Of glitches and gremlinsgreenspun.com : LUSENET : Unk's Troll-free Private Saloon : One Thread
This story is not unique, merely typical---typical of the engineering world and of the world in general. Things go wrong, maddenly, unexpectedly wrong.
From the email of David Pogue, NY Times---
I had a great time working on "Piloting Palm," a new book that tells the story of how Palm Inc. and Handspring Inc. began and grew.
I don't mind admitting that my favorite parts of the story were the disasters. Andrea Butter, a former Palm marketing executive and the co-author of the book, remembers some particularly hilarious collisions of technology and humanity.
Just before the original Pilot organizer was released in 1995, for example, the tiny team at Palm faced a show- stopping glitch: A significant number of the first 10,000 Pilots didn't work when they arrived from the manufacturing plant in the Far East. Days went by, and nobody could figure out what the problem was.
Finally, Donna Dubinsky, the company's chief executive, shut down the production line and called all department heads together. The meeting wouldn't end, she announced, until the solution was found. "What is every single change we've made to the product from the time it was working to the time it stopped working?" she asked.
The only change that anyone could remember was that a small red sticker had recently been added to the inside of the Pilot's battery door, warning owners not to dawdle when replacing the batteries.
This tiny sticker triggered an improbable domino effect. Friction caused by the extra millimeter of paper thickness pushed the batteries away from their contacts just enough for the palmtops to lose power -- and everything in memory.
The solution was as simple as using a different paper thickness for the sticker, but that small change cost the Pilot its original shipping date.
Now, seven years later, history has repeated itself. Remember the m500 and m505 palmtops that debuted about a year ago? As summer turned to fall, Palm began to field mystifying reports from customers whose organizers suddenly stopped being able to HotSync (synchronize data with Macs and PC's). When Palm asked the customers to send in the crippled palmtops for replacement, its engineers could see that the built-in HotSync software was scrambled -- but they didn't know how it got that way.
The demographic patterns were even stranger: the glitch complaints grew more frequent as the weather grew colder, and furthermore seemed to affect customers in colder, northern states. (That, at least, explained why Palm's testers, in warm California, had never witnessed the problem.)
Finally, it clicked: static electricity was the culprit. It turns out that electrostatic discharge was zapping the redesigned m500 HotSync cradle. Palm m500 customers who stroked the cat or walked across a carpet before setting the palmtop into its cradle were loaded guns -- especially if they failed to touch a doorknob or kiss somebody first.
A new cradle contains a beefed-up electrostatic buffer that guards against such problems. It's free to all Palm m500 or m505 owners, whether or not they've experienced problems; check www.palm.com/support/m50XUSBcradle.html for details. (If you see an E or an H on the bottom of the cradle, you've already got the new one.)
Palm isn't always up front about admitting problems with its products, but it did the right thing this time. In any case, both of these stories from the company's lively history make it clear that building hardware is never an easy business. This recent story, in particular, is a vivid illustration that finding out about a glitch in a product you've already shipped can come as -- forgive me -- something of a shock.
Visit David Pogue on the Web at: http://www.davidpogue.com
-- (firstname.lastname@example.org), May 02, 2002
Gee, I really miss "tiger teams", "dynamite rebuilds", etc.
-- (email@example.com), May 02, 2002.
Amazing what a little static electricity can do to you *grin*
-- Cherri (firstname.lastname@example.org), May 02, 2002.
It's the entropy, stupid.
-- (Carville@DN.C), May 03, 2002.
Whatever can go wrong, will go wrong
-- (email@example.com), May 04, 2002.