At this conference, however, Masayoshi was basking in triumph. He owned a third of the stock in Yahoo, then a $6 billion company. He owned a third of GeoCities, too, a Web-hosting service that had inspired the public markets to go gaga (and which has since sold out to Yahoo). Son was finally being accepted as what he wanted to be: a true Net mogul.

As Son spoke, slides flashed on a screen behind him. There were slides with circles, slides with line charts, slides with equations and summation signs. But there was one particular slide that got the audience palpably excited. It showed a small square with the names of some old school industrial companies written below it, a bigger square with new companies like Dell Computer, a still bigger square with the names of huge media companies, and a last, biggest square with the next wave of Net companies. The squares were supposed to represent the relative values of each generation of business giants. And the point, ultimately, Son claimed, was that the top companies of the Net generation would have a market value 10 times that of their predecessors. Of course, it was all hype. But the audience loved it.

I could not help thinking of Masayoshi Son earlier this month when three investors — Son’s Softbank, plus the investment bank Goldman Sachs and the venture capital firm Sequoia Capital — let it be known that they had paid $275 million for 6.48 percent of the stock in Webvan. Webvan, as many readers of the business pages know, is an online supermarket headed by Louis Borders, founder of the Borders chain of bookstores. Webvan currently delivers groceries to customers in the San Francisco Bay Area. In investment speak, Webvan is a national online retail play trying to capture a big piece of the $350 billion dollar a year grocery market.

But more precisely, Webvan right now is a very big warehouse in Oakland, Calif., a headquarters office in Silicon Valley, an attractive logo, a fleet of trucks, some 300 employees, a computer system that links all these pieces together with an e-commerce Web site, a contract with construction giant Bechtel that commits the fledgling company to building 26 more warehouses at a cost of $1 billion, and a lot of hope. That’s it. Six and a half percent of this is worth $275 million, which gives the whole thing a nominal value of $4 billion. That’s an awfully pricey warehouse.

If you are scratching your head wondering how a warehouse with operations in just one small part of the country can be worth so darn much money, however, you’re probably wasting your time. There are certainly ways to calculate the potential value of the business. You can look at the value of Safeway, the giant grocery store chain, whose stock has a total value of $26 billion plus. You can, conversely, look at Peapod, the first online grocer, a service that the Wall Street Journal plugged in 1994 with the headline “Peapod’s On-Line Grocery Service Checks Out Success,” whose stock is muddling along at a single-digit share price. You can talk about profit margins and argue about whether online groceries will succeed in achieving the 5 percent profit margins they hope for — though most backers of online groceries will tell you that traditional supermarkets have margins of just 1 and a half percent and Safeway’s are at 3 percent. The problem is that all of this entirely misses the point.

In fact, there is one overriding reason why this warehouse in Oakland is worth $4 billion: The money managers who have invested in Webvan have already made a whole lot of money in other Net companies. And so, like Son at his conference, they give the entire project a halo of invincibility, no matter how preposterous the financial assumptions. Those money managers are two Silicon Valley venture capital firms, Benchmark Capital and Sequoia Capital.

Venture capitalists right now are the darlings of the business world. In fact, to call them simply “money managers” is asking for a fight. Venture capitalists insist that what they do is not manage money but build companies, and this is partly true. They certain provide management advice, some of which is good; they help recruit employees; and they contribute, to the best of their abilities, at board meetings. But their reason for existence is, put bluntly, managing money, much of which represents the investments of university endowments, pension funds and some of the large number of extraordinarily wealthy individuals. They invest this money in companies — including many technology companies — at an early stage of their development, hoping to multiply their funds 10- or 20-fold with every successful investment.

In Silicon Valley there is a definite pecking order among venture capital firms, and Sequoia and Benchmark are near the top, getting into the most sought-after deals. (Kleiner Perkins Caufield and Byers, which made its money and its name on early investments in Netscape and Amazon.com, has long been at the pinnacle.) Sequoia and Benchmark are on top now largely because each of them has had a single stunningly successful Net investment. For Sequoia, it was Yahoo; Sequoia provided $1 million of startup money for a 25 percent ownership in the company; Yahoo’s stock is now valued at $26 billion. Michael Moritz, the Sequoia partner responsible for backing Yahoo, is also the partner who sits on Webvan’s corporate board. Meanwhile, Benchmark is one of Silicon Valley’s younger venture capital firms — just four years old. Benchmark was the initial investor in the auction site eBay; its share is now worth about $880 million.

But here’s the big irony of the Silicon Valley pecking order: The biggest advantage of having backers from the club of top money managers is the mystique they bring. By their vote of confidence in a company, the top venture capitalists attract other investors, who put in even more money in later rounds of financing. Webvan counts as investors not only Softbank and Goldman Sachs, but also the French billionaire Bernard Arnault, newspaper company Knight-Ridder and CBS. All of them have ponied up tens of millions in financing to go along for the ride with Benchmark and Sequoia.

Well, guess what. Once a company has raised over $400 million, it’s almost certainly only a very short step away from selling stock to the public. Chances are that investors will bite once business magazines report that a company has a “value” of $4 billion. What that means is that the last 6.48 percent of the company was sold for $275 million, which implies that the whole company is worth $4 billion dollars — even though Benchmark and Sequoia got big pieces of the company early on, for much, much less. If the company does go public, the investors in the public markets will no doubt be eager to get stock in a company that comes with the imprimatur of gilt-edged money managers. On top of it all, the fact that Webvan and Yahoo share Softbank as an investor (in fact, Yahoo CEO Tim Koogle has a seat on Webvan’s board of directors) implies a deliriously full future for investors eager to bet on the future giants of the Net.

A scenario a lot like this played out last week, when investors bid up shares in Drugstore.com, an online pharmacy backed by Amazon.com and Kleiner Perkins, the most prestigious and powerful of all of Silicon Valley’s venture capital firms, by 179 percent in one day. The stock of Drugstore.com, a company that opened for business five months ago, is now worth just over $2 billion. (Amazon and Kleiner Perkins have also teamed up on a direct competitor to Webvan — HomeGrocer, which is already operating in Seattle.)

There is not a single venture capitalist who does not say that his goal (they are mostly men) is building “the next Microsoft.” But here’s the ugly truth: It makes no difference to a venture capitalist’s returns if a company grows to dominate an industry or goes bankrupt five years after selling stock to the public. A Boston Market (formerly Boston Chicken) can be as good as a Microsoft because a venture capitalist needn’t wait for the company to prove itself; not long after a company goes public, the venture capitalists can cash out. In fact, a Boston Market — a high-flyer which went bankrupt — can be better than a Microsoft, because all of Microsoft’s stock taken together wasn’t worth a billion dollars until the company had been around for nine years.

What venture capitalists are looking for is — don’t laugh — a “liquidity event.” That’s biz speak for selling a company or taking it public, turning their ownership stake into cash. Typically, a year after a company goes public, venture capitalists are allowed to sell their stake. Sometimes they will do that. More often the venture capital firm will split the stock up among investors in its fund, letting them quietly sell it in small bits and reap huge winnings. Or they might hold it, if the investors think the company really will be the next Microsoft.

The upshot is that venture capital firms are not in the business of taking a long time to build companies. At one time, they might have been, but that is certainly not true in the age of the Net. Like Masayoshi Son at the San Francisco conference, they are in the business of retailing dreams of corporate grandeur. If they succeed in selling those dreams, they can make a lot of money. Whether they come true, however, rarely need be their main concern.

Later, the human contestants made jokes about the “Terminator” movies and robots from the future. Indeed, four questions into the round you had to wonder if the rise of the machines was already upon us — in a trivial sense at least.

Watson tore through a category about female archaeologists, repeatedly activating a mechanical button before either Ken Jennings or Brad Rutter could buzz in, then nailing the questions: “What is Jericho?” “What is Crete?”

Its gentle male voice even scored a laugh when it said, “Let’s finish ‘Chicks Dig Me.’”

Jennings, who won a record 74 consecutive “Jeopardy!” games in 2004-05, then salvaged the category, winning $1,000 by identifying the prehistoric human skeleton Dorothy Garrod found in Israel: “What is Neanderthal?”

He and Rutter, who won a record of nearly $3.3 million in prize money, had more success on questions about children’s books and the initials “M.C.,” though Watson knew about “Harold and the Purple Crayon” and that it was Maurice Chevalier who sang “Thank Heaven for Little Girls” in the film “Gigi.” The computer pulled in $4,400 in the practice round, compared with $3,400 for Jennings and $1,200 for Rutter.

Watson is powered by 10 racks of IBM servers running the Linux operating system. It’s not connected to the Internet but has digested encyclopedias, dictionaries, books, news, movie scripts and more.

The system is the result of four years of work by IBM researchers around the globe, and although it was designed to compete on “Jeopardy!” the technology has applications well beyond the game, said John Kelly III, IBM director of research. He said the technology could help doctors sift through massive amounts of information to draw conclusions for patient care, and could aid professionals in a wide array of other fields.

“What Watson does and has demonstrated is the ability to advance the field of artificial intelligence by miles,” he said.

Watson, named for IBM founder Thomas J. Watson, is reminiscent of IBM’s famous Deep Blue computer, which defeated chess champion Garry Kasparov in 1997. But while chess is well-defined and mathematical, “Jeopardy!” presents a more open-ended challenge involving troves of information and complexities of human language that would confound a normal computer.

“Language is ambiguous; it’s contextual; it’s implicit,” said IBM scientist David Ferrucci, a leader of the Watson team. Sorting out the context — especially in a game show filled with hints and jokes — is an enormous job for the computer, which also must analyze how certain it is of an answer and whether it should risk a guess, he said.

The massive computer was not behind its podium between Jennings and Rutter; instead it was represented by an IBM Smart Planet icon on an LCD screen.

The practice round was played on a stage at an IBM research center in Yorktown Heights, 38 miles north of Manhattan and across the country from the game show’s home in Culver City, Calif. A real contest among the three, to be televised Feb. 14-16, will be played at IBM on Friday.

The winner of the televised match will be awarded $1 million. Second place gets $300,000, third place $200,000. IBM, which has headquarters in Armonk, said it would give its winnings to charity while Jennings and Rutter said they would give away half theirs.

In a question-and-answer session with reporters after the practice round, Rutter and Jennings made joking reference to the jump in technology Watson represents.

“When Watson’s progeny comes back to kill me from the future,” Rutter said, “I have my escape route planned just in case.”

Jennings said someone suggested his challenge was like the legend of John Henry, the 19th-century laborer who beat a steam drill in a contest but died in the effort. Jennings prefers a comparison to “Terminator,” where the hero was a little more resilient.

“I had a friend tell me, ‘Remember John Henry, the steel-drivin’ man.’ And I was like … ‘Remember John Connor!’” Jennings said. “We’re gonna take this guy out!”

——

Associated Press writer Leon Drouin-Keith in New York City contributed to this report.

Meanwhile, Facebook gets the capital to keep buying talent and startups, and to fuel its expansion in all kinds of other ways — and it gets to sell stock in what amounts to a shadow stock market that’s growing faster than regulators seem willing or able to understand, much less deal with.

This looks like a better deal for Facebook than its investor, putting Facebook’s value at $50 billion, which makes sense in today’s increasingly bubble-like market. Silicon Valley is going a bit wild again– not as crazy as the late 1990s, mind you, but there’s a froth element to the local economy.

An interesting question now is whether Facebook will do a a real public offering anytime soon. Federal rules require significant data disclosures when a company has 500 or more shareholders, and surely Facebook is at that point or nearing it. The Goldman deal may be an end-run around the rule, with Goldman not selling Facebook shares to its clients, but rather selling shares in something it (Goldman) owns. If this is the game, and if the SEC lets it happen, the 500-shareholder rule has become meaningless — and markets are all the more opaque at a time when transparency is more needed than ever.

Opacity is a growing issue. A thriving shadow marketplace has emerged for big startups that haven’t done IPOs, so big that the Securities and Exchange Commission is, at least in that space, looking into the wheeling and dealing. For good reason: Many if not most of the investors in these markets have no idea what the true financial picture may be of the shares they’re buying.

Facebook seems like a no-brainer right now. It reportedly has passed Google as the most visited website, and it’s growing in power and people. And Goldman, for all its sleazy ways, has smart people making investment decisions.

But Goldman was also a big investor in the financial bubble that nearly toppled the global economy. It escaped ruin only because we, the taxpayers (actually our children and grandchildren) rescued it and the rest of the banking industry. That was and remains Goldman’s real genius: making giant bets with other people’s money.

But the most important result may ultimately be what this move, among others. does for public understanding of the role of Internet service providers of all kinds. As Amazon.com’s recent takedown of the Wikileaks site it was hosting demonstrates, we are at the whims of the companies that provide the services, and they are increasingly demonstrating that we should be highly skeptical about their commitment to our data’s longevity.

We put our data — our websites, photos, bookmarks, email and more — on their sites. But they can, and do, change their terms of service at will, doing what they please with what we’ve put on their servers. And sometimes they just shut down the services they’ve been providing. They may do it for good reasons, or absurd ones. It doesn’t matter. The point is, they can.

As noted here some months ago, we all need a Plan B for just about everything we do online these days. If we give others a choke point over our communications, we are inviting them to throttle us.

Note: The original version of this piece said Yahoo was closing Delicious. That was based on a variety of credible — and, as noted, unrefuted — news stories that started appearing more than 24 hours ago. They were based, initially, on a Twitter posting that linked to a screenshot taken at an internal Yahoo meeting. The screenshot, which has now been taken down, had Delicious among a group of Yahoo services that were being “sunsetted,” which is corporatese for end of life.

Whatever Yahoo’s intentions with Delicious, my points here stand. Even if the service is sold, a new owner might radically change the terms of service (as Yahoo itself could do at any time). The users’ insecurity remains, whatever the ownership may be.

These changes appear to make plenty of sense for Netflix, because the company will avoid the cost of buying and then mailing the millions of DVDs customers like me have been receiving. And, indeed, on Monday Netflix announced it was going to offer customers an all-online streaming plan for $8 a month.

I suspect there’s been a misstep, however, if I’m any example of the Netflix customer base. I’d been paying $17 per month for a plan that allowed us to have three DVDs out at a time, plus being able to view streaming content anytime. But Neflix has raised our rate by $3 a month, or about 18 percent.

There are way too many problems with the streaming-only plan to even consider it at this point. At the top of the list is the fact that the Netflix catalog of DVDs is vastly, vastly greater than what you can watch online. If the company really wants to be a streaming-only outfit, it needs to persuade the robber barons of Hollywood to digitize everything sooner rather than later.

And while the quality of the streaming is generally OK if you have a fast enough broadband connection — though it doesn’t look as good on my computer as a DVD — network congestion (in my experience) can cause the video to degrade in quality or, in some circumstances, pause altogether. I tried it on a hotel Wi-Fi recently, and finally gave up as the film kept stopping while the stream caught up.

So when the e-mail arrived announcing the price hike, my reaction was: Sorry, no sale. We’ve moved to a lower-cost plan that allows one DVD out at a time, for $10 (also more expensive than that plan used to be), plus streaming. The various plans Netflix offers now range up to $56 a month, and slightly more if you’re renting Blu-ray discs.

Netflix has leveraged the broadband Internet structure like no other company. It now accounts for a significant amount of evening data traffic, by all accounts. I’m guessing that heavy Netflix users are going to pay for the money they save in other ways when they start running into data caps that some carriers have put on their basic Internet service.

Wall Street was thrilled with the latest Netflix maneuver, pushing the stock price way up on Monday (though it eased off slightly this morning). The share price has roughly quadrupled in the past year — evidence of investors’ love for the company, an infatuation I believe has been mostly justified.

But I’m convinced that this move by Netflix is too little, too soon. And I’m betting I’m not the only one who feels that way.

The complaints grew loud enough to persuade Google to revise the Gmail settings so users can turn off conversation view and unravel their messages.

“We really hoped everyone would learn to love conversation view, but we came to realize that it’s just not right for some people,” Google software engineer Doug Chen wrote in a Wednesday blog post.

The aversion to conversation view doesn’t seem to be widespread. Gmail ended July with nearly 186 million worldwide users, a 22 percent increase from the same time a year ago, according to the research firm comScore Inc. Both Microsoft’s Windows Live Hotmail (nearly 346 million users) and Yahoo’s e-mail (303 million users) are larger, but aren’t growing nearly as rapidly as Gmail.