Like Joel, I bootstrapped a software company to similar size (25 employees). I did a calculation and we spend about 1.4% of our revenue today on 5,000 sqft of office space, but have outgrown it and it's no longer the nice spacious office we started with. 10,000 sqft of space in the best class A space in Western MA will cost about 3% of our 2009 expected revenue, with modest but nice build-out included. Given that this would be our 4th move in 6 years, I tend to be careful about long leases and lease buildouts.

I'd rather invest in paying good salaries, offering the best equipment and a nice (but not extravagant) work environment. Then I can afford to launch new products without being beholden to investors.

I really like the glass walls though...

So really, this seems to be a case of spending money to improve quality of life. The other benefits, such as for recruiting and PR are just a nice side effect.

#) Assuming 50 employees, over 5 years, the costs are roughly $4000 per employee per year. That's maybe 15% of their salary, probably less. Given the choice, I would probably give up 15% of my salary to work in a nice place with good food, a good view, good equipment, and so on.

You think that top software engineering graduates in New York City make $26k per year? Recalibrate! Recalibrate!

I think the number you're looking for is "less than 4%". Remember that the office-construction expenses can be massaged by accountants to reduce taxes on them, and that an employee's (e.g.) $80k salary costs the employer considerably more than $80k, thanks to the employer share of things like Social Security and health insurance. To say nothing of the costs of recruiting a single employee. Note that a headhunter will charge you a cool $30k to recruit an employee with a $100k salary:

http://www.asktheheadhunter.com/basics2.htm

...and who knows what kind of stiff that headhunter is going to try to foist off on you?

The chances that the $4k per employee per year is paying off for Fog Creek are much, much better than you think.

The specific upgrades we paid cash for will probably end up costing 1% of an employee's salary, averaged out over everything, over the lifetime of the lease.

Two points and a necessary footnote:

1) 30% is on the high end. 20%-25% is a normal fee (and in this environment, you can probably bargain your recruiter down to 20%).

2) That cost is what you pay for a recruiter, instead of paying your HR people to do the initial screening. Your recruiter might call up someone knowing that there are ten client companies that would interview someone who graduated in the top 10% of her class at one of the top 10% of schools -- but that only two or three of those interviewers would be a good cultural fit. It's worth it for your recruiter to chase the weird ones (if they're not right for XYZ Co., ABC Ltd. would still love to have them!); the equation is different for in-house HR.

Footnote: I'm a recruiter, who -- sadly -- does not get to work with awesome companies like Fog Creek. Working on it, though.

Oops, mental typo. Thanks for the correction.

Between ludicrous offices and the 37s "4 days a week" plan, I'd probably go with 37s.

Are you suggesting that investors keep startups from getting ludicrous office space? From my experience with well-funded dotcom startups, they all tend to spend staggering amounts on sexy office space. We're lightly funded, so we sure as hell didn't.

In any case, Joel isn't spending a shit ton-- he's spending ~4-6% of revenue, as he said. Decent investors would only involve themselves if you were spending an unreasonable amount of money. Joel could be up to his armpits in Series D financing and not have an investor blink at this expenditure.

(if it isn't obvious, I think it's a dumb expenditure for an unprofitable startup, and a fine expenditure for Joel)

(Although I'm unlikely to "let go" of the outside investor thing; I help run a bootstrapped company, and had a bad VC experience).

Or I was idly thinking once of a plan where you sprint for 3 weeks (still with weekends off) and then everyone has an entire week off.

Over and over, ~13 weeks of vacation a year.

I wonder if a refreshed, happy, and likely less compensated team like that could keep up with a full time one.

The key is to spend capital to get the best programmers to create profit. This idea isn't something Joel made up it's an idea by Phil Greenspun, "Building and keeping a good software engineering team" ~ http://philip.greenspun.com/ancient-history/managing-softwar...

So what's your idea?

The flip side is that spending a lot of money on something apparently unnecessary might incline you to spend more money on things that are actually unnecessary.

(I'm assuming "a good ROI" means 4% per year.)

1) Right-click on top of a picture, keep the button down.

2) Flick your mouse slightly downwards.

3) Release the right mouse button.

And there you go. Now, after checking out a picture, you'd want to close the newly opened tab, and get back to where you were:

1) Right-click, and hold

2) Move the mouse slightly downwards, and slightly to the right.

3) Release the mouse button

And there you go again. It doesn't take long to get comfortable with Opera's Mouse Gestures, and you'll realize they're very handy.

In any case, mouse gestures are a good thing and make your browsing more efficient.

http://picasaweb.google.com/spolsky/FogCreekSNewOffice#52853...

http://news.ycombinator.com/user?id=spolsky

What other companies put that much attention towards work environment? I know that a lot of popular startups tend to favor communal environments, but seeing actual offices is pretty rare. All of the places I've worked at tend to fall somewhere between decent to hellish cube nightmare.

The best adaptation I've seen of the open-office idea is a partial retreat from it. People crave acoustic privacy so they can talk on the telephone, but visual privacy is not as important -- they like being able to see what's going on. This has led to a very satisfactory compromise called "cave and commons". Each office worker has a private office, often small, which opens into a generous open area surrounded by many other private offices. The open area has a kitchen, some couches, sometimes tables for sitting around informally, and sometimes a working library, or at least a rack of current periodicals. You can shut the door of your cave and concentrate, or you can leave your door open and keep an eye and ear on who's coming and going in the commons, and whether the meeting or presentation going on there might be worth leaning in on. The feeling is congenial and homey, and it encourages the casual encounters which, research keeps showing, are at the heart of creativity in offices.

I lived in a dorm designed like this for four years of college. I'm not sure I've ever been as productive since. ;)

I would not be surprised to learn that Spolsky has read Brand's book. He's certainly got the idea. Note the emphasis on glass in the Fog Creek offices, providing acoustic privacy but not visual isolation.

http://smashingtelly.com/2008/08/04/how-buildings-learn-uplo...

I'd like to add the critical factor: team size. Once team size gets too large (say over 8-10 people) random noise and distractions take over. I've seen teams of 30 try to do collocated, collaborative work environments and it just falls apart. At that size, the best I've seen so far is cubes, "study hall" and private rooms. That is, cubes for \half the day, a group "study hall" type atmosphere for the other half, and easily available small group work rooms. By "study hall", I don't mean a bunch of people being quiet, just a bunch of people in the same area working on group/individual tasks.

There's an interesting phenomenon that goes on in collaborative work spaces. Many times programmers get lots of benefits from working together but still would rather be left alone. So many times the team requires time that it works together, say for a couple of hours every morning, like it or not. Not everybody likes it so much. Many times there's a gap between "the way I like working" and "the way that I work that is the most productive"

Just goes to show that programming is not a technical exercise, as much as we like to talk functional code, security, and all the other bits and bytes bullshit. It's a human activity, and full of the irrational and quirky nuances that comes from working with hominids.

I can only work in a space somewhere between an 'office with a closed door' and a sensory deprivation tank.

-- Open work space for 3-6 people, IF everyone is working on the same project, IF everyone is committed to doing good work (and won't be surfing YouTube throughout the day), IF there are quite/private places to retreat to for phone calls and solo problem solving, and IF the configuration doesn't stay the same for years. Many people report being more productive for short periods with this sort of arrangement.

Bonus: pair programming seems to lead to a different sort of "flow" than solo programming, which is more amenable to an open work space.

-- Offices with a door for everything else. The upside of offices with a door: reduced interruptions, which are the bane of complex programming. (Try solving a really really hard problem with the threat of interruptions constantly on your mind. Not actual interruptions - just the threat, or memory, of interruptions.) The downside: teams have to be more deliberate in fostering team communication.

The "cave and commons" (elsewhere in this discussion) seems like a good compromise. Even better, in my opinion, would be a modular space where teams could work together some of the time in an open war room, during some phases of development, and where they could work privately during other phases. But that's not easy to find.

http://picasaweb.google.com/spolsky/FogCreekSNewOffice#52853...

Check out all of the famous series 7 chairs:

http://picasaweb.google.com/spolsky/FogCreekSNewOffice#52853...

It's got that not-quite-a-prison institutional look to it. That's great if you're running a high school in the poorest district in Mississippi, but given his budget [1] you'd think it would look nice.

I've got a private office, and the landlord charged my company nothing to throw up walls and doors in the empty space we rented. Maybe it's impossible to avoid Veblen goods in NYC, though.

[1] http://www.gibraltarfurniture.com/malik-mc-c-2201.html That's right. It's 300 dollars.

Wow; apparently my old highschool should sell the chairs in lunchroom to raise a quick million dollars.

Oh and the same room from another angle looks much less like a prison:

http://picasaweb.google.com/spolsky/FogCreekSNewOffice#52853...

That's just an absurd statement. The design is very high-end, perhaps you just don't like the modern look?

http://www.flickr.com/photos/doctony/2486595834/

http://www.flickr.com/photos/stewiedewie/2377292918/

http://la.eater.com/uploads/2008_02_mcdonalds-thumb.jpg

http://www.flickr.com/photos/meteorry/17414425/

There's a McDonald's down the street from me similarly superior to Joel's lunchroom, which resembles none of the above. It was clearly patterned after a high-school cafeteria.

If he spent half the amount of money he spent on this amazing office for a kick ass UI team, he would increase sales by an enormous amount for Fogbugz. I know so many people who would switch if it wasn't so damn ugly and difficult to use. I don't think people use Lighthouse because it has better features.

Joel's priorities seem very odd to me.

It's not that Fogbugz is necessarily difficult, it's that Lighthouse is easier.

I would describe FogBugz as ugly by modern standards, because it appears to have been written a little under ten years ago. And I would describe it as difficult to use partially for the same reason, and partially because some very simple features are missing, e.g. cc'ing people on tickets and bidirectional integration with SCM repositories. The "special" features, like integration with customer email, are not important to us.

Yet what it does is not very difficult, and many other options exist -- free, for pay, hosted, or any combination thereof. I now wonder why people listen to Joel on software.

Now apart from that I completely agree with joel's reasoning regarding building a great place to work for programmers, I just think that he should also hire more designers and ui experts...

But I don't think you'd want female developers. Female sales staff, female secretaries, etc.

Nice office, but without nice pussy, what's the point?

At Bowdoin College the CS professors have offices with an awesome view and a shower and bathroom in each office.

I love HN and the community and still read a lot of interesting stuff here, but, I am beginning to feel that HN is slowly slipping away to be a site that also has a few genuine and invigorating posts/discussions about entrepreneurship and programming.

so no big deal, but windows are nice

Knowing that Fog Creek is so close is tempting me to send in my resume... :)

I love how you're able to market your company. You just got about 14,000 top notch hackers to look and drool over your office space. It's a great recruiting tool.