President-Elect Obama has promised to appoint a Chief Technology Officer of the United States. I’ve argued that what Obama really needs and wants is first and an execution-oriented Chief Information Officer. The job surely requires great technological vision, but execution skills are paramount. Discussion on Slashdot ensued. While many of the comments were off-topic and/or inane, at least two good questions were raised:
- To what extent should the CTO/CIO actually run information systems that are now the province of individual departments and agencies?
- What does a governmental CTO/CIO do anyway, given that most activities are outsourced one big contract at a time?
First, I think outsourcing will continue to be a way of life. The government – even more than private industry — has problems such as:
- Pay scales that limit the ability to hire workers with in-demand technical skills
- Employment structures that aren’t well-suited to accommodate job-hoppers who stay up to date with on-the-job technical training
Second, the government nonetheless needs in-house project managers to supervise outsourced mega-projects. The better the in-house manager, the better the people the contractor assigns to the job. For that and many other reasons, we need outstanding in-house oversight if there’s to be any hope of effective governmental IT use. I suspect that achieving such excellence will require moving managers around among departments and agencies much more than they have in the past. If so, then make it happen.
Third, I think the US Federal Government should adapt a key trick from the large-enterprise private sector – offer cheap centralized services, that individual departments are permitted but not required to use. Obama has ambitious goals for the use of video, search engines, etc. The requisite technology should be developed and provisioned separately – but if individual departments want to go their own way, and can get the appropriations to support that, so be it. (Example of where such independence makes sense: The best search engine for a specific job is often domain-specific.)
Examples of services that should be centralized just because of routine efficiency arguments include:
- Web-site-serving using more modern technologies (e.g., video is or at least should be coming in in a big way).
- Accounting, payroll, purchasing, and related back-office systems.
- Data warehousing of transactions from back-office systems.
- Personal-computing support.
- Network security (run by intelligence-agency professional paranoids).
- Routine telecommunications.
Exceptions would need to be carved out, however, for instances in which there’s a special national security need for extraordinary security and/or good robustness against combat destruction.
But that’s just the beginning. I’d also like to see a central CTO/CIO overseeing projects of the kind that agencies consistently botch on their own. For example, multiple projects to upgrade air traffic control computers have failed since the 1980s. The cost in time, money, and fuel is enormous; many lives have surely been lost as well. The next try should be under the purview of people who actually know what they’re doing. The electronic health records national project is extraordinarily complex, in terms of technology, a shifting health care business landscape, and a broad variety of privacy concerns. Candidate Obama rightly called for this effort to be led centrally as well. Rationalizing civilian emergency communications is another central priority. And so it goes.
It’s a huge job. But somebody has to do it. In a private company, the chief technology/information officer is one of the 10 or so most important executives. The time has come to admit that the same is true in government.
