We have shitty offices, because they are build to suit the real estate market not productivity.
But there are examples which are horrible by design, and not because the property value was important: https://www.washingtonpost.com/news/the-switch/wp/2015/11/30... (first pic)
In other words, unless you're Sebastian Thrun, Chris Lattner, Lars Rasmussen, Cliff Click or Guido van Rossum, your company doesn't care (and rightly so) about your lost productivity because it just simply doesn't matter.
Gentlemen whose names listed above, I am certain, can have any office they want.
It offers no insight in studies how much a single office improves productivity and how that affects overall revenue of the business. I'm convinced it's more then three times of the lease for office space.
And keep in mind that open-plan offices are turning away candidates (see comments in this thread). They are losing talent for optimizing for real estate usage. This needs to be factored in as well.
Clarifying my point: the office is time shared within the company. Different teams would have different days /in/ the office.
1. the author's estimate for non-commoditized software development is ridiculously low (0.1%)
2. even then, those 0.1% are at the risk of not getting private/shared offices. Do you want that for people working on medical devices or other fun critical stuff now embedded in your everyday life? Well, too late. They are working in shitty environment right now, I know that because I'm one of them. Maybe one of the reason they are, is because a "manager" read that kind of blog post which comfort them to be happy to do absurd economies.
3. you can have shared office room with multiple people per room. Even somehow big shared office room, if some trade-off are needed (although around 10 people should be a real max, otherwise above that there is no real difference with a big open-office floor.
4. if you want to do an economic analysis, even if the surface hypothetically needs to triple (which I don't believe for a sec, see above), you need to compare lost revenue per developer (or other interesting figures in case of not yet profitable) -- and think about non-linear factors -- to the cost of extra space renting (mostly linear). "Pay[ing] almost three times less in office rental costs" (which again, I don't believe) is not a business goal.
The lost productivity of a single person might indeed not matter much if the corp is big enough, but since everyone is sitting in the open-plan office, we're actually talking about maybe 10-20% productivity loss across all employees, which even for a big corp is quite massive.
Say about 12m2 (130sqft) per person. In London roughly £50-£100/sqft/annum. £6500-£13000 rent per person. So for e.g. Google in london are pretty central and are probably paying >£10000 per employee in rent assuming they are open plan.
Closed plan
Another article quotes triple the area. Sounds about right. So £19000 to £39000 per person per annum for the closed plan office.
Obviously this is going to be different if you set up in a new campus in a rural location, but tech companies seem to converge on areas with very expensive real estate SF, NY etc.
How do these numbers stack up with the productivity gains you would expect to make?