juggling complicated portfolios of projects with rapidly shifting demands. There was no single best way to tell them to go about their daily activities, nor was there an equivalent of a growing stack of steering wheels or magnetos to indicate how productive they had been.
Amid this upheaval, G.M.’s long-celebrated C.E.O., Alfred P. Sloan, Jr., began to articulate a radical new approach to managing the office: focus on outcomes over execution. Drucker recalls Sloan explaining that a successful manager in this new environment “must be absolutely tolerant and pay no attention to how a man does his work.” It was better to instead provide clear goals and then, later, check to see if they were accomplished. Drucker absorbed these ideas and eventually elaborated on them in his influential 1954 book, “The Practice of Management,” with a strategy he called “management by objectives” (M.B.O.).
When the new digital-technology sector began to take off in Northern California, its managers agreed with Drucker. In 1957, Hewlett-Packard introduced a collection of corporate reforms, which came to be known as the H-P Way, and which were inspired by Drucker’s ideas. “As the company grew and it became evident that we had to develop new levels of management, we applied our own concept of management-by-objective,” wrote Bill Hewlett, in a retrospective about this period. In the early nineteen-seventies, the idea then took hold at Intel, when one of its original leaders, Andy Grove, began to aggressively push his own version of M.B.O., which he called O.K.R., short for Objectives and Key Results. (He complained that his old employer, Fairchild Semiconductor, valued expertise above all else, which led to a lack of “achievement orientation.”)
In 1975, a Harvard Business School student named John Doerr made his way to Intel for a summer internship. While there, he attended a seminar taught by Grove, who extolled the importance of O.K.R.s. As Doerr recalls, “Grove called it a ‘very, very simple’ system, knowing ‘simplicity’ was catnip to an audience of engineers.” Five years later, after Doerr joined the venture-capital firm Kleiner Perkins, he began spreading the O.K.R. gospel throughout Silicon Valley. In his 2018 book, “Measure What Matters,” Doerr writes that an O.K.R. starts with identifying an “objective,” which, true to Drucker’s original concept, captures “WHAT is to be achieved.” This must then be followed by several key results, which “benchmark and monitor HOW we get to the objective.” These should be specific and time-limited, and their completion should be measurable. (It was the addition of these key results that differentiated Grove’s O.K.R. from Drucker’s M.B.O.) Doerr provides the example, taken from the original seminar taught by Grove, in which the objective is to dominate a new sector in the microcomputer-component market, and one of the key results is to come up with ten new designs for a component. The objective is ambitious but vague. The key results are concrete and something managers can track. In the world of O.K.R.s, the question of “What did you do last week?” isn’t casual but instead a request for measurable updates on specified milestones.
By his own accounting, Doerr introduced O.K.R.s to more than fifty companies, with the most famous being Google. In 1999, soon after investing $11.8 million for a twelve-per-cent stake of the young Internet-search startup, Doerr gave a PowerPoint presentation about O.K.R.s to a small group of Google employees, including co-founders Larry Page and Sergey Brin and future C.E.O. Eric Schmidt, who became “tenacious, insistent, even confrontational in their use of OKRs.” As Google exploded into the technology-startup stratosphere, O.K.R.s became de rigueur in Silicon Valley. They’re now embraced to some degree by many, if not most, major companies, including Microsoft, Amazon, Netflix, and Uber.
O.K.R.s, in other words, have become the closest thing Silicon Valley has to an official answer to the question of how to measure their employees’ productivity. On paper, this management-by-objective approach makes good sense. In practice, however, it hasn’t proved to be the silver bullet that its boosters had originally hoped.
A product manager at a major technology company recently told me that O.K.R.s are easy to game. “Imagine an engineer is tasked with launching a new product,” he said. “They might release an incomplete project and arbitrarily declare it done, because that satisfies their first key result, and then spend the rest of their time fixing their own bugs—including some that they maybe intentionally put in there—as this satisfies their second key result.” When it comes time for a performance review, the engineer can then brag about all their “progress” on their milestones. “There can be a fair amount of manipulation that way,” he concluded.
Similar complaints are easy to find online. A Reddit thread from last year titled “What’s your experience with OKR’s” includes a commenter who notes, “We are forced to use it in my company, but nobody seems to use it in reality.” Another commenter admits that many employees at his company retroactively label work they were already doing as key results for made-up O.K.R.s: “They were gonna do that anyway, but it now needs to be labeled as an OKR.” Someone else offers an even blunter conclusion: “I’ve never seen them executed well.”
Issues like these are not the result of easily fixable flaws in Drucker’s original concept. They instead reflect a deeper reality: assessing knowledge-work productivity is a wickedly hard problem. These jobs are inherently ambiguous with ever-shifting arrays of tasks and projects that are tackled idiosyncratically. This freedom isn’t bad. It promotes creativity and a sense of agility that makes work less draining than rote industrial labor, and when harnessed correctly, it can lead to spectacular breakthroughs, like Intel’s microprocessors or Google’s search algorithms. But as knowledge organizations grow too large for their efforts to remain oriented around the drive and vision of singular leaders, this same freedom makes effective management a daunting task. Even Drucker came to realize that he may have been too optimistic about the potential of M.B.O. to fully solve these issues, writing in a 1999 paper, published more than four decades after he introduced his theory, that “work on the productivity of the knowledge worker has barely begun.”
All of which brings us back to Musk’s performative gesture of asking millions of federal employees to report what they had accomplished during the preceding week. Musk would like his fans and followers to believe that this question should be easy to answer, and that he’s reintroducing commonsense productivity to an obstinate public sector. But the reality is that, even in the cutting-edge tech world where Musk earned his original fortune, this simple query has created endless headaches, including for Musk himself. When he acquired Twitter, he replaced its more formal O.K.R. framework with a more informal system through which he could imperiously and unpredictably demand evidence of progress whenever the mood struck, such as when he told all of his engineers to “print out 50 pages of code you’ve done in the last 30 days.” But, according to reporting from The Verge, he soon lost interest in trying to assess individual employees, cancelling the threatened code review. He soon began issuing more erratic declarations, threatening to fire anyone who missed a deadline and demanding that managers rank their staff. Engineers were reduced to monitoring Musk’s Twitter feed to learn about changes to their product plans. Musk’s antics at Twitter might be extreme but it’s true more generally that, after more than seventy years of trying, some of the most prominent companies in Silicon Valley continue to struggle to figure out how to best measure their workers’ productivity. This is a major problem worthy of serious investigation. But if Sloan, Drucker, Hewlett, Grove, and Doerr couldn’t fully solve it, a thirty-seven-word e-mail is unlikely to do the trick. ♦
Sourse: newyorker.com
