NASA says better programming will be crucial to longer missions
NASA's ability to retain programmers who can validate software code is paramount to the space agency's ability to deliver successful future missions, NASA Administrator Daniel S. Goldin told Congress last month.
By Tony Lee Orr
GCN Staff
NASA's ability to retain programmers who can validate software code is paramount to the space agency's ability to deliver successful future missions, NASA Administrator Daniel S. Goldin told Congress last month.
As future missions take unmanned spacecraft farther from Earth, the agency's dependency on software will only increase, he said at a hearing of the House Committee on Science.
A mid-1990s independent study of more than 8,000 software projects across all industries showed an 8 percent success rate, Goldin said. And the successful programs still contained five to 10 errors per 1,000 lines of code, he said.
Wrong measure
A line of errant code caused the demise of the Mars Climate Orbiter. NASA investigations determined that the $125 million craft went spinning off into space or crashed into Mars' red surface because a line of code was written in English units rather than metric units [GCN, April 17, Page 73].
Communication failures within NASA worsened the error because no one raised the alarm that the Orbiter had veered off course. Corrections could have been made from Earth that would have prevented the mission's failure, NASA officials previously testified.
The rapidly changing nature of software development, coupled with the changing demographics for aerospace workers, further complicate the problem, Goldin said.
'While our new hires are knowledgeable and enthusiastic software engineers, they lack the experience of our first-generation space systems engineers,' he said. 'Our inability to pass on this space systems experience is exacerbated by the increased loss of our senior people.'
A flurry of dot-coms have snatched up software engineers eager to make more money than the federal government can pay for talented information technology workers, Goldin said.
'As the nation becomes more dependent on large software systems whose behavior is not well understood, we must invest in fundamental software research to address high reliability and error tolerance, productivity increases, reusability of software and higher levels of autonomy,' he said.
But a failure to double-check software code wasn't the only thing that doomed the Mars Climate Orbiter and later the Mars Polar Lander.
The projects also suffered because there was no continuity of personnel on either Mars program, and NASA left most of the design and development of the Orbiter and Lander to contractor Lockheed Martin Corp., said Edward C. Stone, director of NASA's Jet Propulsion Laboratory in Pasadena, Calif.
For future Mars missions, a project manager, mission operations manager and core team will be assigned for the lifecycle of each project, Stone said.
The Jet Propulsion Laboratory has also instituted a Systems Management Office comprised of the lab's most experienced systems engineers, to conduct independent assessments of projects' requirements, costs and risks, he said.
NEXT STORY: GAO: Energy must better guard travelers' data