3rd January 2003


Back in May 2001 I recommended that the USA develop an anti ICBM interceptor system.  At the time I indicated that the most important expensive part of such a system would be the software.  (It is because software is more intricate than hardware that Microsoft is the largest corporation in the world.)

The charge by MIT scientists that an AICBM (anti intercontinental ballistic missile system) interceptor system is like hitting a bullet with a bullet is a gross understatement, but only if human reflexes were aiming the intercepting bullet.  The aiming mechanism of an AICBM interceptor is not a human, it is a computer.  ICBM's typically travel at around 18,000 MPH, which is (talking in microseconds, which is the speed of computers) about 0.3 inches (a bit less than a centimeter) per microsecond.  If we assume that the ICBM is two hundred miles (300 Km) away, then the radar information on the missile location (assuming lightspeed of 186,000 mi/sec) is less than 0.4 inches (about a centimeter) behind where the missile actually is.  There would probably be a large strategic computer at the radar station, and a tactical computer in the interceptor.  The strategic computer would probably have the primary task of determining which are the probable decoys and allocating targets to 50 or 60 AICBM tactical computers.  The tactical processor for the individual AICBM's would probably be in the nose cone, so the data would become more accurate and less subject to jamming as the interception progressed.  The response time of the tactical (interception) computer should not be a problem, my rather outdated CPU does 1000 calculations each microsecond, which is easily fast enough to interpolate from a table.

Hitting an ICBM with an interceptor is not like hitting a bullet with a bullet, it is more like steering a radio controlled car traveling at half an inch per second into a tortoise moving at 0.3 of an inch per second.

The Times online writes that MIT physicist "Dr Postol and fellow critics say the ability of an interceptor missile to distinguish between an incoming warhead and the decoys likely to accompany it is deeply suspect. Any such doubts would cripple the credibility of the system".  It is of course not only possible but likely that initial versions of the interceptor software would not distinguish between the decoys and the warhead, particularly if the decoys are sufficiently well engineered.  There are however two fairly obvious rebuttals.

  1. The system is initially designed to intercept simple missiles, having either nil or a small number of uncomplicated decoys.
  2. The beauty of computers is that the software is upgradeable.  If the text processor in my computer is inadequate to the task of writing an article, then I can go out and purchase a software upgrade.  In the same way, once having the interceptor hardware operational, software upgrades that can detect increasingly complex decoys can easily be installed as they become available.
One hesitates to suggest that Dr. Postol is perhaps ignoring those considerations because, as Mae West (or someone else) once said, "any publicity is good publicity".  (Mae West was one cool, raunchy babe).

It might be kinder to recognize that Dr Postol is only a physicist, and has thus not completed the rigorous engineering training required to assess practical systems.