(FORTUNE Magazine) – A LONG TIME AGO--records indicate the year was 1994--the futurist Dave Barry wrote a column in which he envisioned a Home Shopping Network program called Interactive Salad Bar during which viewers, "by manipulating a console in their homes, can direct the selection of ingredients in a studio thousands of miles away, and then have the actual salads delivered to their homes, ready to eat, within two working days." He would not be the only one to see around the corner with such precision. Time Warner's Gerald Levin built a cable television network in Orlando to deliver news and movies on demand--at a cost of $5,000 per home. Bell Atlantic's Ray Smith opined that movies delivered over telephone lines would drive Blockbuster out of business. Oracle's Larry Ellison said that it would someday be possible for us to order a pizza--even a pizza with toppings!--using a TV remote control. AOL's Steve Case foresaw a day we could toss away the TV remote altogether because we would henceforth be entertained by "streaming video" on the Internet.

And so you may be forgiven if you raise a skeptical eyebrow at the news that a couple of tech-savvy fortysomething executives from the media and entertainment world have once again begun to chant the mantra so often heard before--that viewers will be able to watch whatever they want to watch, whenever they want to watch it. But bear with me, if you will. Brian Roberts, the chairman and CEO of $18-billion-a-year cable television giant Comcast, and Steve Burke, the firm's chief operating officer, are not painting a blue-sky picture of the future. They are describing a video-on-demand service that they are rapidly rolling out, without much fanfare, to their 21.5 million customers, which is by far the biggest subscriber base in television. What's more, their customers have eagerly seized more control over their TV viewing. In just 18 months, four million subscribers have been hooked up to video-on-demand, and usage of the service is increasing the way Internet traffic did in the late 1990s. Then again, we know what happened to investors who bet big on the Internet. This could turn out to be another Interactive Salad Bar.

Still, advertisers and broadcasters are watching nervously as the cable industry gradually makes video-on-demand available to its 73.7 million U.S. subscribers. Time Warner Cable, the second-biggest cable operator (and, like FORTUNE, a unit of Time Warner), was actually the first to introduce the service, by delivering movies on demand to its customers in Hawaii in 2000. It now has about 4.6 million video-on-demand customers, nearly as many as Comcast. Others, including Cox and Cablevision, are following suit, albeit with strategic twists of their own. But no company has embraced video-on-demand with the fervor of Comcast, and no company can do more to make it happen. Even if they prove to be wrong about the future, Roberts and Burke have to be taken seriously when they say their aim is to revolutionize--once again--the way we all watch television

You see, the first television revolution is over. It was about choice. Beginning in the late 1970s, cable and satellite providers delivered first a few, then dozens, and now hundreds of new and increasingly narrowly focused channels. HBO and CNN and MTV led to a half-dozen flavors of ESPN; Discovery Wings, an all-aviation channel; and Toon Disney en Español. The result is that this year, for the first time, cable and satellite channels as a group--admittedly, a very large group--capture more viewing than the once-mighty broadcast networks do. A second revolution is underway, the people at Comcast will tell you. It is about control--and even more choice. This time viewers will choose from a menu of thousands of programs--they'll watch shows, not networks. Some programs will have mass appeal, but many will target niche audiences. Viewers will watch them when they want, with the ability to pause, stop, restart, and save the programs, which are stored on servers located in what is called the head-end of your local cable system.

"We think video-on-demand is a revolutionary platform," says Brian Roberts, 45, a brainy, soft-spoken executive who grew up in the cable industry. "We're going to change television in dramatic ways." Brian's father, 84-year-old Ralph, started Comcast with a single cable system in Tupelo, Miss., in 1963, and the family still controls 33% of the voting stock. Steve Burke, 46, was also steeped in television from a young age; his father is Dan Burke, the admired former president of Capital Cities/ABC. "Television," Steve Burke predicts, "is going to change more in the next five years than it has in the last 20 or 30."

THE EVIDENCE to support their claims can be found in the living rooms of Comcast customers. I'm one of them. I pay a hefty $130 a month for the all-you-can-eat menu of 80 basic cable channels, 150 digital channels, 50 premium channels--including multiple versions of HBO and Showtime--45 music channels, high-speed Internet access, and video-on-demand. Video-on-demand is the game changer. It means that I watch Meet the Press or the NBC Nightly News anytime, without having to remember to record them. Choosing from a library of about 2,000 offerings, I can find movies like Along Came Polly, Finding Nemo, and Mystic River without going to the video store. I could even, if I so desired, watch such fare as Yoga Retreat, Classic Cars, and Wine World, part of a suite of specialized channels called MagRack.(Alas, Birdsight has been canceled.)

Trust me when I say that this transforms the way my family watches TV. Fetching shows from a library created by a cable operator is fundamentally different from storing shows in the home with a VCR or a digital video recorder of the type pioneered by TiVo. Say you heard about Dan Rather's mea culpa on the CBS Evening News the next morning. You could retrieve it with a click of the remote. At my house we watch less live television than ever before. We don't waste time with shows we don't really want to see. And we don't watch commercials except during news and sports. Television feels more like the Internet than the world of limited choices and network schedules in which most of us grew up. I've seen just about every episode of Curb Your Enthusiasm, and I don't even know when it's on.

Now before we get carried away, a few caveats. While Comcast is trying to change the world, the entire cable industry is in the fight of its life with the direct-to-home satellite-TV providers, newly energized by Rupert Murdoch's acquisition of DirecTV. "It's all about the sky vs. the ground, and nobody knows that better than Brian and Steve," says a media bigwig who sells programming to both cable and satellite. Right now, cable's losing. In the first six months of this year DirecTV and Charlie Ergen's Echostar together added about 1.7 million customers. Comcast lost 61,000 customers. Other cable operators with substandard offerings or poor management lost many more.

The threat from satellite is the biggest reason Wall Street has turned bearish about cable. Comcast's stock, which had been a stellar performer since the company went public in 1972, is down by 15% this year. Comcast's failed takeover bid for the Walt Disney Co. added to the pressure on its shares. Investors reasoned that the Roberts family would not have offered to give up a big stake in Comcast to get Disney if they did not harbor serious doubts about cable's future. It did not help the industry's prospects when family-run Adelphia went bankrupt or when accounting problems led to earnings restatements, executive dismissals, and SEC investigations of big operators Charter and Cablevision.

Comcast is betting that video-on-demand will turn the tide against Murdoch and Ergen. Video-on-demand sets cable apart--cable's two-way platform enables it to store many thousands of hours of programming on servers and let viewers fetch what they want. The satellite companies can't do that.

"The world is clearly evolving toward the concept of TV when you want it and how you want it," says Craig Moffett, senior cable and satellite analyst at Sanford C. Bernstein & Co., who thinks Comcast is on the right track. But, he warns, while video-on-demand will help the company retain its best customers--those who already buy premium channels like HBO and Showtime--it won't appeal as strongly to those viewers who are defecting to satellite because it tends to be cheaper. "If you're playing a mass-market game, the VOD strategy may not be that compelling," he says.

To beat back the challenge from satellite, Roberts and Burke will need to live up to their reputation as the industry's best managers. Their goal is to make Comcast a company that can exploit its scale when necessary and yet remain nimble, customer-focused, and local in its operations. They talk about building a big blue-chip company that feels like a small family firm to its workers and its customers.

Comcast was transformed two years ago when it acquired AT&T Broadband and its 13 million cable customers for about $50 billion. The company's 21.5 million subscribers make it twice the size of Time Warner Cable--of which it owns 21%. Roberts wants to get even bigger: Comcast and Time Warner have decided to craft a joint bid for the assets of bankrupt Adelphia, which has 5.4 million subscribers, and to unwind their partnership so that each can grow independently.

As the dominant force in cable--a business that has always played follow-the-leader--Comcast already shapes the industry's programming, technology, and politics. It is the biggest customer of Viacom, Time Warner, NBC, and the other content giants--and it is reputed to be the toughest. "Scale matters," Roberts says. "Scale allows you to control your destiny, to shape the new ideas, to efficiently manage your costs, and to perhaps enable new businesses and new assets."

Comcast does far more than sell cable. It has become America's No. 1 provider of high-speed Internet access, with six million customers. The company will generate about $3 billion in revenue from selling broadband this year, enough to invest in new technologies, like video e-mail, and new online content, like video clips on its home page. (I watched Martha Stewart's plea to go to jail on a CNBC feed delivered to www.comcast.net.) Comcast employs 3,200 ad salespeople, who will sell about $1.3 billion worth of advertising this year, most of it to local sponsors that are shifting money from broadcast TV to cable. (The company's cable footprint touches 21 of the top 25 markets in the U.S.) Over the next two years Comcast will roll out an Internet phone service with a number of snazzy features, like caller ID on your TV screen so that you know who's interrupting your show (or not). Roberts, a gadget freak who will lead a group of cable executives to Japan and Korea next month to see what's new over there, wants to promote more innovation inside Comcast. "We have to genetically rewire the company to be a technology leader," he says. "For 35 or 40 years we've been a fast follower. John Malone, Time Warner, others set the pace."

Its scale has also enabled Comcast to expand its content portfolio. It owns and operates a bunch of second-string channels--Golf Channel, E! Entertainment, Outdoor Life, and Tech TV--whose earnings are nevertheless growing at a double-digit pace. In a little-noted deal last winter, Comcast acquired 51% of TV Guide's interactive program guide--the so-called gateway to digital television--merely by agreeing to carry the guide. More recently its last-minute support for Sony and a consortium of private-equity investors gave that group the edge in the competition to acquire MGM. Next up, most likely, is a children's channel, to be developed with PBS and Children's Television Workshop, the people who make Sesame Street.

Here's the worry, though: No one has run a cable company close to this big and this complicated before. Cable was a virtual monopoly for years, and the biggest owners--first Malone, then AT&T--were lousy operators. The cable guys could barely bring themselves to answer the phone and show up for appointments and keep pictures on the air when they sold nothing more than TV. Now they must manage a variety of digital packages, the video-on-demand platform, high-definition TV and DVRs, as well as computers and modems and telephones and the blizzard of customer-service problems they bring. Comcast employs nearly 12,500 people just to answer the phone, and it expects to get about 200 million phone calls this year. "In a competitive world," Burke says, "if you don't take care of your customers, someone else will." Can Comcast persuade its customers to love their cable company? Its big ambitions won't amount to much if it fails to do the little things right.

IT'S THE MONDAY after Labor Day, and Roberts and Burke have come to the Comcast Center, the new basketball arena on the campus of the University of Maryland, to unveil a big deal with the NFL and its commissioner, Paul Tagliabue. Hundreds of Comcast employees from Maryland, Virginia, and Washington, D.C., have arrived to cheer them on. Comcast and the NFL announce that they will produce, every Sunday, extended highlights of all NFL games; all the important plays and some that are not so important will be stitched together into ten- to 20-minute packages. By Monday morning all the condensed games will be made available on Comcast's video-on-demand platform.

"Comcast's whole strategy," explains Roberts, "is to make our version of television more interactive and put the viewer in control. If you're a die-hard football fan, this is going to be terrific."

Roberts seems distracted during the NFL event, and later that day he calls to tell me why. He has been engaged in intense negotiations by phone with Sony and its private-equity backers as they try to cement their MGM deal. It, too, is about video-on-demand. Comcast entered the deal in its final stages by agreeing to develop new cable channels using the MGM and Sony libraries, which together have about 7,500 movies and 42,000 TV shows. (It takes only about 300 movies to feed a 24-hour cable channel for a month.) For its part Sony has agreed to cooperate with Comcast to make new movies available for video-on-demand. That is a breakthrough. Hollywood up to now has been loath to give new releases to cable when they get to the video stores for fear of eroding the lucrative revenues the studios now generate from video rentals and DVD sales. "The potential for the movies-on-demand business is enormous because it's a $10 billion rental market today," says analyst Craig Moffett.

Initially Comcast did not put any money into the Sony/MGM deal. But its willingness to create new channels from the library and to distribute them emboldened Sony and its partners to increase their offer and outbid rival Time Warner. Since then, Comcast has invested $300 million in the venture.

That was, as Roberts sees it, the latest affirmation of Comcast's bet-the-company acquisition of AT&T Broadband. "Why do they like us?" Roberts asks. "It not that Brian's a nice guy. It's 22 million subscribers. It's the same reason the NFL went with us."

The NFL and MGM deals are part of a Comcast game plan that had to be rewritten when the Disney bid failed. Comcast went after Disney for two reasons. The first was opportunistic--Roberts and Burke thought they could exploit Michael Eisner's weakness, acquire unique assets at a good price, and manage them better than Eisner and his people had. More than that, though, they wanted to use Disney programming to accelerate the growth of video-on-demand and personalized TV. "Imagine if we could have offered all of ABC, ESPN, and the Disney channel on demand," Burke says.

Without Disney, Comcast has had to build its content library bit by bit. (Pun intended.) Comcast has used its leverage to get content for video-on-demand included in all its new programming deals. For example, Viacom was persuaded during negotiations for MTV, Comedy Central, and its other cable properties to give the CBS Evening News, which it also owns, to Comcast for video-on-demand. A recent deal with Discovery Communications will include high-definition TV shows on demand.

What Comcast has been unwilling to do is pay extra for on-demand content, arguing that it already pays $4 billion a year to content companies for their programming. That is a mistake, the company's critics say. Cable operators invested heavily in programming during the 1980s to drive the industry's growth; they financed BET, Discovery, and Ted Turner's TNT, among other networks. Now, if they want to jump-start video-on-demand, they should pay for the rights to the most popular shows. Because Comcast won't, you can't get CSI or The Apprentice or American Idol on demand. So far, Disney and News Corp. have refused to license any of their broadcast programming for video-on-demand. As News Corp.'s Peter Chernin said recently, "Our view is that if people are using our content to build businesses, we need to be paid for it." That is hardly unreasonable. Of course, News Corp. won't go out of its way under any circumstances to help Comcast, which is DirecTV's biggest competitor.

Comcast could also do a better job of packaging and marketing video-on-demand. To be sure, the service is not easy to describe. (Look at how TiVo has struggled to market its DVRs.) But cable operators have never been good at marketing or, for that matter, at creating content. Their DNA is mostly about distribution and deals and what cable programmers refer to, derisively, as "dumb pipes." The creative types in the cable industry, people who create shows for MTV or HBO, respect Roberts and Burke for their business acumen, but they say the company has no creative sizzle. Comcast is led for the most part by cable lifers in suits. It doesn't feel like an entertainment company when you walk the halls.

I felt the need for a creative spark when I watched the NFL replays on Comcast. The concept is fine, and die-hard fans may not care about the presentation. But the replays consist of plays strung together, one after another, using the live audio from game day. There's no narrative, no drama, no fun, no resemblance to the half-time highlights on Monday Night Football once delivered in bombastic fashion by Howard Cosell or even to ESPN's SportsCenter on an average night.

Roberts responds that the people who count--Comcast's subscribers--are embracing video-on-demand. He has numbers to prove it. The NFL replays alone generated 600,000 viewing sessions in their first week on the air. In July, Comcast customers used their on-demand programming 51.6 million times, an increase of 20% from the previous month. In Philadelphia, where it has been available the longest, nearly 70% of Comcast's digital customers use it frequently, about 20 to 25 times a month. HBO, Nickelodeon, and PBS shows are the most popular so far.

Remember, too, that the video-on-demand business is still evolving. Time Warner, for example, charges even people who already pay for HBO and Showtime an extra $6.95 a month to get their programs on demand. Comcast, for its part, wants to get the technology accepted by the masses, reverse the loss of subscribers to satellite, and then "monetize"--yes, shades of the Internet again--the on-demand platform. How? Well, Comcast and a partner in Philadelphia recently introduced Dating-on-Demand where, for a fee, people can place video personal ads on cable. It's a hit with the audience too--in just the first month, viewers tuned in 287,000 times. Other forms of advertising are sure to follow. Imagine long-form commercials introducing a new car or touting vacation destinations. Comcast could charge advertisers that want access to their audience, perhaps on a pay-per-view basis. The possibilities are many.

THE BATTLE for the TV remote control will be led by generals like Roberts and Burke and Murdoch and Charlie Ergen. It will be won or lost by foot soldiers like Brenda Windsor. A 52-year-old mother of three, Windsor is one of 200 call center employees at a Comcast regional office in White Marsh, Md., just up I-95 from her hometown, Baltimore. I listen in for a while.

Windsor has an answer for every question that comes her way. "Yes, you can pause live television," she says to a woman who does not understand how a DVR works. She tells another, "As long as your TV is cable-ready, there's no charge for digital in the second room." Just as important, she chats comfortably with people--they are her neighbors. She asks one customer if she's heard that Olympic gold medalist and Baltimore favorite son Michael Phelps is getting his picture on a Wheaties box. She asks another if she's excited about moving to a new home, and they discuss the local schools. She gives everyone her name and phone extension and work hours, and invites them to call back if they have any problems. Her customers are charmed. "You have helped me so much," says one. "Most people don't even give you their name." One caller sent her Girl Scout cookies.

If Comcast could clone Brenda Windsor, it probably would not have had to launch an extensive, expensive companywide program called Think Customer First, as it did this year. Unlike AT&T Broadband, which outsourced its call centers, Comcast handles most of its customer calls in-house, and locally if possible. It built eight new call centers after the merger. New hires get five weeks of classroom training, followed by three weeks on the floor taking calls while a supervisor listens in. (It costs about $7 to handle each call.) Comcast's goal is to get 90% of calls picked up within 30 seconds.

People like Brian Lynch, who runs Comcast's cable systems in three suburban counties around Baltimore, have to make the Customer First effort work. Lynch, 45, is responsible for 850 employees, 350,000 customers, and a business unit with annual revenues of $350 million. He's been in cable since he graduated from college, beginning in a call center, then doing installations. He's seen the industry remake itself in response to competition. His systems, he says, now will schedule installations between 7 and 9 A.M. and 5 and 8 P.M., as well as on Saturdays or Sundays. "Meeting customer expectations is no longer good enough," he says. "We're going the extra mile." His compensation, which had been based on his unit's cash flow and subscriber growth, now includes a new metric tied to surveys of customer satisfaction.

AT&T tried to run its cable business like a phone company, with 4,000 people in a central office in Denver making most of the decisions. They were bureaucratic and too far from customers and employees. Comcast pushes decision-making into the field, when possible. It would like to exploit the fact that unlike the satellite companies, it has 67,000 workers scattered around the country. Lynch runs his Baltimore-area unit like a small business, overseeing sales, operations, local marketing, and even some aspects of programming. (He made sure Comcast offered Korean and Russian channels to reach the local immigrant population.) He'll march in a parade or arrange for a booth at the county fair.

Roberts and Burke have thought a lot about Comcast's culture. They wrote a customer-focused credo a few years ago, and they created Comcast University, which offers courses, in person and online, on topics ranging from Executive Leadership and Cable 101 to time management, team building, and e-mail etiquette. They promote lots of volunteering and community service. At company retreats, senior managers hear from business stars--Howard Schultz of Starbucks and Jerry Yang of Yahoo talked about managing growth; Bill Gates of Microsoft and Carly Fiorina of Hewlett-Packard shared their visions of an interactive future. A Wharton School graduate, Brian Roberts said his father was the one who really taught him to think about building a company for the long run. "My dad always thought of us as a little IBM," he says.

Financially, Comcast is starting to resemble a blue chip. After years of borrowing heavily to make acquisitions and upgrade its plant, its debt has dropped from a peak of $34.8 billion after the AT&T deal to $22.4 billion. It expects to generate about $7.5 billion in operating cash flow in 2004--an increase of 18% over last year--and $2 billion in free cash flow, after paying for all capital improvements, interest, and taxes. Comcast even produced net income of $327 million for the first six months of this year, a rarity in the cable industry.

Two recent transactions, one small and one big, have led some investors to take a fresh look at cable and Comcast. The small one involved Berkshire Hathaway, which disclosed that its Geico insurance subsidiary had purchased five million Comcast shares. (When it was incorrectly reported that Warren Buffett made the investment, Comcast stock rallied. The purchase was actually made by Geico, where Louis Simpson, a former Comcast director, is president and CEO of capital operations.) The big transaction involves Cox Communications, the nation's third-largest cable operator. The Cox family wants to buy the 38% of publicly traded Cox Communications that it does not own for about $8 billion and take it private. Obviously the family believes cable stocks are undervalued. "That's the best possible vote of confidence our industry could have gotten," Roberts says. "But the market doesn't see it."

The thing is, investors want more than a vote of confidence. They want results. In the short term--say, the next year or two--Comcast will be able to deliver financial results by selling more high-speed Internet access and rolling out telephony. Its fervent hope is to make those new products distinctive enough so that viewers will pay a premium for them. But Roberts and Burke recognize the possibility that over time, telephony and Internet access could devolve into low-margin commodities. (Their home city of Philadelphia is right now talking about providing free or low-cost wireless Internet access throughout its downtown.) This is why television remains the lifeblood of Comcast. Long before satellite came along, cable became a great business by providing viewers with a better television experience than they could get anywhere else. In that regard, Comcast's vision of the industry's future resembles nothing so much as its past.

FEEDBACK mgunther@fortunemail.com