So, President Obama is promising to appoint a federal chief technology officer. First question out of the gate: Do we even need one, given that CIOs and CTOs already abound across U.S. government agencies?

The difference here is that this presidential appointee (Obama hasn't named anyone yet) would sit at or near the Cabinet level and thus would be in a position to drive "transformational change, not incremental improvements," says Norm Lorentz, who during the Bush administration served in a more limited capacity as CTO within the Office of Management and Budget.

Obama needs to pick someone who knows his or her way around Washington, but not someone so grounded in "government IT" that fresh thinking becomes impossible. He must lean toward a practitioner, a multinational CIO or someone with similar chops who understands tech architecture, purchasing, deployment, management, systems integration -- and people leadership -- on a massive, visionary scale. Don't bring in a vendor executive who will bring product and technology baggage to the post.

The candidate must be supremely knowledgeable and tenacious, of course, and his or her responsibilities must be clearly defined from the get-go. Denis O'Leary, former CIO of Chase and now a professional investor, nailed it when he said: "Without the power to create change, this is a symbolic rather than sustentative event. There's a lot of roadkill of well-intentioned corporate types getting chewed up with the governance processes of Washington."

The industry leaders we interviewed tended to fall into two camps: those who see the federal CTO as an inside-the-government change agent, and those who want a leader of tech innovation more broadly. In the latter camp is Sybase CEO John Chen, who fears the fed CTO will "jump into the technical stuff, under pressure of getting tangible, short-term results, and miss the focus on frameworks and the innovation environment."

With all due respect to Chen, who lays out some very thoughtful ideas, I get nervous whenever talk turns to frameworks and environments. Yes, the CTO shouldn't just focus on quick fixes or toil in "the bureaucratic weeds," as Intel's Peter Cleveland cautions, but let's also not create a position whose charter is so grandiose or nebulous that it becomes little more than a pulpit.

The fed CTO needs to think big—but not so big that it becomes impossible to get things done. That means focusing on the government's technology and data architectures (cloud computing, data center consolidation, Web applications, data warehousing, etc.) and setting procurement and other standards across agencies. Leave the "innovation" and U.S. competitiveness master planning to the Commerce Department -- or better still, for smart entrepreneurs and captains of industry to figure out with only minimal government assistance or interference.

I like the thinking of EMC CTO Jeff Nick: The fed CTO can play the role in the Cabinet or at a presidential advisory level that a private-sector CTO does for the board of directors, and that's to help agencies understand the technologies at their disposal and bring some consistency to how those technologies are acquired and managed, rather than make all the tactical tech decisions.

Finally, the fed CTO should steer clear of setting industrial policy—picking specific winners (wind turbines, WiMax, etc.) and by exclusion relegating losers. Ultimately, consumers must decide how much more investment is needed in green, broadband, and other areas, and which specific underlying technologies are the future. On this point I'm in violent agreement with Sybase's Chen, who says: "We don't want anyone to dictate any one technology. ... That actually will hurt rather than help."

Rob Preston,
VP and Editor in Chief
rpreston@techweb.com

To find out more about Rob Preston, please visit his page.