13 June 2005

Star wars (no, not ROTS)


What a shock. The "Star Wars lite" system rushed into deployment in Alaska, despite never having succeeded a real-world test, isn't working.

<digression> Way back in college, one of my programming classes did a segment on ethics (the only class I ever took, aside from philosophy, which talked about ethics), and one of our case studies worked on the question of whether we, as individuals, would be willing to work on the SDI. Many people's reaction was that of course they would — they would work for whoever paid them. Others went the other way, refusing to accept military money, no matter how good the cause. Those of us in between came up with a secondary question, which was this: given that the system was estimated to need some tens of millions of lines of computer code, which could never be tested in real-life conditions (nuclear explosions going off, etc) until it was needed, what was the likelihood that the system would work? Nearly everyone agreed that the probability was vanishingly close to zero. OK, fine, given that, we asked, would you still be willing to work on it? Some still were, thinking that the basic research towards an unattainable goal would still be useful. In the end, though, a clear majority of the class said they would not be willing to waste the taxpayer's money that way. </digression>

At least the Pentagon's panel is correctly identifying the root problem with the Alaska installation:
In any case, the panel makes clear that the U.S. decision to press ahead with the antimissile system in the face of production and testing delays has come at considerable cost in assurances of its reliability. Pentagon officials have blamed a recent string of system flight-test failures on minor technical glitches. But the panel argues that the setbacks reflected a larger preoccupation with deadlines.
This sounds about right to me. When I advise managers on fixing problems with engineering development, I always ask them when the last time they spoke with an engineer, and why. Invariably they say it was because of a late project. If asked if they've ever spoken with an engineer about the quality of his work, they usually say no.

There's a very good reason for this. It's dead easy to tell if a project is late, and very difficult to see if it's of poor quality. But no manager should ever be surprised by a poor-quality engineering product if the manager has never demonstrated their awareness of quality.

So when the president promised an installation by the end of 2004, a poor outcome should have been expected. It's no wonder he was a failure as a businessman.