In response to my last column, in which I argued that government (and big business) IT should run more like a startup, a colleague wrote to object: "Given the statistics on startups," he said, "that means going out of business."

Not quite, for a number of reasons.

A large, mature IT organization is no more "one small startup" than a mutual fund is one small stock. It's absurd to think that any large IT organization could reinvent itself as one startup all at once.

What I'm suggesting is that we run government and enterprise IT activities in the same way that lean startups are run. That means engaging in small experiments, measuring the results, learning and iterating based on the metrics and learnings. Even if you wanted to, you couldn't successfully transform most government IT shops all at once because of the sheer scale involved. And "all at once" isn't exactly startup thinking.

Think of a large IT organization as a portfolio of activities. Focusing on one particular "startup activity" failure is counterproductive. It's sort of like saying: Some people fail when they attempt 50-kilometer mountain trail runs, so that's why I don't exercise at all."

Of course startups fail, and some of your organization's IT and business process innovations will fail. But that doesn't mean you shouldn't engage in them. Experiment with some of those activities; continue the regular bureaucracy with the other activities. But dare greatly, and seek great rewards.

If all your organization wants out of IT is operations, it would have outsourced it. My IT organization tracks and reports on hundreds of thousands of dollars of benefit to our city each year that's a result of innovation, not operations. Innovation can and must be part of in-house IT.

I'm not sure that the number of startup failures exceeds the number of bad IT shops out there. I'll tell you this: I hear a lot more about bad IT than I do about failed startups, and I'm pretty active in both circles.

Mature, Fortune 100 companies are learning from the startup community. At this year's Lean Startup conference, Beth Comstock, a senior VP at General Electric, talked about bringing Lean Startup to life at GE, a company with a $230 billion market cap. GE isn't too worried about these practices bringing the company to its knees, so I suggest that you don't worry too much about these practices screwing up your IT organization.

Small Failures, Big Benefits

Why should we in IT care about startup best practices? Because we have a key role in process and business innovation at our organizations. Let's face it: The way that established companies do things isn't all that super duper. We need to challenge our long-held aversion to experimentation and allow for incremental failures.

The challenge for every organization is to take small, discrete risks rather than bet the farm on the next innovation. These risks, properly managed and taken in aggregate, can yield big benefits.

Global CIOs: A Site Just For You Visit InformationWeek's Global CIO -- our online community and information resource for CIOs operating in the global economy.

Back to the idea that startups "mean going out of business." An economic developer I know, Chris Miller of Illuminomics, kept track of the actual wages created per dollar spent in attracting Daimler-Chrysler to build a factory in the Savannah, Ga., area, versus a program that did the same for startups. Miller found that Daimler-Chrysler created 50 cents in wages for every dollar the government spent on activities surrounding the company recruitment; the company pulled out of the deal after two years of negotiations. In comparison, he found that the startups created almost $12 in wages for every government dollar spent. Another comparison: Area taxpayers laid out $106,666 per job Daimler-Chrysler created, while paying $4,912 per job startups created.

So if we're slandering the startup way, let's include big companies in on the fun. According to a study by the Kauffman Foundation, startups create most net new jobs in the U.S., while mature companies are destroying them through layoffs and offshore relocation.

I'll take a portfolio of "startup" operations (the equivalent of a portfolio of experiments in an IT organization) over traditional big organization practices any day. Yes, some of these activities will fail, but it beats wallowing in mediocrity.

What is technology except innovation? What are startups except the leading edge of innovation? Instead of holding on to what you know, open your eyes and help yourself to a portion of amazing.