There is a chapter on startups in my book Accidental Empiresin which I give all sorts of advice on how to make it big with your own high tech company. This was one of those instances of a pretty good observer coming to some sound conclusions that he is willing to share, but not follow. That explains my bank balance. The last of my many suggestions in the chapter was for readers to take me to lunch. Over the years, my phone has rung with dozens of lunch invitations, some of them even from people who went on to make a buck or two. Or three. This week, one of those companies I advised in their earliest days — Excite — was sold for $6.7 billion. Here's the story of that early contact and what came of it.

In the fall of 1993, I heard from a guy named Joe Kraus, the 23 year-old president of a six-man company called Architext. Shortly thereafter, I took the entire company out to Fresh Choice, a salad place. The bill came to less than $60. Here's a good rule: Take a vegetarian to lunch. The company consisted of five guys who were 22-23 and one old man of 25. They had met in their freshman dorm at Stanford University and stayed together ever since. The 25 year-old was the resident adviser. Five of the six were students of computer science or linguistics. Joe — el presidente — studied political science. Figures.

Listening to their story, it seemed to me that they were breaking at least one major rule right from the start. They had the wrong reason for founding their company. To my simple mind, there were many wrong reasons for starting a company, and only one right reason. The wrong reasons include everything from "to get rich" (that's the worst one of all and usually dooms the venture) to "I'll show him/her/them that they were wrong." The only right reason to start a company, I thought in those simpler days, was to invent something that didn't exist and was so cool that it was worth founding an entire company just to have whatever the invention was for your own personal pleasure. This is still a very good reason for starting a high tech company, but now I'll allow that there are other valid reasons. In the case of the Architext boys, their reason for starting a company was so they could continue to spend time together. I am not making this up.

They had no compelling technical vision, no ego, no chips on shoulders, no big point to make to the world. They didn't even know what it was they wanted to do. I wonder these days had they settled on, say, pizza delivery, if the world would be a different place today.

The smartest nerd in the very smart bunch was Graham Spencer, who is a charming and modest guy with the IQ of a third world nation. It just fell naturally to Graham to come up with some idea to keep the boys together. And so he did. Combining their skills in computing and linguistics, Graham took a 30 year-old idea called concept-based searching and made it practical. After about 18 months of hard work, Graham and the others came up with a way to find ideas hidden inside text without the specifically searching for keywords. And they found a way to make the searches very fast.

What they didn't set out to do was create an Internet company. That part happened pretty much by accident. The original concept was to search large text databases, perhaps on mainframes or local area networks. The World Wide Web wasn't such a big thing in 1993. The only reason they bothered with the Internet at all was because it provided a free source of the millions of words of text needed to perfect and validate the searching system. Their early demos were all conducted on newsgroup data.

To my credit, I knew instantly that this bunch was different. They had the technology, and they also had an easy willingness to adjust their vision of the company. This is important, because rigid startups are the first to die. I immediately offered to join them.

To their credit, the boys were suspicious. Technology was paramount to them at the time, and they probably had an innate sense that my expertise lay mainly in dead computer languages. We talked, I offered to write checks (14 percent of the company would have cost me $3,000) but their lack of enthusiasm was obvious. What good could they get from a 40 year-old gossip columnist who ate meat? It was a fair question, and so I withdrew, but not before finding them their first customer (International Data Group, my employer at the time) and their first venture capitalist (Steve Coit at Charles River Ventures).

From that point on, any advice I gave was bad. They had a funding offer of $2.2 million from Vinod Koshla at Kleiner Perkins, and I suggested they tell Vinod to go to hell. (Those were my exact words). Kleiner Perkins wanted to turn Architext (by then called Excite) into something that came to be called an Internet portal site. I thought this was crazy and told them so. Yes, I am an idiot. Here's another good rule: Know when to stop listening.

Other than doing several TV interviews over the years, I haven't had much to do with Excite since 1994. The company has reinvented itself a couple of times, gained adult supervision from a former TV producer named George Bell, and done its best to grow as big as possible as fast as possible, It was this strategy that led to the announcement on Tuesday that Excite would be purchased by @Home, the putative provider of Internet service to cable TV customers.

@Home is in its own phase of grow or die. The company has done a pretty good job of building (or at least planning to build) a fat data pipe to millions of cable customers. But just as Excite saw the need for revenue sources beyond advertising, @Home saw the need for revenue sources beyond the simple pushing of bits. They needed to own at least some of their own content. Excite provides that content and at a cost of less than nothing.

Less than nothing?

Less than nothing. It is an all-stock deal, and Internet stocks, while they appear to represent a lot of money, are still just made of paper. The deal will be classed by the IRS as a pooling of interests, and won't even be taxable. That alone means the deal cost nothing. What made it cost less than nothing was the fact that the day the deal was announced Excite shares gained $42 and @Home gained more than $12. One plus one equaled about 2.6 by my calculation, and anything over two counts as a negative cost — a profit.

Excite, like it's bigger, badder competitor Yahoo, is entirely about branding and brand awareness, so the name won't go away. Excite is better known than @Home. Current management at Excite won't change, either. Only the pockets get deeper. So in exactly the same spirit in which a little Mississippi long distance company became MCI-Worldcom, look for more content deals from Excite and more customer-acquiring deals from @Home, sucking-up smaller ISPs.

The one thing that has changed in all this is the identity of the competition. Unable to beat Yahoo at its own game, Excite is using @Home to change the game. The new target is America OnLine.

Makes sense to me. I only wish that somewhere along the line I had thought to buy the stock. See, I AM an idiot.