19 min read

ProdFund 1.10: Where We Work

While the modern tech office has evolved a long way from its predecessors in 1720s London, there's a surprising amount of continuity. Work from home in tech, meanwhile, traces back to two surprising origins in 1970s California.
Open-office style desks off to the horizon, an open sky overhead as the sun sets in the distance.

Product Fundamentals is a podcast dedicated to spreading the core knowledge software product people need in order to succeed. Season 1 is structured as a history of how we make software.


In the penultimate episode of season 1, we're finally addressing the silent partner in all our work: the place we do it.

We'll track how we got from the first large office buildings in early modern London, through the rigorously optimized work spaces  of the early 20th century, on to the cubicle, the open-plan office, and finally, the the modern tech campus. We'll also cover the history of modern work-from-home, which has a lot more to it than the recent pandemic.

The audio for this episode is embedded below, and the episode transcript follows.

You can also find this episode of the Product Fundamentals podcast on the show website and through all the usual podcast services.

Transcript

Hello friends, and welcome to the Product Fundamentals Podcast, Episode 10: Where We Work.

This episode has been a bit delayed by some technical challenges and by me fighting with a cold – hopefully, both those issues have been sufficiently resolved now and my voice won’t be too raspy for this episode. My apologies, and thanks for bearing with me.

All right.

So far, we’ve spent nine episodes talking about the history of how we work as a matter of technology and process. But underneath all the changing methodologies and technologies has been the silent partner of place. And the place where we work, once seemingly so stable, was wildly disrupted by the COVID-19 pandemic. It has become a surprising new axis of change in the way we work.

This episode, to round out the story of how we came to work in the way that we do, and to understand the background of these ongoing changes, we’ll cover the history of the office and of working from home.

Early days

Let’s start with the office. The first organizations large and complicated enough to have offices were government- and government-adjacent. The British Navy and the British East India Trading Company, built the first large office buildings in 1726 and 1729, respectively. It had its most direct precursors in the counting houses of banks and trading organizations. 

Of course, organizations like the Navy and the East India Company predated large office buildings. But prior to the construction of their large offices, their operations were often scattered across small facilities, including houses that had been repurposed for business work.

The backbone of the early office workforce was the clerk. These early clerks were often undifferentiated workers. Anyone familiar with the modern Japanese notion of the job salaryman and the office lady, who are corporate workers that may move between functions without a single strong specialization, will recognize the British clerk. The clerk was responsible for all sorts of work requiring basic literacy and numeracy. In large organizations like the government and the East India Company, clerks could be shifted among various responsibilities within the bureaucracy; in small firms, the clerk were jacks-of-all-trades.

By the early 1850s, the British government had developed a sprawling but only loosely-coordinated bureaucracy, which was increasingly viewed as inefficient and corrupt. This led to the commissioning of the Northcote-Trevelyan report of 1854, in which prominent Conservative politician Stafford Northcote and Permanent Secretary of the Treasury Charles Trevelyan investigated the state of the bureaucracy and proposed reforms. Their report led to standardized entry exams to the British government bureaucracy – perhaps modeled in part on the Chinese imperial service exams – as a way to counter nepotism. They also recognized that there were two kinds of office workers, which they called mechanical and intellectual. The distinction between them would have a lasting legacy on our workplaces.

Mechanical workers weren’t machinists – rather, they were clerks whose jobs were simple and repetitive, such as those who processed forms, copied documents, and updated ledgers. Intellectual workers, meanwhile, were responsible for strategic and managerial tasks.

Administrative reform was a particular passion of Trevelyan, who worked on the issue from varying dimensions for many years. In 1856, Trevelyan argued to a Select Committee of the British Parliament that the government must make significant investments in the physical office spaces for the bureaucracy. He wrote,

"next to the appointment of proper officers, the thing most conducive to the public service will be to have proper offices."

Trevelyan was arguing from a perspective of financial prudence, as spending on large consolidated offices to replace scattered smaller facilities would pay for itself through workplace efficiency. Trevelyan wrote,

"One clerk in a well-arranged office, with ready access to all its papers, and in constant communication with his superiors, can do nearly as much work in a day, as two who are placed in distinct buildings, separated from each other, from their chiefs, and from the records of the office."

In constructing these new offices, Trevelyan thought the government must respect the distinction between mechanical and intellectual labor that he and Northcote had identified in their report two years prior. 

"For the intellectual work, separate rooms are necessary, so that a person who works with his head may not be interrupted; but for the more mechanical work, the working in concert of a number of clerks in the same room under proper superintendence is the proper mode of meeting it."

That is to say, managers need individual offices where they can focus, while most of the rank-and-file should be in one large room where they can be supervised.

Trevelyan was enthusiastic about colocation at a higher level of abstraction as well. Not only should the collections of scattered houses that had made up some ministries be consolidated into a single large building, but various ministries should be placed near one another to reduce communication costs and lags among them. In effect, he argued for a government office park.

The Northcote-Trevelyan report, and Trevelyan’s subsequent work, were important drivers of administrative reform that created the early modern British Civil Service, They informed construction of major ministerial offices in London, and helped shape one of the archetypal Weberian bureaucracies that were so influential on early 20th century business.

Scientific Management

By the time Weber was writing about work during  the 1910s, the modern office was going through another important evolution. The Scientific Management ideas of Frederick Taylor were extending from the steel mill, where they were first developed, to the office. This extension was largely done by followers of Taylor, such as William Henry Leffingwell.

In 1917, William Leffingwell – I assume no relation to the Dean Leffingwell of Scaled Agile Framework notoriety from our previous episode – wrote the book Scientific Office Management. As with Taylor’s work on factories, Leffingwell is laser-focused on how to fully optimize workers’ tasks, incentives, and environment to maximize the profitability of the business.

For our purposes today, the most interesting bits are about the structure of the office space. While Scientific Management has a well-deserved reputation for exploiting workers, there are plenty of touches that add to human comfort here as well: Leffingwell writes that while many states have laws requiring one toilet seat for every twenty workers, actually one toilet for every ten workers leads to more efficiency. He writes that businesses usually under-invest in heating and cooling for their workers; uncomfortable workers are inefficient workers, so managers should be sure to keep offices at a moderate temperature. Water fountains should be placed in multiple locations throughout the office, so that workers do not waste time on long walks to their nearest fountain. Leffingwell calculates that if 100 clerks walked 100 feet for a round-trip to their nearest water fountain five times a day for a 300 day work-year (remember, this was 1917) then the workers would collectively waste 3000 miles of walking per year just for their hydration. Finally, workers that use pins to hold together loose paper should be given pin cushions rather than dishes to hold pins, because pin cushions will reduce finger pricks which would otherwise lead to infections and poisonings – another source of inefficiency.

The Scientifically Managed office embraces both standardization and specialization. Functions can benefit from specialized equipment, such as specially-designed tables for letter-openers to better manage the inbox. However, workers within a department should use standardized gear, to reduce costs and make it easier to propagate the "one correct way" of doing a given task across all workers. And lest the core of Scientific Management be overlooked, Leffingwell is clear that the most important piece of equipment in any office is the stopwatch. It is the stopwatch that enables the manager or the consultant to measure the time-efficiency of workers actions so that the company can determine the right way to do a task and spread that method across all its workers.

Leffingwell’s scientifically managed office space was an open-office space, with clerks dispersed in rows of desks in one large open space, with straight paths breaking up each group of two or three desks to provide easy access to a nearby water fountain. In an office of 100 workers, just the four managers get their own private offices.

Before we move on, I hope you’ll indulge me one partial aside here: I was tickled to see a few pages in Leffingwell’s Scientific Office Management on the importance of cultivating esprit de corps to boost productivity. Leffingwell writes approvingly (pp. 49-51) of work by Joseph & Feiss, a Cleveland-based clothing manufacturer that had been an early adopter of Taylorism. Factory manager Richard Feiss had made a point to marry Taylorist principles with an apparently earnest commitment to Progressive-era welfare capitalism. Even though their workforce was mostly made of easily-exploitable recent immigrants, Joseph & Feiss had added reading and recreation rooms to their factories, hosted dance classes and encouraged sports leagues, and had found that workers who participated in the company choral league tended to be especially productive in their day jobs.

Shortly after Leffingwell’s book was written, Feiss would implement a five-day work week at his factory, down from the six-day industry standard. This was four years before Henry Ford would be credited as a visionary for doing the same thing in 1921.

So, the next time someone tells you that it’s just spoiled techies at cult-ish startups who get to enjoy cozy workspaces, extracurricular activities, and shorter working hours, you can tell them that actually we’re just trying to keep up with early 1900s immigrant garment workers in Cleveland.

Action Office and the cubicle

All right, let’s push forward to the modern office. By the mid-20th century, affordable electric lighting and climate control allowed for larger floor plates in offices. It was no longer so important to position workers near windows, or to otherwise maintain access to natural light. Along with these technological changes, post-World War II skepticism of the industrialized bureaucratic society put Taylorism on the backfoot.

As Europe in general, and West Germany in particular, rebuilt from the war, they built work spaces that broke up the monotonous rows of desks associated with the Taylorist era, using dividers, potted plants, varied furniture, and so on, to create spaces that invited teams to sit in clusters, and that created a more organic and unregulated flow of circulation as people moved through the space. This design movement, called Bürolandschaft, meaning "office landscape," rapidly spread through Europe in the 1950s, eventually reaching America through inventor Robert Propst.

In 1960, Propst wrote that

"Today’s office is a wasteland. It saps vitality, blocks talent, and frustrates accomplishment. It is the daily scene of unfulfilled intentions and failed effort."

Propst worked at furniture company Herman Miller, and in 1964, the company introduced its Action Office 1 product line, designed by Propst. The first Action Office series was built with small professional offices in mind, meant to encourage movement and flexibility. and its pieces were expensive and hard to assemble, and they sold poorly. However, the Action Office II product line, released in 1968, found a much more receptive market. Action Office II was built for the corporate office worker, using a variety of interchangeable standard components to let each worker create their own area in an open floor plan, with low walls creating a semi-private personalizable space.

Propst and his work partner, industrial designer George Nelson, intended for the Action Office line to be flexible enough to create a wide variety of working styles, and for workers to be able to change the configuration of their space frequently. Action Office included standing desks as well as sitting ones, and while some private space was valued, the kit was usually demoed showing each worker as having a 120 degree wedge to work in, rather than a 90 degree box.

However, Herman Miller’s corporate clients generally did not follow Propst’s vision. Instead, Action Office became – of course – the cubicle. The interchangeable components were cheaper than traditional office furniture, and the most space-efficient use was to place workers in individual square cubicles in a seated position, which made it easy for supervisors to walk a line and monitor the productivity of subordinates. Thus, the cubicle farm was born.

Not only was the Action Office space efficient; it even got a boost from the American tax system: new rules introduced in the 1960s let businesses claim tax deductions for various business expenses, and office equipment like cubicles had a much shorter, and therefore more attractive, depreciation schedule than permanent improvements like adding real walls and doors.

The cubicle’s originators were not happy with how their inventions were used. George Nelson contributed to Action Office I and split with Propst over the Action Office II line. He would later write in 1970, 

"One does not have to be an especially perceptive critic to realize that AO II is definitely not a system which produces an environment gratifying for people in general. But it is admirable for planners looking for ways of cramming in a maximum number of bodies, for 'employees' (as against individuals), for 'personnel,' corporate zombies, the walking dead, the silent majority. A large market." (source, pp. 219-220)

Propst would come to share the same sentiment, commenting to the New York Times in 1997 that his intention had been to, 

"give knowledge workers a more flexible, fluid environment than the rat-maze boxes of offices,"

but that this concept got lost as companies embraced open-plan cubicle farms to cut costs. Propst continued, 

"The cubicle-izing of people in modern corporations is monolithic insanity."

Nonetheless, by the mid-90s, the International Facility Management Association estimated that 40 million Americans worked in cubicles, which amounted to 60% of the white-collar workforce.

The tech office

What of the technology industry? It turns out we have long been divided about dividers.

Bell Labs’ main facility in Holmdel, New Jersey, which opened in 1962, offered most engineers small offices, and then had deliberately wide hallways and open areas which were intended to stimulate spontaneous conversations. Hewlett-Packard, whose HP Way we touched on in Episode 3, used open floor plans from its early days in order to stimulate more impromptu communication.

Andy Grove’s Intel proudly used cubicles for everyone, including executives, from shortly after its founding in 1968.

Microsoft, meanwhile, spent its early decades firmly committed to each engineer having an office. For fellow software workers who are confused by what I just said, an office in this context means a small room with a door, and that room was for the sole use of just one worker. I know, it’s hard to imagine. But Microsoft wasn’t alone; Sun Microsystems offered private offices for 75% of its workers in the 1990s.

But then, the tide turned. IBM, which had also long offered private offices to its office workers, began to switch to open floor plans in 1997.

Around this time, the cubicle farm that had dominated the 1980s and 1990s began to give way to the “bull-pen” style of open low desks that had long been associated with journalistic newsrooms.

Indeed, the dot-com bubble seems to have marked an inflection point in the shift toward open – and more densely packed – offices. New companies like Netscape embraced low-walled cubicles and open spaces that were meant to foster teamwork. Data show that commercial real estate per office worker across all industries in the US reached its modern nadir in 2001, at 171 square feet (or 16 square meters). 

Space per office worker rebounded after the dot-com crash, and again during the Great Recession, but had nearly slid back to its 2001 low in 2019 before the COVID-19 crisis threw everything out the window. 

In 2005, Google renovated its offices into the Googleplex, the first of the modern wave of tech office campuses replete with amenities far removed from traditional office functions. Of course, it was built around large bull-pen style spaces. Facebook would follow suit in 2012 with its own campus on the site of Sun Microsystems' former offices in Menlo Park. Shortly thereafter, Facebook began work on an expansion space for their headquarters: a single open space for at least 2300 workers. Mark Zuckerberg described it thus:

"The idea is to make the perfect engineering space: one giant room that fits thousands of people, all close enough to collaborate together. It will be the largest open floor plan in the world, but it will also have plenty of private, quiet spaces as well.'

The bull-pen office design has its defenders and its supposed advantages, beyond just the cost savings of fitting more people into a fixed space with lower construction costs. Some of these advantages were meant to dovetail with software methodology: a team that shared a tight work area would presumably communicate more. The New New Product Development Game of 1986, which had inspired Scrum, had extolled the advantages of co-locating members of all different disciplines in one room so that they could each see what the others were doing and cross-pollinate ideas. Extreme programming was predicated on pair-programming, which is hardly doable from separate private offices.

But the bullpen-design open office has also come in for plenty of criticism. The common complaint in polling of employees, is, of course, audible distractions. Serious noise canceling headphones are part of the standard tech worker toolkit for a reason. The loss of privacy and the added pressure to always appear focused and productive, as well as the need to be seen putting in at least as much time as colleagues, are other common complaints.

There are challenges to the better-collaboration hypothesis too: research reported in a December 2019 article in the Harvard Business Review found that when firms switched from cubicles to open offices, face-to-face interactions between workers actually fell by 70%. The researchers hypothesized that this was because the open office collapsed the distinction between public and private space, and as a result, workers treated their colleagues as if they were always in a private mode, even when they were in a nominally public space.

Finally, the health consequences of the crowded open office have been a concern since before COVID-19. An academic study from 2011 based on a survey of Danish workers found that workers in open offices took 62% more sick days than workers in individual offices. Subsequent studies from Sweden and Germany also found more sick days used in open offices.

One last note on the modern tech office: The rise of the tech campuses saw facilities like on-site laundry and barbers enter the workspace, and these became emblematic of Big Tech. But it’s worth noting that large companies across industries have long offered fringe benefits as a way to attract talent, and to use business tax-deduction rules to offer essentially untaxed income replacement to workers. 

I touched briefly on the welfare capitalism perks enjoyed by garment workers at Joseph and Feiss in the early 20th century. Mining towns and company towns in the industrial era also offered free or subsidized services, meals, sports fields, and so on. By the 1980s, it seems that large companies like General Electric and PepsiCo were offering fitness centers. 

But perhaps nothing was a stronger cultural shibboleth of the American tech industry in the 2000s than free lunch. In addition to the upsides gained by the company from keeping workers in the office and in contact with one another for more of the day, the free lunch phenomenon was boosted by tax treatment. Until 2018, US tax law allowed companies to deduct the cost of serving lunch to workers as a business expense. Now, the IRS is cracking down on the practice, and by 2025, the cost of free lunches will be treated as taxable income for the worker, making the deal much less attractive. Between this tax change and reduced office headcounts in an age of hybrid and remote work, we may soon see this symbol of the 2000s tech boom fade away.

Working from home

Meals and fringe benefits aside, with all the drawbacks of the modern bull-pen open office space, it is unsurprising that a significant share of software workers have eagerly embraced working from home during the COVID era. 

I’ve tipped my hand to holding a few controversial opinions over the course of this podcast season, but now it’s time for my most scandalous confession of all. I will take a stand on the most divisive issue in tech today and admit: I’m an office guy. While I acknowledge that there are lots of drawbacks to the open-plan bullpen office, and it’s certainly nice to have the flexibility to work from home sometimes, on balance I much prefer working from a nice tech office most of the time rather than working from home.

But in researching for this episode, I was somewhat surprised to learn how common work from home has been in the tech industry – since well before COVID, and even since before the Internet.

Don’t worry, I’m not going to launch into the premodern history of working from home, other than to acknowledge that for most of human history, most work was done at home or in a shop attached to the home. As we’ve already discussed, the modern office building is only perhaps 300 years old, and before that, the small offices of lawyers and bankers only push that history back a bit more.

Modern large companies have only really had employees working from home since the 1970s. The terms “telework” and “telecommuting” were coined in 1973 by Jack Nilles, an American rocket scientist turned consultant. Nilles became a persistent advocate for “telework” – that is, getting work done someplace besides a company’s central office – pushing large companies, the US federal government, and the California state government to study the topic. He finally achieved some successful pilot programs in the late 1980s and early 1990s.

Nilles often cast the benefits of telework in terms of the hot button issues of the age: air pollution and oil prices. By saving office workers from commuting, gas consumption and traffic could be reduced. 

In his 1994 book, Making Telecommuting Happen, Nilles imagines that most workers will still go to the office occasionally. How often they need to go to the office will be determined by the share of their job tasks that require face-to-face communication. Concentrate tasks that require face-time onto a small number of days, and then telework the rest of the time. 

Nilles was not expecting that employees will necessarily work from their homes. Rather, he imagines most employees going to telework centers, either as small satellite work locations belonging to the worker’s company, or local branches of a network like today’s coworking spaces. Actually working from home, he writes, may only apply to a relatively small share of workers.

Nilles wrote his 1994 book with a skeptical manager in mind, recommending that telework be offered only to the workers who could handle it responsibly. Working from home is not for everyone! Workers with compulsions, such as overeating, workaholism, and drug abuse, should not work from home. The company may need to inspect the worker’s home to assess whether it is a suitable place to work.

As dated as some of Nilles ideas may be, he was prescient when it became to the future of the technology. Writing in 1994, he discusses at length Moore’s Law, and the inevitable arrival of high speed access to "public meta-networks," by which he means the Internet. If anything, he was a bit optimistic, anticipating that "video conferencing will be almost a standard item on mid-cost personal computers in the late 1990s." He also expected collaboration software to rapidly improve, enabling workers across the world to share notes and drawings.

Jack Nilles may have had a touch of quixotic futurism, but working from home also has a pedigree with some of the most established organizations around. In 1979, IBM started an experiment in remote work – apparently in order to reduce the strain on a lab mainframe in San Jose – by installing work terminals in five local employees’ homes. The experiment was seen as successful, and IBM rapidly rolled out work from home to more functions. By 1983, some 2000 IBM staff were working remotely. The company leaned into work from home, and by 2009, 40% of IBM’s 386,000 workers had no office. By offloading unneeded office space, IBM said it saved about $100 million per year.

Remote work got more big organization credibility in 2010, when the United States passed a law requiring federal agencies to establish guidelines under which most government employees could work from home. 

More technology companies, including Yahoo, made a name for themselves with permissive work from home policies. Remote-first and remote-only companies emerged as well. In 2020, devops software company Gitlab proclaimed itself the world’s largest pre-pandemic fully remote company, with 1200 remote workers.

The COVID disruption

In 2019, just before the COVID-19 pandemic made work-from-homers of us all, a Gallup poll of US workers with "remote-capable" jobs found that 8% worked exclusively remote. But another 32% described themselves as hybrid – that is, working at least 10% of the time remote, but less than all the time.

No listener to this podcast in 2023 needs a recap of the COVID-19 pandemic. I’ll just note that in May 2020, 70% of Americans with remote-capable jobs were working exclusively remote. By May 2023, that exclusively remote share had fallen from 70% to 29%, while the hybrid share stood at 52%. Just 20% of workers in jobs that could be done remotely worked full-time at an office.

Software workers almost certainly have more remote and hybrid work arrangements than those economy-wide figures indicate. In a separate 2016 poll that broke out data by industry, 57% of workers in the computers and information systems industry reported working remotely least some of the time, compared to 43% of the overall population.

The pandemic was a rollercoaster of changing company policies. Many software companies initially leaned into remote work, suggesting that work-from-home might remain a viable option for employees forever. We’ve since seen some of these companies, such as Google, apply a mix of carrots and sticks to pull staff back to the office.

While it’s easy to see this turn as a one-time post-pandemic adjustment, it’s actually entirely consistent with the modern history of work from home, which has involved lots of swinging back and forth.

IBM, which saved so much money through its decades of embracing remote work, drew criticism for requiring its 2600-person marketing organization to return to its offices in 2017. IBM was hardly alone – in 2013, Yahoo canceled its work from home policy; in 2014, Reddit did the same. Plenty of other non-software businesses vacillated on remote work before the pandemic as well.

There’s a credible enough story that suggests this is rational behavior. There is increasingly voluminous academic and corporate research on the relative merits of remote work, hybrid work, and office work. Each claims its real advantages. Some of the findings conflict with one another.

To a terrible loose first approximation, it seems that remote work may result in higher productivity for workers doing familiar solo tasks and tasks that require high concentration. In-office work results in richer and more accurate communication, better alignment and interpersonal trust, and more collaboration.

Critical to the persistent utility of the office seems to be the Allen Curve, first published about by MIT business professor Thomas Allen in 1977. The curve describes the inverse relationship in communication frequency between people based on physical distance. Empirically, we are four times as likely to communicate with someone 6 feet away as someone 60 feet away, and almost never communicate with someone located far from us. Strikingly, this finding holds true across media: emails and phone calls suffer the same drop-off. 

This actually lines up well with one of the principles published alongside the Agile Manifesto in 2001: 

"The most efficient and effective method of conveying information to and within a development team is face-to-face conversation."

A more recent study by MIT researcher and consultant Ben Waber found that software engineers discussing a problem exchanged 38 acts of communication if they were in the same office, versus just 8 messages if they were in separate locations.

If, as research and intuition suggest, different tasks favor different work settings, then we should expect that individuals, teams, and entire companies will go through cycles when remote work, office work, or some combination of both will be optimal. 

The way we work is shaped by where we work, and the right place to work may be just as shaped by what we’re working on.

Wrapping up

That’s all for this episode.

Join me next time as we bring this first season to a close by teasing out the themes and forces that shaped how we work today, and identify what the past tells us about what we can expect in the future of how we build software.

As always, your comments and feedback on this episode are very welcome. You can find a transcript, links to sources, and ways to reach me on the show website at prodfund.com

And if you like this series, and you want to hear more, do me a favor and share it with someone you think would enjoy it too.

Thank you very much for listening.