Fog Creek Software
Discussion Board




Private rooms vs. "radical collocation"

Recently, after reading a few of Joel's articles, I got into discussion with the PM of my current project team about working conditions.

It appears that my company is deliberately putting people into open space offices to achieve better communication level. Also, they say that some "serious research" conducted by "many rich companies" show that this is the right thing to do. They call it "collocation" or "radical collocation".

Actually I came up against some results of such research, eg. http://intel.si.umich.edu/crew/Technical%20reports/Teasley_Covi_Krishnan_Olson_radical_collocation_12_20_00.pdf (the problem is this comparison was between "war rooms" and cubicles, so it's not quite adequate).

This is quite the opposite from quiet, private rooms Joel and "others" are advocating. I wonder if you have some experience in this matter? Which approach is better?

Maciej Kolodziej
Friday, November 28, 2003

How about this:

If you have excellent employees, who tend to be self-motivated, then it's best to have a private office for each.

If you have mediocre employees, who would rather slack off when left alone, it is best to put them in "war rooms", so they know and feel that others are watching what they are doing.

MX
Friday, November 28, 2003

[If you have excellent employees, who tend to be self-motivated, then it's best to have a private office for each.
If you have mediocre employees, who would rather slack off when left alone, it is best to put them in "war rooms", so they know and feel that others are watching what they are doing. ]

Hmm. Ok, I'll try not to be offended.

The above is a gross generalization, and not a very accurate one at that.

If you're doing XP or Scrumm or another agile process, then private offices are probably not the best choice, regardless of how self-motivated your employees are. If you have enough space to support common work areas and smaller, private, personal spaces for employees, that's probably the best way to go.

Characterizations like that above are not helpful. I could as easily say that developers who need to cloister themselves in order to think or get work done are inferior to those who can work in a group environment.

anon
Friday, November 28, 2003

I saw the article referred to in the OP a few years ago.  As noted, their comparison is between a "warroom" and cubicles.  It is not surprizing that there was some improvement.  There seems to be a lot of agreement that cubicles are about the worst possible workspace for software development.

You can include me among the "others" who advocate private offices for software developers.  I have worked in quite a variety of work spaces: open office spaces, cubicles, quadricles, shared two person offices and private offices with door and window.  The private office is superior by far, if it is done right.

At least, for the type of development I do (embedded systems, engineering support, satellite comms, scientific data processing) the private office is best.  In previous discussions on JoS there was one poster who told us he put his developers in private offices and it was a disaster.  Very low productivity.  But he was doing game development.  This is something I have no experience with.  It may require continuous interaction.

Private offices are necessary, but not sufficient, for good workspace.  Quality requires attention to details.  In addition to providing the space, you have to be concerned about noise sources such as HVAC noise, speaker phones, cell phones, paging, etc., and other problems.

If you read the Santa Teresa report [see reference in the Bionic Office Examples thread following this one], you will see that developers spend a certain fraction of their time working alone, and another, possibly larger fraction, working in groups.  Once you have provided the private offices so developers can get their individual tasks done, you also have to make it easy to interact in small groups.  Arranging offices close to each other, making offices large enough so that 2-3 people can meet, having central large meeting rooms, a convenient kitchen, etc.  so that meetings as needed are easy is also necessary for a good work space.

mackinac
Friday, November 28, 2003

I think that one of the key indicators for success for "radical collocation" is the nature of the conversations that go on at work.

For example, if you're intently concentrating on developing a new feature for the Foo module and you're interrupted from your concentration by a conversation between your co-workers about a nasty bug that has been found in the Foo module, well, that's good.

But if the conversation between your co-workers that interrupted you was about who was booted off the latest popular reality show last night, well, that's bad.

I think that all the annecdotal evidence about whether "one big room" works or not comes down to the nature of interruptions in the particular workplaces. In some places, people are focused on the job and most conversations have some productive value, and in other places most conversations have no work value.

Unfortunately I don't see any way to measure this.

Bill Tomlinson
Friday, November 28, 2003

Executives and managers generally have the highest per-hour cost in the company. Anything that improves productivity should be exercised on them first.

In other words, if management gets private offices, then it sounds to me like "radical collocation" is a study trying to justify the results. :-)

Now, if you can put *just* the team in a workspace, then you might get some communication benefit. But this is going to mean moving people around a lot or having a very structured team setup. It also means your team communication benefit had best outweigh the complete loss of people to concentrate.

Philo

Philo.
Friday, November 28, 2003

Wow - the study says there was a ONE-HUNDRED PERCENT improvement in productivity!

Fred Brooks was wrong, folks -- there IS a silver bullet!

Dina Sudahay
Friday, November 28, 2003

"I could as easily say that developers who need to cloister themselves in order to think or get work done are inferior to those who can work in a group environment."

You could say that but you wouldn't have any studies to support your statement, whereas there are many studies showing productivity improvements when going from a noisy environment to a quiet one that allows designers to focus.

Dennis Atkins
Friday, November 28, 2003

I read the first page of that report.

1. They couldn't quantify the changes that were related to radical collocation and changes in management methods.

2. How much of this was the Hawthorne effect? They said they studied productivity in new radical collocation situations versus metrics they had from the old way. Well how much of that productivity came from the fact that things had chaned and they were being studied?

www.MarkTAW.com
Friday, November 28, 2003

What is best for executives might not be what's best for developers.  SRA International did a study of comparing private offices and cubicles.  An article describing the study and its results was published in the Washington Post Business section for October 4, 1999. This article never mentioned Santa Teresa, but the results sounded very much the same.  At the end of the study they decided to put the chairman of the board in a cubicle and most other employees in private offices.  Managers and executives do spend a lot of time interacting with people, so it does make some sense that they'd be the ones in cubicles.

mackinac
Saturday, November 29, 2003

Just my opinion, but I think that most programmers perform better in a "collective" environment.  I've met a few (very few) superior programmers who would benefit from a private environment. 

But most are not.  Most programmers cloister themselves, often cluelessly writing code without context, that doesn't get unit tested, and properly integrated into the whole project.  I'm happy as a clam, 'cause I'm living in my shell.  Curiously, none of the bugs the testers write up ever repro for me.

Anyhow, my boss would benefit from working in the collective environment.  He's often clueless regarding what we stuggle with and seems to wonder why features take so long and what the risk points are.  Yup, he lives in an office separate from his staff.

horse with no name
Saturday, November 29, 2003

Here is a link to a research report that indicates radical colocation might have adverse effects:  http://www.news.cornell.edu/releases/Jan01/noisy.offices.ssl.html

mackinac
Sunday, November 30, 2003

And here's an article about the hidden costs of putting people in cubicles: http://www.chacocanyon.com/pointlookout/020417.shtml

mackinac
Sunday, November 30, 2003

*  Recent Topics

*  Fog Creek Home