Is Joel Spolsky suffering from an Edifice Complex?

Thanks to his outstandingly written blog Joel on Software, Joel Spolsky is a popular software development guru. That makes a lot of sense. Spolsky is sometimes further perceived as an expert on the software business, to the point of leading well-attended conferences on the subject. That is much more questionable.

Spolsky's company, Fog Creek Software, was founded in 2000 and -- according to the New York Times -- has 25 employees. This is, I must admit, more growth than has ever been achieved at any company I've ever founded, excepting only one where the business plan revolved around an acquisition "roll-up" strategy. Even so, it's not what I'd look for in a business guru whose main message is "Do what I do," even if those employees are all the insanely great fruits of Spolksy's hiring process. (Unless, of course, the company has an astronomical level of revenue per employee. I see no reason why that would be the case -- but that possibility is one reason for the question mark in the post title.)

In fairness, Spolsky seems to have his hand in a multitude of projects -- company, blog, conferences, StackOverflow.com, etc. -- and perhaps his methods would be more successful if he focused. Still, for a company that presumably has a lot of great, superbly-managed developers, Fog Creek hasn't produced much that the world willingly exchanges a lot of money for.

And now things are apt to get worse. One of the most reliable rules of thumb I learned as a stock analyst from my boss, the legendary Bob Cornell*, is that when a company moves to fancy new headquarters, it's time to sell the stock. This was true of Comserv 25 years ago. It was true of Symbolics. It was true of Oracle when it moved to its present campus. It was emphatically true of Informix when it arranged to expensive new headquarters. And it's also usually true of private companies.

At the risk of oversimplifying and not accounting properly for regional variations in architecture -- most successful technology companies are located in light-colored two-story buildings on the West Coast, or in dark-colored medium-height office buildings in other kinds of locale. My only contribution to this theory was to give the phenomenon a name -- the "Edifice Complex." Spolsky has the Edifice Complex bad. At least, that's what a New York Times article describing Fog Creek's new headquarters -- and crediting Spolsky's hands-on involvement in the design -- seems to suggest. So does the prominence of interior design in the graphics both on Spolsky's famous blog and on Fog Creek's corporate site.

Actually, this emphasis is hardly new. Fog Creek's own About page raves about an earlier version of the interior design:

To recruit the best programmers, we’ve invested in the nicest work environment we can get. In 2003, we moved into a new office that was custom-designed by a top architect to be the ideal programming workspace, with private offices, windows everywhere, a lounge with a big plasma TV, and every feature a programmer could ever want. We even have twenty power outlets at each desk, at desk height, four with UPS power. That’s how fanatical we are about catering to programmers.

The logic behind the Edifice Complex theory goes something like this: There are many useful things a CEO can do for a company. S/he can hire, fire, manage, sell, market, develop, raise money, or just plain cheerlead. But one of the least useful things the CEO can do is design office space and furniture. A move to lush new headquarters, lovingly supervised by top management, is an indicator of both management distraction in general and questionable priorities in general. About the only less useful thing a CEO can do is devote most of his or her creativity to several different outside projects, unless those projects have clear benefits back to the company, as Spolsky's blog presumably does, but his other ventures probably don't.

Bottom line: Take Joel Spolsky's programming and programmer-management advice to heart. And cool working environments are -- well, they're cool, and there should be many more of them. Even so, be very skeptical of anything else Spolsky has to say about the software business.

*Bob Cornell was the #1 analyst following GE, Westinghouse, et al. for the first 10 years Institutional Investor had All-Stars in that industry. Bob trained me, and I was the #1 software analyst. Bob hired and trained Steve Smith, and he was the #1 hardware analyst. Bob hired and trained Jack Grubman, and he was the #1 telecom analyst (and famously dishonest, but not on Bob's watch). Bob even hired and trained Andy Kessler, who confuses his fact and fiction a bit much for my taste, but was a pretty good analyst even so, and very funny to boot. Bob was trained himself by Ben Rosen, who went on to found the newsletter that became Release 1.0 and a venture fund that backed Lotus, Compaq, and many others. When a brief foray into investment banking didn't work out for him, Bob went back to being an analyst again, and was highly ranked for another 20 years. Basically, Bob Cornell is one of the greatest stock analysts ever.