|
At a recent professional meeting, a system safety engineer commented, “We do software safety by not letting software do anything important.” That would
seem to be an excessively harsh solution, yet it strikes a chord for most of us, as rarely does a day go by when we don’t find ourselves wailing bitterly about something that
software did or didn’t do. Newsworthy incidents, such as the Christmas-weekend grounding of an entire airline or the release of thousands of credit card numbers onto the
Internet, remind us just how unsafe software can be — and meanwhile, our everyday encounters remind us constantly of the many ways it can be so downright annoying! Who on
earth thought that “It looks like you’re writing a letter” would be helpful, or that anyone would want their checkbook program to auto-fill in the name of someone they wrote
one check to, five years ago, instead of the utility company that they pay every month? Come to think of it, that system safety engineer may have hit on something.
The key word seems to be “letting” software do things. We don’t have to let our everyday software drive us nuts — we can turn those annoying features off,
although (admittedly) the manufacturers make us work very hard to find out how. A larger question would be why software should be allowed to control something as important as
the operation of an entire airline, or why there was a path that could be opened between those credit card numbers and the Internet. What cannot happen generally doesn’t.
Software is a tremendous convenience, and we certainly love to let it save us time and money. Is the occasional serious incident part of the price we must pay
for that? If our answer depends on our personal experience — i.e., whether we were among the victims or just one of those who saw the news report and said “oh, what a shame” —
then we aren’t taking our responsibilities as system safety professionals seriously enough.
So how do we come up with guidelines for the level of power given to software in a system? Here’s a suggestion. Think of software as your worst employee. It
can’t really think for itself. It generally does what it’s told (if and when it works at all), but not necessarily in the way you want. It’s given to sudden bouts of deviant
behavior, and it may never tell you why. Above all, it just doesn’t care. It has absolutely no feelings for your business or your staff and customers, or even a social
conscience regarding the public at large. Now, how much power should it have?
Keep a tight rein on that employee. And if it gives you any flack, fire it.
I hope you’ve now resolved to think carefully about what software can and can’t do in your system. If so, here’s the reward for all of you who’ve been
screaming “tell me how! tell me how!” since paragraph 1 of this article:
To turn off those infuriating “hints” in most versions of MS Word:
Right-click (control-click on Macintosh) the Office Assistant. On the shortcut or
contextual menu, click Options. From the Options tab, un-check the “Using Features more effectively” box and click “OK.” (Try not to laugh too hard at the words “more
effectively.”)
To get rid of that ancient name in most versions of Quicken:
From the Lists menu, select Quick-fill Transactions. This will open up a list of previous
payees. Highlight the ones you want to get rid of and hit “Delete.”
|
