I found your March Editorial "Mutant Chips," in which you told us not to be overly concerned about the Pentium bug and also implied that we wouldn't want to fly with an airline that was programmed using heuristic methods, to be something of a giant leap from reality. According to published reports, between 40 percent to 60 percent of all pilot-reported faults are not corrected at the depot repair shops. These randomly occurring failures (bugs) are generally referred to as CNDs (Can-Not-Duplicates). CND rates usually increase as multiple unrepaired faults accumulate.
You think we're asking for trouble if we use neural networks; we've got trouble already. If pilots or autopilots rely on the data from the avionics equipment to fly the plane safely, and half or more of the faults are never fixed, how safe are airplanes really? The diagnostic problem gets worse as avionics become m
ore complex. Complexity demands that we use neural networks and other heuristic methods, if not as the final solution, then to help us figure out a more direct means to sort it all out.
Brent Sorenson
Vice President, Research and Development
Universal Synaptics
Ogden, UT
Flexible C++
Matthew Wilson
My approach to software engineering is far more pragmatic than it
is
theoretical--and no language better exemplifies this than C++.
BYTE Digest editors every month analyze and evaluate the best articles from Information Week, EE Times, Dr. Dobb's Journal, Network Computing, Sys Admin,
and dozens of other CMP publications—bringing
you critical news and information about wireless communication,
computer security, software development, embedded systems,
and more!