Old Wine in New Bottles

Today controversy erupted over the Homeland Security Department's new "virtual border" system that is supposed to use biometric technology to identify visitors to the United States and link disparate databases at INS and other agencies into a seamless whole. Calling the US-VISIT program "a striking failure," Rep. Jim Turner lambasted DHS for naming Accenture as the prime contractor for a project that could be worth as much as $10 billion in coming years. "It appears that in their rush to get something out there quickly, they've gone down a path that's basically a repeat of our old technology systems."

What this illustrates is a problem inherent in all technology-driven applications. If all that is done in re-architecting a system is to develop new interfaces for legacy databases, then the information-retreival problems of the old systems never go away. As Forrrester Research reports, "When discovered midproject, data defects and data structure anomalies in legacy databases lead to time and cost overruns." In technical terms, a "mismatch between instance information and schema information." Duh! It's not much unlike what happened to Microsoft when it tried to build Windows 95 on top of old DOS code. Not until Windows NT and later Windows XP, which started fresh, did things really get straightened out.

So the lesson for government, once again, is to learn from the private sector. Except this time, don't try to emulate them. Learn from their mistakes. Ten billion dollars is a lot to spend when you're just putting old wine into new bottles.

 Posted by glenn