I'm more than a little embarrassed to say that 14 years ago I boldly predicted the death of the automotive assembly line. It's not the dumbest thing I ever wrote, but it's up there.

Mind you, a number of automakers - including Volvo - did indeed try to eliminate the assembly line with alternative vehicle-build strategies, but they never worked.

Peek into just about any assembly plant nowadays and "the line" is grinding along not much differently than it did when Ford Motor Co. was building model Ts. But in the early and mid-1980s the appetite for new technology seemed insatiable, and it was easy to get caught up.

"The traditional auto assembly line, arguably the world's largest sacred cow, is on the way to the slaughterhouse," I breathlessly announced in the June 1985 WAW. "Detroit's entire concept of building cars, pioneered some 75 years ago and changed little since then, is disappearing in a blaze of new technology - and new thinking."

The "line" would be replaced by a new type of conveyance that doesn't grind irreversibly through a plant. I explained new types of conveyors would come to dominate, ones that pause, stop and break off into "myriad tributaries" so that vehicle modules and subsystems could be assembled at their own best pace.

Plus there would be hundreds - and some day thousands - of automated guided vehicles scurrying to and fro with components and assemblies. Coupled with the then-new concept of modular assembly, I promised a wave of high-tech would bury the old assembly line ideas.

I wasn't just making this stuff up. I got lots of encouragement - and great quotes - from high-level manufacturing engineers and some well-known independent experts. No one was more enthusiastic about introducing exotic technology to the factory floor than Roger B. Smith, then chairman of General Motors Corp.

"Automaking is on the threshold of a technological revolution that will dwarf all our past accomplishments," he said effusively as GM launched its new high-tech "Poletown" plant in Detroit. Spurred by a series of "inspired and almost miraculous" innovations, he said automakers were "ready to usher in a new and golden age of opportunity."

I remember feeling euphoric after conducting dozens of interviews with automotive executives, consultants and robot salesmen about the avalanche of new technology that was coming to the factory floor: Besides robots and AGVs there were artificial intelligence, supercomputers and more.

This new industrial revolution would eliminate the vaunted $2,000 per-car cost advantage the Japanese enjoyed. Labor content would plummet and the U.S. would re-establish itself as an economical place to produce cars. Everybody talked about "leapfrogging" the Japanese.

I thought we were all dizzy with enthusiasm. Now I realize we were just plain dizzy.

While the automotive world was abuzz with talk of the "robot revolution" the robots actually were starting to revolt. They were painting each other instead of cars at GM's Poletown plant, and refusing to work at dozens of other facilities.

A year later, I wrote about how it all was falling apart. Our cover for that piece in 1986 depicted a cartoon of a robot sitting on a psy-chiatrist's couch, complaining that people expected too much from him.

Oddly enough, reams of bad publicity and hundreds of millions of dollars worth of canceled orders for robots and AGVs still didn't force the technology pushers out of town, and automakers didn't sit down and fundamentally rethink their strategy.

Consultant James Womack, president of the non-profit Lean Enterprise Institute in Brookline, MA, and co-author of The Machine that Changed the World, probably the most influential manufacturing book of the last two decades, says that during much of the 1980s the productivity problems plaguing U.S. carmakers were perceived as being simple: The Japanese were beating us because they had cheaper labor and more robots.

Machine, published in 1990, explained how companies could dramatically improve their performance through the "lean production" approach pioneered by Toyota Motor Corp., which - contrary to the perception - uses minimal automation.

For the decade prior to Machine, Detroit's knee-jerk response was to add as many robots as possible and then move whatever labor that couldn't be automated to low-cost locales such as Mexico. Cheap labor was equated with competitiveness.

I'd like to forget about all the "gee whiz" reporting I did in those years, but it pops into my mind every time I see a glowing report about the modular manufacturing miracle automakers and suppliers are creating in Brazil, and when I hear about Toyota allegedly being able to deliver custom-order Solara coupes to buyers in five days instead of the usual six to eight weeks. (For the record, Toyota says it can't).

As we stand on the edge of the new millennium and look ahead 20 years, absolutely no one is predicting totally automated "lights out" factories, where robots toil in the dark assembling modules and saving on the light bill. Not now, not 50 years from now. That's a huge change from the early 1980s when "lights out" seemed like a given in the 21st century.

Mr. Womack says totally automated factories might make sense "100 years from now," but for the foreseeable future, the capital expense of buying an automated plant and the cost of maintaining and programming the equipment far outweighs any labor cost savings, he says.

Some manufacturing experts and futurists are very excited by what automakers are doing in Brazil with modular manufacturing (see WAW - Aug.'98).

But when Mark Hogan, then GM's vice president and general manager-Small Car Group, described Yellowstone to me last January, alarm bells rang in my head. As it was proposed, Yellowstone would replace aging small car plants with new greenfield plants using the latest modular techniques. These latest techniques, including breaking up the car into 12 to 15 modular chunks that are assembled off-line at suppliers and then delivered to the assembly plant, could have been plucked straight out of my 1985 "gee whiz" manufacturing story.

On paper it sounds great. But the United Auto Workers union found the concept so offensive that the term "Yellowstone" now is banned at GM unless you're talking about the national park.

So if auto manufacturing isn't blinking lights and robots in 2020, what will it be? Mr. Womack and others suggests the following:

More regionalized production strategies. Global manufacturers will likely adopt a manufacturing strategy similar to that of Honda and Toyota, where products, platforms and suppliers are grouped into specific areas such as "North America." The idea of "global platforms" is nice, but probably not practical when fuel prices and consumer tastes vary so widely between Asia, North America and Western Europe. Honda, for instance, no longer tries to sell family sedans and minivans that are the same in Japan and the U.S.

A new distribution system. The current finished product distribution system will definitely be changed, but strategies vary widely. It is estimated that 30% of a vehicle's retail cost occurs after it leaves the assembly plant, making it ripe for cost-cutters. Automakers estimate that a better distribution system could chop that 30% by as much as half. Modifying the distribution system, though, starts at the assembly plant and includes the supply base that feeds it.

Cheap labor may become a lower priority. For automakers to reach the often stated goal of delivering a custom-order car to a buyer in five days instead of six to eight weeks, the supply chain has to be able to react faster. That means different product sourcing strategies that will speed the "velocity" of raw materials and components through the supply chain. That translates into establishing a close-knit regional supply base that has quick throughput times.

In this scenario, sending parts far and wide to take advantage of low labor costs will have to be re-evaluated. "If you really want to move to a world where the customer can order a car and get it in a couple of days, you really haven't got the time to send the parts to Malaysia and back," says Mr. Womack.

He calls today's tiered supplier base a "trucker's paradise" where too many low-value components are ferried back and forth among the lower tiers, while the top tiers just snap together components.

He uses the example of a wiper module that is shipped to Mexico to take advantage of low labor costs. The module has only 1 minute of direct labor, yet it travels 6,000 miles and spends seven weeks in the production pipeline. No one even knows how to calculate the cost of managing and tracking the logistics of this travel, he says.

Not only will the travel of individual parts be scrutinized much more closely, but all the coordination and logistics labor involved with tracking, transporting, storing and expediting components also will be watched more closely. The end result likely will be fewer white collar logistics jobs and a pared down travel schedule for parts.

Modular Manufacturing. The concept of saving production time and labor through the use of modules is as old as mass production. Henry Ford used modules extensively at his Model T plant in Highland Park, MI, including fuel tank, front and rear suspension, axle and engine modules. They were built up inside the plant and then funneled down ramps to the assembly line. Bodies were built by independent suppliers and dropped off.

It all worked because Ford minimized complexity. Not only did model Ts come in only one color, they also had no options. When competition forced Ford to add lots of paint colors and engine options, the natural leanness of his operation fell apart. In the 21st century automakers and suppliers will continue to struggle with the issues of modules and complexity.