Are ‘ghost engineers’ real? Seeking Silicon Valley’s least productive coders.A Stanford researcher says data on programmer productivity suggests 14 percent of remote software engineers get barely anything done.
https://www.washingtonpost.com/technology/2024/12/08/ghost-engineers-programming-productivity-coding/Yegor Denisov-Blanch is a Stanford business school researcher, not a medium, but he’s spent recent weeks communing with phantoms. Many have been angry.
The deluge of messages began after the 32-year-old grad student posted a summary online of his findings from analyzing data on the productivity of programmers from hundreds of companies worldwide.
About “9.5% of software engineers do virtually nothing,” Denisov-Blanch wrote on X, contributing so little code he suspects they are either slacking off or secretly cashing two tech paychecks. He termed such freeloaders “ghost engineers.”
After Denisov-Blanch’s post was viewed nearly 4 million times, some self-professed ghost engineers contacted him. In messages that ranged from defensive to profane and enraged, emails viewed by The Washington Post showed, they confessed to taking advantage of what they view as broken corporate cultures — often claiming it wasn’t their fault.
Denisov-Blanch shared his findings, from ongoing research not yet published in peer-reviewed form, after a Silicon Valley investor reignited a long-running debate over the existence of ghost engineers.
“Everyone thinks this is an exaggeration but there are so many software engineers … who I know personally who literally make ~2 code changes a month, few emails, few meetings, remote work, < 5 hours/ week, for ~$200-300k,” Deedy Das said in an X post one morning last month.
Das, who didn’t respond to an interview request, listed 13 companies where he said such behavior is common, from networking pioneer Cisco to cloud giant Salesforce. He detailed a few tips and tricks on how ghost engineers get things not done. They include heavy use of the “in a meeting” status in workplace chat apps and using low-cost gadgets called mouse jigglers to simulate constant activity.
Aaron Levie, CEO of Box, one of the companies Das named as a hotbed of underperformers, saw the post that morning. Late that evening, he responded on X: “This has been a particularly constructive day.”
In a phone interview, Levie said that while he did not immediately fire anyone, the online debate sparked fresh conversations at his company about an issue leaders were already tackling.
Over the past four years, as remote work has ballooned within the tech industry, the company has become focused on measuring all workers’ productivity, not just engineers. Box has trimmed teams to avoid overlapping responsibilities, Levie said, pared back meetings and become choosier about research and development.
Denisov-Blanch, who taught himself to code as a teen, said in a phone interview that he didn’t set out to hunt techie ghosts.
But the phenomenon became obvious, he said, after creating a machine-learning algorithm that digests a company’s collection of code to analyze programmer productivity, conceived in collaboration with Stanford associate professor of organizational psychology Michal Kosinski and entrepreneur Simon Obstbaum, former chief technology officer of anime streaming service Crunchyroll.
The team’s data showed large companies are more susceptible to ghost engineers, although smaller ones are not immune.
It can be difficult to assess engineers’ productivity, Kosinski said in a phone interview, because of the complexity of the work and labyrinthine nature of some large tech companies. But not identifying and rooting out underperformers “promotes charlatans and decreases rewards that the good-performing employees are going to get,” he said.
The Stanford research comes at a time when some large tech companies are rolling back remote work policies instituted during the pandemic and after a wave of layoffs at firms including Google, Amazon, Meta and Microsoft. Beginning in January, Amazon wants its workers in the office five days a week. SAP, AT&T, Dell and Zoom are among those that have rolled back more flexible policies.
Denisov-Blanch said the Stanford research found more top-performing programmers worked from home rather than the office, but also that ghost engineers are more likely to work remotely. They made up 14 percent of remote software engineers, compared with 9 percent who work in the office at least part of their week and 6 percent who work in an office daily.
Denisov-Blanch sees ghost engineers as more defeated than devious. “It almost always starts with frustration with their jobs and not seeing a clear link between effort, reward and recognition,” Denisov-Blanch said, a conclusion he made after emailing or speaking with dozens of ghosts. “They lose motivation — and as they lose motivation, they start performing lower and lower and lower.”
Over time, that reduced productivity moves from being passive to active, Denisov-Blanch said, with workers adopting tricks such as putting artificial time blocks on their calendar or posturing about the importance of their workload. “Sometimes it’s hard for managers to know the truth,” he said.
Early in his career as a software engineer, Krunal Patel experimented with ghosting, not out of laziness but to give his manager a reality check. For one week, he and a co-worker gave their manager detailed twice-daily updates on engineering tasks that obscured the fact they were accomplishing nothing. Their boss didn’t notice the pair were slacking — until they confessed.
“He was shocked,” Patel said in a phone interview. “We sat him down, and we explained that we were frustrated.” Patel and his colleague asked for less micromanagement of daily tasks and more deep engagement on the problems they were trying to solve.
Patel’s manager took up their suggestions, and the engineers got more done. “We were a little more efficient. We enjoyed working,” said Patel, now a tech executive and entrepreneur with 20 years’ experience in the software industry.
Sudheer Bandaru experienced a similar outcome while managing a team of software engineers at a medium-size company.
When performance appraisals came around, the “brightest” engineer, who was talkative in team meetings, turned out to produce almost no code at all, Bandaru said in a phone interview. “It was a shocker,” he said.
When Bandaru talked with the worker, he turned out to be misdirected not malevolent. “He was more of a research person, who never wanted to sit in one place and do coding,” Bandaru said. Once switched into a different role, the employee thrived. Bandaru said that multiple instances like this one inspired him to create Hivel, an analytics platform aimed at helping companies develop software faster.
Used the right way, ghost-busting software that monitors techie productivity could bring about more heartwarming managerial moments. But Patrick McKenzie, a tech industry writer and former software engineer, says tools that merely look at the lines of code engineers have written risk causing false accusations.
Some senior engineers “can very reasonably commit no code,” McKenzie said, for example because they are planning out software architecture or training new engineers. “That person isn’t a ghost, and that person isn’t coasting,” he added.
Denisov-Blanch said that to avoid that trap, the algorithm he and his collaborators used to gauge productivity can monitor the impact that both teams and individuals have on a company’s overall collection of code.
He’s exploring how to commercialize it after his viral findings caught the eye of investors. That might allow companies to decide what to do about the ghosts haunting their payroll.
But Denisov-Blanch said he aims to help ghost engineers, not snitch on them. “My mission isn’t to find these people,” he said. “My mission is to understand why this phenomenon occurs and make it not occur.”
Total Posts: 10261