I’ve worked both in private offices and in open plans. And my experience is that private offices are good for those occasional long deep work sessions but overall the team in total is more productive and happier in open plan. Not everything is about your personal productivity and it’s all too common to see separate team members deeply focused on non-aligned items which makes the joined work much much less than the sum of the parts.

I've worked in private offices, private cubes, and open office.

As much as I personally prefer isolated offices/cubes, the last 4 years in a mostly open office have definitely given me sympathy to the argument for them. I personally have noted a level of collaboration and information sharing that just didn't happen in the more isolated environments. There is something pragmatically useful about having people be able to jump in with their feedback or expertise.

The more isolated people are the more you have to go out of your way to ensure information dissemination and feedback channels are available and utilized. It certainly can work, but I haven't seen it work as well.

Open rooms promote teaching, spontaneous design discussions, etc, that just doesn't happen in the same way over e-mail, chat, stand-up meetings, etc.

There are arguments to go both ways. I used to hate open offices, now I'm sympathetic to them.

A key to making the open office tolerable is that it should not mean no isolation. There are ways to address their concerns. eg, headphones are essential. Obscuring line of site is very helpful. My current situation has me in an area of "open rooms", where rooms with ~11 desks are arraigned in an open office-esque fashion. It's pretty decent. Not super-open, but people can "share space" with the people relevant to them.

> I personally have noted a level of collaboration and information sharing that just didn't happen in the more isolated environments

I think it's important that we don't overly focus on one isolated experience point like yours, because empirically this is highly disputed.

In my experience, information sharing has plummeted after moving to an open-plan layout. People are forced to use noise-cancelling headphones to avoid impromptu exchanges, because the impromptu exchanges are value-destructive in the first place. And when someone gets co-opted into an impromptu discussion, they try to keep it superficial and share less substantive info, to get away from the unplanned distraction as fast as possible.

Not to mention the huge increase of totally not-work-related or irrelevant distractions, like loud sales calls, product discussions that don't affect anything for my team, discussions about weekend plans, etc.

It's a very shallow notion of information sharing, which could happen asynchronously through code, documents, email, or with a scheduled video call, or a short scheduled meeting, etc.

As I mentioned elsewhere, this horribly misguided idea that somehow constant, real-time audio communication == sincere collaboration or information sharing, this idea is really destructive and doesn't map very well to how engineers actually work.

> I think it's important that we don't overly focus on one isolated experience point like yours, because empirically this is highly disputed.

I'm not saying my experience generalizes, I doubt every environment would see the same benefit. My point is that even though I started exclusively preferring one side, I can see value on the other side.

> It's a very shallow notion of information sharing, which could happen asynchronously through code, documents, email, or with a scheduled video call, or a short scheduled meeting, etc.

Far from shallow. Technically anything can be communicated in any environment, that's not very interesting. What's interesting is how pragmatically an environment actually works and what it's real life pros/cons are.

But an open-plan office layout assumes that all communication happens most effectively with only one possible type of communication: real-time, constantly preemptible audio streams.

It’s shallow to say that “information sharing” happens this way, since for many people it obviously impedes or completely prevents information sharing.

> “What's interesting is how pragmatically an environment actually works and what it's real life pros/cons are.”

I agree on this, which is why it only requires such a short analysis to see that open-plan offices fail so one-sidedly. They are empirically shown to be widely disliked, to lower morale, to lower producivity (both individually and overall), to lead to more superficial interaction and less deep communication, to lead to more defects in knowledge work outputs, to increase communicable disease transmission and negatively affect sick time and vacation time habits, all while entirely discounting the most pragmatic working styles of at least one huge group of people (introverts) and, when all is said and done, they don’t even save money except in the shallowest, short-term sense, and often companies spend on opulent luxury features in order for the workers to appear essentially as decorative office furniture for when investors or upper management walk by.

It is more than fair to call this phenomenon shallow.

I'm in a similar setup right now - open rooms with 12 people per room. We have huge desks that leave plenty of space between each other and dividers separating desks that face each other. You only see another person if you turn 90 degrees from your forward facing position. We also have private rooms of various sizes that can accommodate one, two or five people if you need to be alone. Some of my colleagues use the rooms a lot, I personally rarely do. Overall, I really really like it, my experience mirrors your own.

The best for me was at a prior company with high partition walls blocking the office into large cubes each with 3 or 4 people in. It was flexible enough to reconfigure and as long as you were on the same wavelength as your cubemates regarding distractions, it offered the privacy to focus too. I think that company shifted to fully open plan with hot desking sometime after I left.

I've also worked in a variety of different floor designs, and have found open-plan to be dramatically inhibiting, not just for me but also for direct reports who tell me the open plan setup prevents them from getting things done most hours of the day.

What do you propose for developers whose jobs consist primarily of deep focus tasks that require quiet, privacy, and states of flow?

(I would argue this is the majority of developers, but that is beside the point. Even if it's a minority, yet their work output is very important for a given company, it would still seem that embedding them in an open-plan layout they must sit in for the majority of the time would clearly be throwing away more money they could possibly be saved on the real estate. Or you disagree?)

You say, "Not everything is about your personal productivity" but this seems mostly irrelevant, because we're in a section of the possible solution space that focuses on never accounting for developers' personal productivity. Separately, if you're on a team where developers have to invest deep focus into disparate parts of a system or disparate solution approaches (the majority of teams I've ever seen), then it does boil down to the sum of individual productivity for most things.

Absolutely, this discussion always devolves into good vs bad and understandably so because it's very frustrating when your work environment doesn't facilitate your performance. But the typical knowledge worker actually functions on some blend of the manager and maker schedules -- sometimes they need to do deep work without interruptions, sometimes they need to look up from their desk and communicate to make sure their output is aligned with the company. A private space free of distractions facilitates the former but an open plan is actually pretty good for the latter.

The best solution (I'm surprised more companies have not implemented this) is to provide both environments and give the employee some guidance in terms of where they choose to work. Most knowledge workers will not benefit from the monastic strategy described in this article but almost all will benefit from a bimodal or rhythmic strategy.

If you already have an open plan office this is as easy as telling the employee he/she can work remotely a few days a week as long as they select a quiet space that's free of distractions (so if they have kids running around at home during the day, maybe better to advise them to go to the library instead).

The manager should provide some guidance in terms of how much time the employee spends remotely vs. "on the floor" with everyone else. Graham's maker vs. manager article is great on this topic,engineers often err a little too far in the direction of isolation, managers err too far in the direction of having everyone in the collaborative environment, the solution is a little dialogue.

With very light guidance and very little additional cost to the business you can improve both productivity and morale this way. Our team functions best very far down the deep work end of the spectrum -- we have one day a week where everyone goes into the office or gets on calls and gets aligned. For the rest of the week communication is mostly async and work is mostly remote.

The trouble is that the distribution of work is not an even split between work that requires deep focus and other work that benefits from disruptive audio communication.

Most engineering work requires deep focus. So if the office was designed to allow the majority of work to be private, quiet, and deeply focused, but with occasional meetings or break-out social rooms for the dynamic discussions, that would be great.

Instead, it is designed in the wrong-headed, opposite way: all work is embedded into the dynamic, real-time audio distraction stream. You end up needing to compete for conference room reservations, or hide form people, or listen to music when you don't want to, or abuse a work-from-home privilege, just to get work done, because every day you need privacy and quiet for most of the day, and the default setting of the environment disallows it.

I would agree with your comment if the work divide was more 50/50 between work that needs dynamic communication and work that needs privacy.

But that's just not how reality actually functions.

(A separate part of this which irks me is the assumption that employees can (or want to) 'just work from home' to solve it. It externalizes the costs of privacy onto workers with all kinds of trade-offs not in the employees favor. And a lot of companies actually micromanage this option and are bitterly strict about limiting work from home time.)

Your comment is not wrong but it's all problems and no solutions. If I was designing an engineering office I would not go open plan (or the open plan area would be small). But today we have a lot of open plan offices that inhibit productivity. I identified a way that a company can quickly and inexpensively improve productivity and happiness by creating more opportunities for deep focus. Perfect is the enemy of good.

That is a fair point.

My concern though is that when companies see an example in which someone deflects on addressing a real need for private space, and externalizes the cost onto the employee (via micromanaged work-from-home), it sets a precedent that further entrenches open plan designs for new offices later on, and also discounts the value of investing to rebuild office dividers and spending to change from an open plan back to offices.

I’ve had fully remote jobs before and jobs where I could generally WFH when needed, but neither option provides a good solution.

If your spouse lives with you, they might need to generate noise at home, especially if there are young children. Or you might just live in a cramped city apartment with no space for a desk, or noisy neighbors, etc.

The point is that this pushes the costs onto the employee: “here, you figure this out.” But providing good tools to get the work done is the employer’s responsibility. WFH solutions let them try to absolve that duty, often without actually resulting in private or quiet space for the worker anyway.

A better proposal might be that an employer will pay the cost to fully rent dedicated private offices at a coworking space, and then let the employees who want offices simply work from that company-rented private office space.

This way the employer bears the costs, and doesn’t view it as “right” to just lob the grenade back at the employees by leaning on WFH as the only possible way to work in a quiet setting.

There is middle ground where the team is together, but only the team and it means 4-10 collaborating people. You are not alone and you are not in one room with other teams (especially often calling teams) and not with over 40 people - which generates too much

It should be not one extreme against another.

In my experience, even when managing a team of engineers, it doesn't matter if the people in your nearby open environment are working on the same team or a different team. It's much more about the basic environmental factors (e.g. not having people walking around nearby, having adequate privacy, being able to customize your own lighting, have a personal window to glance out, etc.)

Even when there are only 5-10 people nearby and they are all working on my same team, the open-plan environment is completely untenable and really damages productivity. I would say above all, you need privacy to get into states of flow, and after that, you need to know that you personally control your own schedule, and are in charge of planning when interruptions will possibly affect you.

Whether someone can interrupt you with a relevant question about your team's work, or someone can interrupt you with annoying chatter about weekend golf plans, they both equally prevent flow and productivity.

Environmental noise is much lower with 6 people then with 30. It is huge difference in amount of noise and interruptions. On average, 30 people generate five time as much "weekend golf" discussions and all on-topic debates of other teams are off-topic to you. It is also easier to police interruptions with around 6 people, unless they are jerks - but chance of jerk among 6 is still lower. By policing I mean saying something like: "you are noisy today and I cant focus, take it please elsewhere".

You don't get zero interruptions, but there is significantly less of them and the barrier to communication in team is still small.

Unless the colleges are very noisy, I would say that most people can focus in reasonably large room with 6 people. I guess it is individual, but the need to be completely alone with everything completely as you like it is rather on the more extreme side of spectrum. Most people can make compromise about lighting, are able to share window etc and can still focus. (I am absolutely cool with home office or whatever for those who cant. Just that the average worker should be both able to work in non-perfect privacy and simultaneously not to consistently disrupt those who are focused.)

If the relevant questions happen too often and you are not an analyst or pm or senior responsible for teaching new person (in such case it is your job to answer questions as they arise to speed up process), then it warrants organizational discussion in team and bundling questions into one meeting. After all, if private office would stop them, then they are not that necessary and it should be possible to lower them down by discussion.

The example of colleges seems odd to me, since it’s a common resource at most colleges to get a study carel or private study room at the library, and it’s a place where even in common areas, there is rather strict enforcement of silence, lack of cell phones, and certainly nothing like modern open plan office noise or distraction.

Also, I think it misses the point to talk about whether or not people can make compromises. The question is about how to empower and enable workers to generate their best output.

For example, you could also say something like, “even though people like having two 27-inch monitors, I find they can make compromises and just use one 17-inch model.”

It’s myopic, because the question has nothing to do with whether or not workers can compromise. It’s about whether it’s cost-effective to pay for quality tools (monitors, offices, private workspace).

So when you say something like, “Just that the average worker should be both able to work in non-perfect privacy and simultaneously not to consistently disrupt those who are focused,” it just misses the mark.

The question is, why would a company wrongly think that providing “non-perfect” privacy is somehow good when empirically it’s known to be bad, even for the company’s own bottom line.

Lastly, I think it’s important to totally avoid framing the desire for adequate private space as if the worker wants “perfect” privacy or they are inflexible and uncompromising. This is a false and worker-unfriendly way to look at it.

Having a private office is not “perfect” or overly demanding to request. Rather an office is just a simple, cost-effective tool. Workers who use offices are still good at compromising to have good communication and collaboration, for example with open door policies, scheduled meetings, and all sorts of non-audio collaboration, in addition to getting to use their extra productive time for dedicated focus on things like code review, to directly collaborate in ways that help the whole team.

Open plan offices on the other hand represent zero willingness to compromise on the part of the employer. The employer is saying they will not invest in good tools for you, and instead dogmatically insist there is only one type of communication (real-time, constantly preemptible audio stream) that is permitted, and anything else has to require contortions and inconvenience on the part of the worker (e.g. working from home, using headphones when it is uncomfortable or distracting to do so, etc.)

So really, we have to move past the anti-employee attitude that a private office is some type of primadonna special request from someone who doesn’t compromise. That is what greedy employers would want us to falsely think.

In reality, an office is just a simple, cost-effective tool literallyno different than ergonomic desk chairs, monitors, or the company’s commuter benefits.

I can't imagine any non-senior employee (for instance) asking for a dedicated private office and not being laughed at, of course exempting any employees who require a private office to perform the core function of their job (not sure if any of these exist, though).

I can't imagine how you believe that private offices are not a luxury. Depending on where the company is operating out of, the rent for the floor space of their office could cost more than the company pays the employee.

This comment seems like pure provocation or something. What does it matter what you can imagine?

Stack Overflow, for instance, famously gives private offices to all developers, even junior ones, and even in Manhattan.

I worked previously at a defense research lab in the eastern US, and had my own office straight out of undergrad (it didn’t even occur to me to ask).

I have former colleagues or friends working in private offices in: computer graphics form film, defense research, hospital research, quantitative investing, adtech, large ecommerce retailer, and education tech.

Private offices are utterly not a luxury. They are a basic tool. One simple reason is because they are cost efficient, so you don’t spend more on offices in any sense but the most narrow-minded. Other reasons include all of the decades of research on their basic ergonomic benefits.

This would be like calling an ergonomic chair or a trackball mouse “a luxury” because they superficially appear to cost (slightly) more than basic alternatives, without accounting for the cost-savings they cause. And, like offices, it’s a trivial extra cost for the company.

> “Depending on where the company is operating out of, the rent for the floor space of their office could cost more than the company pays the employee.”

Given that this is not true in Manhattan (e.g. see Stack Overflow’s big write-ups on it), can you provide data to show where this is true?

You're absolutely right about the floor space, I had not considered 3x4m glass boxes, and I apologise that my post came across that way (pure provocation) -- perhaps I was in a sour mood. I have not worked for a significantly wealthy company in my history, and so it has never been the case that a private office has been the most cost-effective way to improve profits. To make a reasonable argument: Typically the equation is simply money-in vs money-out. If you can spend an extra 100k a year on private offices for developers and get 300k a year in chargeable services/development effort, vs spending 100k on another salesman and get an extra $1m in net profit, which would you choose? The choice here is clear and 2 monitors vs 1 monitor is a $100 cost for 1.5x efficiency, and is an absolute no-brainer. An office on the other hand is usually not the most cost-effective option, but if you have enough cash to fund all of the other more cost-effective strategies plus private offices, then it is another place in which management can increase profits.

Please follow up with your thoughts, and again, I apologise if what I wrote was inflammatory or came across that way. It was likely coming from a position of "unknown unknowns".

The Cal Newport book on Deep Work that this article is based on has inspired a push for the Eudaimonia Machine [0] (an updated version by Newport himself here [1]). It’s a concept for an office that is designed to promote effective deep and shallow work. It’s a wonderful idea, though discussions on HN about it have pointed out that it’s difficult in startups, especially during high growth: allocating space for deep work (sound proofed rooms for individuals) takes up quite a bit of space that would be otherwise needed for a fast growing employee base.

[0] https://medium.com/@jsmathison/i-cant-stop-dreaming-of-eudai...

[1] http://calnewport.com/blog/2016/10/19/the-opposite-of-the-op...

> space that would be otherwise needed for a fast growing employee base

I'm not sure what "otherwise" is supposed to mean here. Isn't the whole point of the "space" in the first place to enable the "employee base" to be more productive there than at home or remote?

Maybe that employee base wouldn't need to grow so fast if the existing employees could be more productive by, say, using existing space more effectively.

Lastly, it seems to be something of a cliche how hard and/or expensive it is for startups (especially, as you mention, when growing) to deal with space. I am, however, skeptical that the decision of having open plan offices is based on any kind of rigorous decision making rather than trend-following and following the path of least (initial) resistance.

Thank you for sharing. Some of these design ideas remind me a lot of the discussions in The Timeless Way of Building, particularly the hub-and-spoke idea compared with Christopher Alexander's passages about the importance of windows, with examples from Scandinavian designs that prioritize surface area of new buildings to give more people dedicated, personal windows (at least at the time of his writing).

The reason is because management is almost never driven by science or productivity maximization but maximization of either narrow subdivided metrics (like costs for a particular function) or absolute cargo cult conventional wisdom. Even when there is an empirical, data driven culture somewhere in an organization, it rarely extends to the board, often doesn't extend to the executive suite, and almost never extends into, say, facilitues management.

I'm not sure I agree with this because if it were thus, I think 'most organizations' would be out of business quickly.

I think the answer is more plain: shared offices cut cost, and and the resulting immediate an ongoing increase in profit is 'the most important of all benchmarks' for most companies, in fact it's not even a 'benchmark' ... it's 'the point'.

So - essentially, 'private offices' are an 'investment' and an investment has to be clearly and obviously justified, otherwise the default is to go to open office plans.

The vast majority of work simply is not deep work - and yes, even things like 'Marketing' - although most operational marketing people have a 'creative bone' - and brilliant marketers are as useful and brilliant as equivalent Engineer, the actual creative/deep thinking Marketing stuff is usually done in agencies, which are like architecture firms. Similarly I don't doubt creative financial structuring or deal making, 'deep time' is just not as consistent for them as it is for most Engineers.

There is actually rather scarce 'deep work' to be done, and so, as so few companies willingly require people to be 'learning very important new skills' ... offices become less common due to the simple economics of it all.

“Deep work” is the vast majority of necessary work for software engineers and knowledge workers in most companies.

In fact, these jobs are almost defined by the characteristic that almost all of the work is deep work, requiring privacy, and states of flow.

I keep seeing this unsubstantiated claim that deep work is rare, or at least a 50/50 split with shallow work.

But it’s flat untrue across the board.

"software engineers and knowledge workers " make up a tiny portion of workers in the economy. Tiny. Is my point.

Also, having written software for quite some time, in my experience would be that yes, at the end of the day, 50% of an Eng time is meetings, standups, blackboard discussions, chatting with others, having breaks/wasting time, eating, going to the bathroom. etc.. I think it approaches 50/50. The other issue being that one simply cannot be 'in the flow' for much longer than 4 hours anyhow, pro musicians don't practice 10 hours a day, more like 3. Though for myself, I can see much greater than 4.

Why would the relative fraction of jobs taken up by software matter? Most jobs aren’t police officer jobs, but we still give police officers handcuffs and police cars, because (for that subset of jobs) those tools are needed, just like private workspace is needed for the subset of jobs in software.

> in my experience would be that yes, at the end of the day, 50% of an Eng time is meetings, standups, blackboard discussions, chatting with others, having breaks/wasting time, eating, going to the bathroom. etc..

Other than when a company mandates a poor meeting-oriented policy, like mandating Agile meetings even when it’s counterproductive for everyone, I’ve never heard of anything remotely like what you describe.

Mandating an inappropriate Agile-style meeting schedule is common, but that’s exactly the sort of thing I am saying the open plan office engenders. The fact that unproductive meetings make up 50% of work time in these companies is in no way any kind of endorsement that software engineering (or even other) workers actually benefit (in terms of productivity).

In other words, the engineering tasks demand 80-90% deep work focus, but engineers are forced, like a square peg in a round hole, to contort unproductively to wedge that somehow into a meeting-centric company policy that hurts both the engineers and the people who mistakenly organize & believe in the udefulness of the meetings.

I always just figured it's the "oil change" problem. The day after the oil change interval is up that you skipped, all the trucks are still running great and it feels like you just saved a bunch of money...

It can be very difficult for a bean counter with a spreadsheet to connect the extra broken down trucks at the end of the year to the oil changes that were skipped in February.

Every open office I've ever been in was to put more people in less space for less money. The subsequent burn-outs and defections were, of course, due to "the market" (and once, even uglier, "the poor millennial work ethic").

I don’t buy the argument that bean counters are unable to account for vaguer aspects of productivity loss. For one, bigger corporations invest huge amounts of money into productivity research and take the topic very seriously. They aren’t unsophisticated or naive and corporations will really study the internal incentives they create for cost cutting, to make sure they aren’t mistakenly cutting costs that actually lead to other, unanticipated costs. Sure, bureaucracy makes this imperfect, and some companies are worse than others, but it wouldn’t explain 80% or more of all companies shifting to known productivity-destroying office spaces.

Plus, it’s widely and popularly disputed that open plans are actually cheaper, e.g. even in pop articles like this:

< https://business.linkedin.com/talent-solutions/blog/hr/2016/... >

I’ve also seen sources claiming that an open plan developer is less than 9% cheaper in real estate terms than a private office developer, without any consideration about interruptions, distractions, communicable disease, lower morale, changed vacation or sick time habits, etc.

9% might sound like a lot of savings on a naive glance, but when these employees are costing you $150K of wages and a bunch more in overhead & benefits, you would be hugely sensitive to big, publicly discussed sources of productivity reduction, because it’s immediately obvious, especially to corporate finance, that saving 9% on floor space means nothing if you’re losing 15% of labor ouput from distractions, 5% from increased sickness or vacation habits, 2% from widely reported morale issues, etc.

I think really the issues are much more status driven, seeking to force flow-requiring knowledge work into a commodity, to push down the age and wages of employees.

And part is cosmetic. If you think of a start-up itself as a product, and the investors as the user, and the VCs as the company, then VCs optimize cosmetic attractiveness to create start-ups they can sell for as little substance as possible.

So if grand visions of hoodie-clad developers sipping craft coffee in an exposed brick, glass, ductwork open plan hell will convince the dimwitted actor or retired athlete to invest on visions of the next unicorn, then why would they care about worker productivity?

The whole point is credential and cosmetics to lever up the known-to-be-unrealistic valuation to foster an inflated IPO or acquisition, and it’s not a function of actual economic product of employees.

Then after some unicorns hammer this strategy, everyone else (including big corps trying yo look hip) just start cargo-cult copying it.

My only remaining surprise is at how long this open plan bubble can persist before people realize these companies are not about any type of productivity.

> Plus, it’s widely and popularly disputed that open plans are actually cheaper, e.g. even in pop articles like this:

It is now, but when did that start?

I vaguely recall when it was still actually controversial (perhaps due to lack of data), but that may have been as long as 10 years ago. I also recall when open plan wasn't even a thing and the discussion was merely cubicles vs. hard walls.

> My only remaining surprise is at how long this open plan bubble can persist before people realize these companies are not about any type of productivity.

If I may be flippant for a moment, I'm shocked.. shocked!.. That there is wastefulness going on this VC-fueled establishment.

I'm certainly less surprised, as I believe productivity (and its less sexy cousins efficiency and profit) have long been disregarded as being far less important than growth, ever since the dot-com boom, in the world of (many, if not most) VC-driven startups.

What startles me is just how much VC money is flowing directly up to Amazon via AWS (sometimes actually encouraged by the VCs themselves), as that smells more like inflation of a bubble than merely paying programmers to be distracted.

Your reply made me chuckle and you are right.

For me personally, I derive a lot of self-identity from my own productivity, and end up caring about my work and treating it like a craft.

I wish I could turn this off and just accept better paying jobs that embed me in a distraction-filled room where nobody really cares about my productivity so long as I superficially appear to try hard and evangelize the company line.

But I just can't. I am too intrinsically quixotic. For me, and the I guess dying breed of people like me, the side effect of the VC behavior you mention is like death by a thousand cuts.

I'm not sure we're so much a dying breed as that we were always a small minority to begin with.

We just have to hold out hope for the even rarer manager that recognizes such objective value and can transform it into subjective value to sell up the chain.

Not all open spaces are created equal. If a bunch of the square footage is allocated to non-work purposes (entertainment, artwork, airy high ceilings, food, staff services, etc) then reducing cost is NOT the primary goal. But in most IT shops, where cost cutting by management through reducing IT's physical footprint is seen as a major win to trimming the budget (wasted) on IT, the savings isn't a mere 9% but more like 75%.

For example, I work at Merck & Co, where management has decided not only to move all of IT into much smaller open spaces, but at my site, they've decided that only 200 chairs are now needed to house the 600 staff. Thus the intended savings due to downspacing is FAR greater than 9%.

Since I can't believe Merck's management is unique in its enthusiasm for human bin packing, a small number like 9% likely applies only to startups in high rent districts where 'open space' truly is a design concept that implies more is sought than just physical space minimization.

But the producivity loss would be correspondingly gigantic in that case. For a big conglomerate like that, it seems more likely it is an IBM-style ageism strategy, to make the conditions so undignified and fundamentally unworkable for experienced adults that over time they quit and are replaced by young workers. The company is placing a wager on the idea that it does not need actual productivity, and benefits from warm bodies doing menial tasks with limited productivity.

It probably should be viewed as a direct statement that the company doesn’t value engineering talent, and experienced people should plan to leave unless they’re OK with this.

This seems more likely than the idea the company believes it’s somehow saving 75% on space with some open plan hot-desking approach without losing engineering productivity.

Maybe all that research you write about does not translate down to workers. Because the corporation around here don't seem to implement any of it and decisions don't seem to be much influenced by productivity considerations.

They are not result of complete stupidity either, but they seem to be result of price, how "representative" space is and similar down to earth considerations.

The research is only focused on the end result for workers. It does translate. Corporate managers just have different (often irrational) incentive schemes related to cosmetics, cargo-cult copying, value signalling, etc. And in some cases outright don't care about worker productivity (e.g. if they get paid just because a credentialed worker is associated with their firm, and not based on that worker's output.)

Reminds me of business ideas that are not instantly obvious vs a sexy "The Team!" page.

I'm split on this personally. I don't like the type of noisy open plan office where you work next to the ping-pong table while someone rides a bike through the office or something like that, but on the other hand, I don't feel like being in total isolation helps to focus either.

I actually like a little bit of background activity, I work better when I listen to music (mostly classical or anything without vocals) or a not too noisy but not totally silent location like a library or a small open plan office with people spaced out

There have been studies going back as far as Peopleware and The Timeless Way of Building that suggest people are generally adversely affected when the noise is more than noise they personally generate (e.g. keyboard clicking), and not in their control (e.g. choosing to listen to music may be fine, but when you’re forced to listen to music because the alternative is a distracting conversation between teammates it has adverse effects).

The other thing, especially mentioned in The Timeless Way of Building, is that you need privacy to reach a state of flow. It doesn’t necessarily mean you’re isolated or that no one can interrupt you, but it means nothing in your immediate vicinity is triggering your peripheral vision or any sense of a fight vs flight reflex because someone’s walking around right behind you, etc. Most of that research advocates that individuals should be allowed to customize these aspects of their personal space, like privacy, quiet, lighting, ergonomics..

More generally though, it’s not about what works for you or for me. It’s about what we can empirically learn about states of flow, and on that front it’s utterly beyond dispute that open-plan offices are disastrously bad.

Christopher Alexander's work isn't based on empirical studies, though, is it? Never seen any. It's just his, uh, inspired opinion.

In large part yes, you are right, his writing was more qualitative and sociological, sometimes philosophy.

But seeing your comment made me start trying to dig up a 1987 paper from Alexander titled, "Toward a Personal Workplace," and I am actually amazed and unsettled by how hard it is to find anything online. That article surveys more empirical properties of office interiors. But yes, I would look to the studies cited in the Peopleware chapter called, "Bring Back the Door" for the empirical extensions of what Alexander had inspired.

Here's the best link I could find on that thread of Alexander's work:

< http://zeta.math.utsa.edu/~yxk833/Chris.furniture.html >

You aren't kidding about it being hard to find anything about that paper. I tried finding it a couple of years ago and just spent an hour now and had no luck. Thanks for the link I hadn't seen that before...

I see an office one time with oddly long hallways 3 times the height and width one is used to. It had a marble floor that echoed loudly and ended in a Kremlin sized door with a single employee behind it. When I asked why it was that way they explained it was to force the visitor/coworker to repeatedly reconsider if his thing was really worth disruping that person with. To the question if that realy worked the answer was a description of the previous office where coworkers would fly in and out out of convenience. They would stick their head around the corner for 5 word sentences. Apparently productivity went though the roof in the new building.

The tech industry is not rational. The tech industry right now is a gold rush and you are the shovel.

The only thing people high up in the org-chart know is that if they dig hard enough they might find gold.

Then, they don't even have to find the enough gold, they only need to make it seem that they will find gold and use that hype to raise money, be acquired or go public.

And even if in the end 0 gold is found, a bunch of people will walk out with money and zero liability.

The primary reason is middle an upper management's holy grail quest to commoditize knowledge work and turn software development into an automotive assembly line.

My personal guess is that it's somehow related to fundamentals differences between introvert and extroverted personalities and that the people making theses decisions (floor plans) are mainly extroverted. Just my opinion.

I agree this is part of it. For example, as a high-achieving and socially outgoing introvert who is very sensitive to background noise, I found this paper really depressing:

< https://static1.squarespace.com/static/55dcde36e4b0df55a96ab... >

I definitely learn things from people I don't expect to teach me anything. I'm used to pair programming being frustrating, but if you can put egos aside you pick up on different ways of thinking and way more exposure to life/programming experiences others have had. I think the shared office space is forced information exchange between people.

It's not comfortable, it's not necessarily conducive to YOUR methods - but you put enough interesting people in the same space and you're bound to jumpstart something. Maybe an HR incident, but something.

I actually am prevented from learning when I am forced to overhear exchanges between other teammates. Most of the exchanges are not relevant for me, often they are not discussing anything work-related as well.

It also swiss-cheese's my schedule, because instead of setting aside dedicated time for pair programming and for information exchanges or meetings, like you would think you should do in a professional workplace, those events are now thrust upon me even when I might be trying to finish some other work or I might have to urgently focus on something else.

The entire premise that continuous real-time audio is somehow synonymous with idea-sharing or collaboration is just insane.

Sincere collaboration is respectful of each individual's time and planned work schedule, as well as individual preferences for quiet, privacy, etc. Sincere collaboration can be asynchronous, textual, visual, or auditory (or a mix), can be remote-friendly, etc.

What you describe as "forced information exchange between people" is appropriate maybe for a trading floor or mission control, but the attempt to stylize a bullpen of software engineers in the same way is disaster, and at best it is superficial collaboration.

I definitely agree that it's a balancing act between "find the right person to pair with this other person" and "I can't get anything done, this person is only giving me irrelevant information". Often times it's just forced, and arguably not successful.

I meant that pair programming was often beneficial for me, and I work well with most partners.

It just has to be planned, put in a calendar item I can plan ahead for. If it’s constant pairing or if it’s ad hoc pairing that is thrown, unplanned, into my schedule, then it’s always a bad idea and makes both people worse off.

You also have to ensure the noise generated by pair programming doesn’t affect other workers who happen to sit nearby. They don’t benefit from my pair programming, and shouldn’t be forced to reduce their own productivity by putting on headphones, working from home, trying to find a conference room, etc. just because of my pairing session.

I found loss of autonomy and inability to use own brain completely killing my motivation. It might be fine with someone with high sociAl skills, but pair programming with someone who confuses own subjective preference with "objectively better" and have sever micromanagement tendencies is hell. And don't tell me that we should fire all bordeline autistic coders, I think we should cooperate in a way that does not make it hell to them nor to neurotypicals who work with them.

I don't think we should fire borderline autistic coders. I agree with you that it's not always the best approach to take to force coders to interact in a shared space. I used to enjoy something I had with a friend where we would watch git diffs and write back and forth to each other about the changes.

Personally, I think it's because people don't really believe the research.

I mean, sure, most people aren't aware of the research at all and just follow the herd, and there's a bunch of incentives problems here with people caring more about specific metrics like time saved, etc. And sure, you can't really say that most people are rational.

But more importantly, I think this is one of those ideas that just doesn't really pass the sniff test. People think back to open plan offices, for various reasons believe that it actually worked well (e.g. most major companies started that way, wall street works that way, etc). And therefore, they treat the research like a curiosity - "oh you have some research? that's nice. But it's probably bullshit since I've seen open plan offices work great". And considering the replication crisis happening, it's not a crazy position, either.

This may be part of it, but it’s hard to believe there isn’t a more insensitive / tone-deaf dismissal of the research in favor of different, superficial preferences.

It’s such a widely and publicly disputed topic that it’s not a question of dismissing “mere research”. Rather it’s more like, “can I get away with not giving this important tool to workers?” and some degree of confirmation bias to avoid looking into it at all, like you say.