Every now and then we read about errors that we just have to laugh at. And now and then I get tired of writing serious columns. So today, either indulge me or just ignore this contribution altogether.
Just the other day, I received an advertising flier from a major company (politeness accounts for my suppressing the name) that had me in stitches. Here are the bargains advertised in the flier:
• 640GB internal hard drive: was $14999; now only $7999
• 640GB external hard drive: only $9999 – Limited time. Limited supply.
• 250GB external hard drive: was $14494, now only $7899
• Super DVD 22X Writer: only $9999
Amazing what happens when you leave out decimal points, eh? This is a classic example of a failure of data integrity: what went out were not the data that the sales department wanted to advertise!
I wonder how many thousands of customers thought that they had received a blast from the past? When I started work at HP in 1980, our largest disk drive was the HP7925, which had a staggering capacity of … [wait for it]… 120MB. It cost $25,000, which in today’s money would be around $100,000 (my 1980 Honda Accord cost me $7,000 that year).
One of my favorite examples of a spectacular data integrity blooper occurred in 1999 when for unknown reasons, the BUY.COM online store Web site listed a $588 Hitachi monitor at only $164.50 — and staff failed to notice the error until two days later, by which time there were 1,600 orders for this incredible bargain. The potential cost was estimated by the company at $320,000. BUY.COM filled 200 orders and told all the rest that they were out of luck. They also posted new language on their Web site addressing the non-validity of erroneous prices. This case was reported in the RISKS FORUM DIGEST.
Analysts speculated on the cause of the error. One intriguing possibility: the BUY.COM online store had a policy of underbidding any price on the ‘Net and may possibly have used knowbots to scour the Web looking for prices of products it was selling. Speculation had it that if a competitor accidentally or deliberately posted a bad price, the unsupervised knowbot could very well poison the BUY.COM Web site database. The same technique could be used in an information warfare attack to ruin a competitor. Even worse, the same problem could occur if two companies inadvertently used the same policy of underbidding all competitors and then simultaneously launched automated processes to lower the price without human intervention. Talk about a deadly embrace. . . .
When I was a child, I was raised by a generally abusive father who punished me for making mistakes while he was forcing me to learn stuff that was years ahead of what my little buddies were studying – not conducive to popularity among the 8-year-old crowd, but I did finish high school math by the time I was 9. One consequence of his, ah, teaching style was that I learned – the hard way – to check my work before I gave it in to him. When I reached McGill University in 1966, I carried on as I had learned and I remember my astonishment at finding out that other people did not do the same. One day, for example, I nearly caused a heart attack in another student when he said, “You’ve been punching in a lot of numbers on the calculator, haven’t you? How many numbers are you dealing with?” “Oh,” I answered, “I’m summing 1,200 numbers.” “You are adding 1,200 numbers?!?” “Yes – in groups of 100 – and each group twice (once forward and once backward) to be sure I’m entering them right.” He practically fainted. It was shortly after that that I turned to the FORTRAN compiler and punched cards to replace the calculator for that kind of work.
Strangely, I think that my father’s cruel response to mistakes is one of the reasons that I have never embarrassed or humiliated students who make mistakes in my courses – I just encourage them to figure out the root of their difficulties and solve the problems.
Anyway, enough of this airy persiflage. Could we just all agree to apply some simple quality control to all aspects of our work? Especially the stuff that gets sent out to lots of people??
