Oct. 13, 2004

Software will stay put

Wynn Quon

Forget the alarmist talk about software development moving offshore. The pessimists are wrong because they've missed two things: the nature of software work and the coming boom in software demand.

They are peddling nonsense about the future of the North American software industry these days. It isn't pretty. The North American software industry is in trouble. Indian software developers making US$20,000 a year are poised to drop-kick their overpaid North American rivals out of the stadium. It's the hollowing out of the service sector. It's globalization.

It's also flapdoodle.

The hysteria over offshoring is wafting up from a loud crowd of economists, labor leaders, and politicians, none of whom have any expertise in software.

Take economist Paul Craig Roberts for example. In a recent Businessweek editorial he portrays offshoring as pure labor arbitrage. The Next Big Thing in North America's technology industry will be a loud, gurgling sound as the service sector sinks, torpedoed by cheap foreign labor.

The hysteria over software offshoring reminds me of the last time unqualified pundits jawed their way into a technological debate: We ended up with the paranoid episode known as Y2K. Forget the doomsayers. The good news is this: if you're a talented software developer, your job prospects here in North America are bright, if not scorching hot. The news needs to get out because the fearmongering has hit hard. Enrolment in computer science studies in North American universities over the past year has dropped by twenty to thirty percent at a time when we can least afford it.

Where do pessimists like Roberts go wrong? First, they misunderstand the fundamental nature of software development. Secondly they've missed the big picture, the fact that global demand for software will skyrocket in the next few decades.

Software development is one of the most, if not the most difficult intellectual activity that people do. (We're talking here of industrial and commercial software systems, those that control a vast range of functions and capabilities rather than the software programs you might write at home that perform simple tasks.) Cheap overseas labor would indeed be a threat if writing software was like cobbling shoes. But it isn't. It's more like making a movie or writing a book. (Why is it that Hollywood, one of the most expensive places in the world to live, still dominates the global film industry, even though yes, an Indian director makes a tiny fraction of his American counterpart?).

Here are the key facts that are sorely missing from the debate:

· In any group of software designers, the best individual will out-produce the worst by a factor of 10-to-1. (This amazing statistic has been verified empirically with repeated consistency. See Frederick Brooks' seminal software management book, The Mythical Man-Month for this and other insights).

· The same productivity difference of 10-to-1 exists among software teams which means that all-star software teams need skilled designers as well as skilled managers.

You can see how the would-be offshorer who thinks he's getting a cut-rate deal on software work could end up with a disaster instead. The 10-to-1 productivity gap among software teams dwarfs the salary gap between North American software workers and their overseas counterparts. Offshore your software project to an unknown software team and you're jumping into the void. If the finished software is bug-ridden, maintenance and bug-fixing can easily multiply costs ten-fold. On the other hand, consider a $50,000 North American programmer. She may look expensive but if her code works the first time and has zero defects, you're way ahead. I suspect that in an unhealthy number of cases, especially in tech startups, CEOs are using offshore development as an excuse to run away from the challenge of managing software projects. They're seduced by foreign contract companies who offer them software nirvana at cut-rate prices. These bean-counting executives are like rubes going for the lowest price in a used-car lot. I can't feel sorry for these folks. If you can't set a good example, you can serve as a terrible warning.

Indian outsourcing companies are increasingly sensitive to software quality issues. The latest marketing tack is to obtain something called CMM Level 5 certification. Unfortunately hiring a company to do your software solely because they are certified is like hiring a chef just because he tells you he's recently got a new set of knives.

This is not to say that offshoring is always bad. Offshoring is effective when you either send out software work which is low risk and low-complexity or you create your own software organization on the ground overseas. Outsourcing low-complexity software work (e.g. translating legacy COBOL programs to C for example) means you're more likely to avoid the extreme 10-to-1 performance hazard. Creating your own software organization overseas accomplishes the same thing. If you set up your best management practices, you could end up with a top-notch Indian team at lower cost. But the kicker of course, is that only large firms have the resources to do this. Not only that, the clock is ticking. Indian software designer salaries are rising at double-digit rates, the window of savings is rapidly closing. Software talent is so rare that it's more likely that Indian software salaries will rise to North American levels than vice versa.

The second mistake the pessimists make is in missing the big picture. We are on the brink of an explosion in demand for new software. Despite the disillusion of the dot-com and telecom bust, it shouldn't be forgotten that new technologies often start their most robust growth in sober aftermath rather than in the midst of speculative mania. In the crazy '90s, glitzy money-losing websites came and went. Now the focus is much more down-to-earth. It's all about enabling the back office with web technologies, squeezing out costs and speeding up transactions. It's the Webification of the economy in every sector, whether it's government (the U.S. federal goverrnment alone will spend US$60 billion on information technology next year), health-care, just-in-time custom manufacturing or mass-market retail. It's everything from the newly emerging fields of bioinformatics and cyber-security to the booming gaming and CGI industry: It's hard work of real substance, there's a lot of it and it's all software.

There is yet one other bright spot on the horizon, ironically in the very regions that are causing today's offshoring anxiety: India and China. With GNP growth of four percent and eight percent respectively, these emerging giants are going through an unprecedented transformation. They are morphing from agrarian/industrial societies into information societies. Think about the software needs of two billion people, the software that has to be written to fit their home-grown cultural, financial and social requirements. Think of the software budgets of a million Asian businesses as they bootstrap themselves into the 21st century. That Golden Horde of Indian and Chinese programmers that alarmists love to cite as the decimator of the North American software industry? There will barely be enough of them to keep up with local demand.

Or picture the worst-case. Thousands of top software designers decide to work for free. Not only that they're going to give away their code too. Isn't this the ultimate doomsday scenario? Well "the worst" has happened already. It's called open-source development. Linux is probably open-source's best-known offspring but go to SourceForge.net and you'll see a list of over 80,000 ongoing open-source projects. Has all this zero-cost development hurt the wider software industry? No way. Just one example: IBM's going to sell US$2 billion dollars worth of Linux servers this year. Such is the global appetite for software, that even when it is produced for free, the economic benefits are bountiful.

In the end we've got two choices. We let our talented software people rise up to the challenge (and get rich in the process), or we can scare ourselves silly with peddled nonsense. Pick one.

Wynn Quon is chief investment analyst and former code-monkey at Legado Associates - www.legadoassociates.com (e-mail: wynn_quon@hotmail.com)