By Paul Murphy, author of The Unix Guide to Defenestration
Quick: what's the single most dangerous thing that could happen to your data center? Right, a senior executive with a stupid idea.
On any standard measure of IQ you're probably smarter than the people who run your organization and you certainly know a lot more about computing than they do. Genuinely talented people tend to be humble because they're comfortable with themselves, something most executives aren't.
In most cases that means they adopt arrogance as a defense when forced to deal with people they know are smarter or more dedicated - that's why the most unreasonable demands from on high are often accompanied by put downs. The offensive behavior is a defense; part of the fight or flight mechanism governing so much of our behavior.
If you've ever wondered why otherwise sane executives accept technical advice from the ninnies who write In Flight magazines, daily newspaper tech columns, or other pulp press garbage, this is why. It's another manefestation of the executive's basic insecurity that he looks for stuff he can easily understand and make you do as an assertion of his superiority.
What that means is that when your boss falls in love with some inanity and demands that you immediately implement it, arguing with him is exactly the wrong thing to do. It may not actually be possible, you may already have something better implemented and in use, or it may be absurdly expensive and ineffectual, but telling him any of these things is a mistake. Instead, you should do nod sagely and promise you'll set about making it happen.
That has two effects; both positive. It reduces his threat perception, thus opening the way to rational behavior; and, equally importantly, it buys you time for things to settle out.
The first thing to do with that time is look closely at the idea. Perhaps it's your reaction that's clouding perceptions and this idea is not so stupid after all? If not, not making this an issue may help it simply go away, or you can launch an exploratory or pilot project that wastes so much of his time that the whole thing dies of its own inertia.
The very best thing, hwoever, that you can usually do with something like this is to use his idea as cover for something of real value that you couldn't otherwise get budget approval for.
In one case where I got this wrong the business owner became convinced he needed an Intranet and, more importantly, that intranets could really only be done using Lotus. At the time, we were using Apache, webDAV and Hypernews to manage a half dozen roaring internal discussion groups and some on-line manual development projects but my reaction, in pointing this out, was dead wrong. For three years afterward this guy kept pushing Lotus, and made knowledge of Lotus a condition in hiring my replacement.
In another I got it right. The VP finance misunderstood some discussion about PC reporting tools he'd read in a newspaper and demanded that we implement "planar analytics" as soon as possible. I still don't have a clue what that is, but we got a special budget to implement an Oracle dashboard and hire a support analyst out of it essentially by labelling the start up icon "planar analytics."