FORTUNE home comp_profiles investor careers smallbiz tech tech
SEARCH FORTUNE

GET QUOTE



Click here to subscribe to Fortune.
Provide banner and clickthrough for non-JS people.
Fortune Datastore
Magazine Customer Service
Fortune Conferences
Special Sections
Magazine Advertisers

Current Issue
Archive
Site Map
Frequently Asked Questions
Press Center
Contact Fortune
Advertising Information




Michael Schrage
Michael Schrage
BRAVE NEW WORK
Good Network, Bad Data
Why zippy new computer networks can actually make more stuff wrong.
FORTUNE
, 11,
By Michael Schrage

Send to a Friend Print Subscribe to Fortune

True story: Fortune 250 firm builds superb network that seamlessly links its managers enterprise-wide. The network is so good and so fast that top executives see all the performance data at the exact same time the line managers do. HQ doesn't hesitate to pick up the phone or zap an e-mail to underlings, demanding explanations and clarifications. When the line complains of micromanagement, the top dogs bark, "You expect us to ignore data that run the business? Get real!" The result? The line managers con the network. They delay posting data online. A few even post misleading and inaccurate data to mollify their bosses. Within six months, the technically excellent network is carrying crap data.

Almost-true story: Capacity-squeezed supplier starts to limit how much product it allocates to customers. Customers now get barely 60% of what they need. Not being morons, customers have their supply-chain software pad their orders by 40% in the next cycle. But the cagey supplier cuts the allocation by 50%. Customers then double their already inflated orders. The supplier slashes the allocation yet again. Customers double-down one more time. But by now the supplier has boosted capacity. Customers can get everything they over-ordered. Now there's a glut. Sound familiar?

Hypothetical story: Colleague sends urgent e-mail. Needs to know ASAP when you will finish the Epsilon Project. You're confident you'll be done by the 15th. To be safe, you respond that you'll finish by the 18th. She attaches that due date to a memo for her boss, but she warns that you have been known to miss deadlines. Her boss can't afford to mismanage expectations, so he tacks another week onto the delivery estimate. But Sales, having been burned by him in the past, has programmed its project-tracking software to automatically add another four days to delivery dates from his department. The customer gets the e-mail announcing Epsilon's ship date. Because the customer has built a four-day cushion into its own expectations, this now puts the Epsilon delivery a full 72 hours beyond its acceptable production deadline. The customer immediately cancels its order. Your boss e-mails you that Project Epsilon is dead.

What do these three examples have in common? All parties behaved rationally based on the best information available to them--and ended up dramatically worse off than before. Network technology effectively amplified and accelerated that rational pathology. Rational pathologies appear to infect even the best of companies. No company is wired better than Cisco, for example. Yet though it has publicly bragged that it can close its books within 24 hours, Cisco had to take a $2.2 billion inventory write-down--almost 50% of its sales--in the first quarter of the new millennium. Was it bad data? Bad luck? Bad timing? Or something else?

As information becomes ever cheaper, data integrity becomes ever dearer. In a marketplace where overbidding is a rational response, disclosing how much you are overbidding is irrational. It's rational to build slack into a process, but openly disclosing precisely just how much slack may be irrational. It's rational to assume that managers would prefer to manage based on timely and accurate information. But when that invites constant interference--and added costs--it may simply be more cost-effective to misrepresent.

We've all heard the techies sigh, "Garbage in, garbage out." That maxim is misleadingly true: If you put bad information into a network, you'll inevitably get bad information out again. But the truism obscures the fact that networks themselves invite abuse. Perversely, they often provide greater incentives to distort than to disclose. Why do networks facilitate their own befouling? Because their designers neglect to account for the true costs of disclosure. When the perceived costs of reasonable disclosure exceed their perceived benefits, then distortion and deception reign. The market "clears" only when the costs of the distortions outweigh their benefits.

The solution? There isn't one. These problems can't be solved; they can only be managed. Getting people to honestly disclose how much they will honestly disclose is wishful thinking, not business reality. But business reality dictates that organizations that commit to strategic networking must invest as much effort in designing the incentives for honest disclosure as they do in designing the technical infrastructure itself.


Send to a Friend Print Subscribe to Fortune
 
©Copyright 2002 Time Inc. All rights reserved. Reproduction in whole or in part without permission is prohibited.
Privacy Policy   Terms of Use   Disclaimer   Contact Fortune