The Unreliable Reliance On Computers

Building a bridge or a walkway is, as a minimum, a daunting task, loaded with responsibility, for the safety of its users.  Many actions must be performed with perfection in mind; to this end the following  must be taken into consideration, as minimum, to ensure a safe end product.

1.   Incompetent and/or negligent design.
2.   Lack of adequate supervision and maintenance.
3.   Lack of competent information about maximum loads expected.
4.   Meeting unreasonable cost estimates.
5.   Contractors taking shortcuts  in construction.
6.   Contractors substitute inferior materials.
7.   In competent and/or corrupt inspection 

While the above listed items are, without a doubt, paramount to a successful project, there is one more consideration, not to be taken too lightly.  It is the unreliable reliance on Computer Aided Design (CAD) systems.  These systems are first of all designed by humans, who may be, or not, competent, to specify the many parameters necessary, for the system to work.  These systems consider, as much as possible, historical data on prior recorded failures, strength of materials based on laboratory testing.  From my own experience, in another field, I have seen laboratory material test data used, without considering the inherent statistical variations present.  As a result, engineers, more often than not, work with nominal material test data; i.e. worse case is seldom evaluated, thus unknown.

In days gone by, the engineer, without all these computer systems, had to constantly evaluate his calculated results, whether or not they passed the test of reasonableness because the slide rule, in most cases, did not provide the location of the decimal point (safe log scales).  Thus, the engineer of old, had to know pretty much, approximately, what the result should be, based on experience, i.e. be reasonable.  In those days, calculated results were not always super accurate, but acceptable for practical purposes.  Engineers allowed for this shortcoming by increasing the safety margin of their design accordingly.

Today's engineers have a great digital arsenal with which to get more accurate calculated results.  This, however, is not necessarily a blessing, because if the underlying computer code introduced an error by mistake, negligence, or incompetence, the erroneous calculated result, no matter how accurately expressed, can be the cause of a massive project failure, making the national news.  The modern engineer, is now conditioned to believe that the computer can do no wrong.  As a result, the engineer might not add an extra sufficient safety margin, in his design, and could in fact choose a margin coming closer and closer and closer to 1.0, at which time the safety margin has disappeared.

Silly program errors can easily sneak into the world of today's engineering world.  An example comes to mind, a corporate entity required the reporting of defects in PPM, instead of the usual percent defective.  The calculated results were expressed with great precision, oozing confidence, and some.  The corporate entity was very please with the low PPM defective, and held up the local division, as example for others.  However, upon my closer inspection, the calculated result did not pass the test reasonableness. Here is why, by example.
 If one had 25 pieces defective in a shipment of 1,000 pieces. 25/1000= .025 or 2.5% defective.
 . 025 X 1,000,000 = 25,000 PPM defective.  This is to say, we take the fraction defective, 0.025 and multiply it by one-million, to get the PPM defective.

The engineer who programmed the PPM calculation, instead of multiplying the fraction defective by 1, 000,000, actually divided by 1,000,000.  When we, by error, divide .025 by 1,000,000, instead of multiplying, we get instead of the correct 25,000 PPM defective, .000000025 PPM defective.
 Needless to say, corporate did not apply the test of reasonableness, when receiving this obvious bogus report.   When I pointed this out to the engineer and dept. manager, they, for some time, refused to see their error; their argument was, it was calculated by the computer, how could it be wrong?  Herein lies the danger, our modern engineers are subject to the unreliable reliance on computers.