OUTSOURCING….Dan Drezner, aka Outsourcing-R-Us, links to a New York Times article today suggesting that outsourcing of software development might be harder than it looks:

Another Indian executive in the United States who has soured on outsourcing is Dev Ittycheria, the chief executive of Bladelogic, a designer of network management software with 70 workers, also in Waltham. Bladelogic, whose client list includes General Electric and Sprint, outsourced work to India within months of going into business in 2001. But it concluded that projects it farmed out ? one to install an operating system across a network, another to keep tabs on changes done to the system ? could be done faster and at a lower cost in the United States.

That was true even though programmers in India cost Bladelogic $3,500 a month versus a monthly cost of $10,000 for programmers in the United States. “The cost savings in India were three to one,” Mr. Ittycheria said . “But the difference in productivity was six to one.”

This fits with my own experience. Several years ago, the company I worked for was acquired by a Swiss company that had offices around the world, including one in Kuala Lumpur. This made it very easy to experiment with contracting some work out to Malaysia, but the experiment turned out to be more expensive than we had thought. In the end, my conclusion was that the raw cost needed to be about one-fourth the cost of doing the job in California in order to be worthwhile. The Malaysians only charged about half what it cost to do it at home, and eventually we gave up.

(The problems are the obvious ones. You have to spend a lot more time writing very detailed and clear specs if the work is done overseas. The coding takes longer because communication is more difficult. The overseas programmers ideally should set up the exact same coding environment as headquarters, but they can’t always do it. Management overhead is greater. Etc.)

I’m no longer there, but after I left they tried again, this time with programmers in Vietnam. I had lunch with one of the company’s manager last week, and she told me that it’s worked out better this time because the raw cost is about 20% the cost of U.S. programmers. Even after the usual markups, it still saves the company money.

Of course, this varies from company to company and project to project. IT departments building internal systems are usually able to make more efficient use of overseas programmers than companies building consumer software. And certain types of projects (ones that are more easily broken into modules) lend themselves better to outsourcing than others.

But overall, my guess is that outsourcing isn’t quite the bogeyman in the software industry that it’s made out to be. It definitely exists, and it will certainly grow in the future, but people who try it often find that in the end it doesn’t save nearly as much money as they hoped. My rule of thumb remains the same: the raw cost needs to be in the neighborhood of one-fourth what it would cost to do the same job at home in order for the cost savings to be worthwhile. Caveat emptor.