Wired visits one of the first interactive TV tests - in TCI's backyard in Colorado - and discovers that the only problem with the idea that movies-on-demand will drive the creation of the infobahn is, well, the demand.
This is the best job I've ever had," Fran says, with giddy enthusiasm.
Sporting shiny reddish hair, oversized eyeglasses, and a fresh pair of sneakers, Fran is the video-on-demand gopher for the sleepy burb of Littleton, Colorado, entrusted with keeping her eye on a big computer monitor hanging from the ceiling. When an order for, say, Coneheads, comes in, she rushes to the giant videotape library on the far wall, retrieves the correct title, and hurries to insert it into the appropriate place in a gleaming bank of VCRs. During prime-time hours, two people share this job. But right now, on a Wednesday afternoon, it's all Fran.
Fran may or may not realize that her job category will soon be phased out. Her employer, cable colossus Tele-Communications Inc. (TCI), is conducting this market test with AT&T and US West. But all the other interactive television trials just getting underway are powered by computers, not by people like her. This particular test, called VCTV (for Viewer-controlled Cable Television), happens to be decidedly low-tech. The only point of this trial is to gauge what consumers want - and what they will pay for when all the technology is ready.
Two and a half videos per month. That's the golden nugget of this entire market research effort. Since the VCTV test began in July 1993, the 300 households participating in the trial here in Littleton have each ordered, on average, between two and three movies every month. "That's twelve times the national average for regular pay-per-view," boasts Jeffrey Shomper, VCTV's marketing coordinator. He's comparing the finding with traditional pay-per-view, in which viewers dial an 800-number and order from a narrow choice of movies shown every hour or so. According to the Pay TV Newsletter, published by Paul Kagan Associates Inc., the figure is only slightly lower than twelve times the national average, with typical households ordering just 3.5 movies per year.
To find this out, TCI, AT&T, and US West have chipped in more than US$10 million.
But considering how much these 300 families have been prodded, prompted, and probed, this 2.5-per-month figure is less than impressive. To recruit participants, VCTV sales reps went knocking on doors in Littleton on up to three separate occasions. If that wasn't enough, flurries of telemarketing calls and direct mail pieces with pictures of Kevin Costner were supposed to do the trick. And testers waived the monthly subscription fee for the purposes of the market test. Viewers only pay on a per-movie basis, usually about $3 or $4 a shot. Once 300 households accepted the service, they were also given three "barker" channels showing the trailers for the latest video releases around the clock. "We are hitting these customers pretty hard," says Shomper, in a bit of an understatement.
That's only the beginning of the interactive fun. Half of the participants' TVs are fitted with electronic Nielsen monitors allowing analysis of viewing decisions the families make; VCTV also tracks viewers' tastes using its own system. If someone shows a hankering for any of 19 different movie genres, the system takes note; those watching Westerns, for instance, immediately start getting hit with direct mail touting all the John Wayne, Clint Eastwood, and Gene Autry movies they could bear.
Market research purists might call this "polluting the experiment." No market-driven company in its right mind is going to spend this much money and effort recruiting and tormenting millions of customers. In the real world, the average household might never want to order 2.5 movies per month with a remote control. And even if it did, would that generate sufficient revenue to build an entirely new, digital, interactive television infrastructure?
That is one of the cable industry's most persistent and vexing questions.
Watching any movie, any time - video-on-demand - is supposed to be the killer app that propels dozens of new interactive TV services into American homes by the end of the century. The home shopping, the custom news programs, the play-along game shows, the dozen or so simultaneous football games, the home banking, the on-demand Roseanne, the whole 500-channel scenario that cable companies have been hyping - all of this is supposed to piggyback on the raging success of movies-on-demand. But there's one glitch: there's not much demand for movies-on-demand.
No matter. Market leaders are charging ahead, the less-than-stellar results from TCI's test notwithstanding. The trials by Time Warner in Orlando, Florida; by Cox Communications in Omaha, Nebraska; by Bell Atlantic in Alexandria, Virginia; by AT&T and Viacom in Castro Valley, California; and by TCI and Microsoft in Seattle are all set to start by early 1995. All of these companies have developed unique ways to replace Fran - they scan thousands of films into digital format and load them onto massively parallel computers. These video "server" machines will be linked by fiber-optic and coaxial cable to the newfangled set-top boxes on the TV sets in the homes of test customers.
Meanwhile, all these companies are aware of an extreme version of the 80-20 rule. It's not as if 80 percent of the viewers want to watch 20 percent of the thousands of titles that are available. It's more like 95 percent of the viewers want to watch the same five movies at any given time. That's just the way the industry works. When Mrs. Doubtfire is released on video, almost everyone wants it that evening.
Computer scientists have tried to come up with a way to accommodate this requirement. The goal is to allow as many as 10,000 people to watch the same copy of a popular movie at 10,000 different start times. Just as on conventional TV or in theaters, movies stored on computers must be shown at a rate of 30 frames per second. Because the computer can grab the video at rates up to 240 frames per second, in any given second eight viewers can access the same portion of the memory.
But eight is not enough, especially when Jurassic Park comes out on video. A technique called "memory striping" increases viewership by chopping up each movie into, perhaps, a thousand small snippets. And each snippet is automatically placed on a different portion of the memory. This way, eight thousand viewers can simultaneously watch the same small segment of a movie. When a viewer finishes with a given snippet, special software knows how to jump seamlessly to the next one, keeping the movie on track. The software could even let viewers rewind, fast-forward, and pause the video, as with a VCR.
This all works in laboratories at least. No one has shown that this can work for a large group of fussy viewers. And even if the technology can operate as planned, there are basic limitations built into it. A single video server can handle the viewing demands of about 30,000 households, according to Ben Linder, director of technical marketing for Oracle Corp.'s Media Server software. Oracle is supplying video database technology to Bell Atlantic and other companies conducting trials. The software giant impressed cable companies earlier this year with claims that it could supply the technology for approximately $500 per household. Some industry analysts don't believe it. Taking all the costs into account, many analysts estimate that it would cost $30 million to set up a typical 30,000-home network (about $1,000 per household). Time Warner's Orlando system, meanwhile, is reportedly costing $5,000 per viewer. No matter whose figures you use, the conclusion is the same: households will have to spend a lot more than what it costs for 2.5 movies per month for cable companies to break even on this stuff in a reasonable amount of time.
The technology also has some stiff and obvious competition. The corner video store seems anything but an endangered species these days. Indeed, the video rental industry shows no sign of retreat. A full 80 percent of American homes have VCRs. In 1993, these households spent record sums, renting 3.2 billion videos - an average of nearly one per week. There are roughly 28,000 video stores in the US, each serving an average of about 2,600 families. Although Blockbuster and a few other chains have dominated the business press a lot, most of America's video stores are independently run. These mom-and-pop operations are an entrenched part of neighborhoods all over the country.
Believe it or not, many people actually enjoy visiting them. And walking through a typical video store is actually a highly efficient way to search for what you want. Movies are all categorized and displayed in full color. When you select one and bring it home, you are actually transporting more than 100 Gbytes of data with you. Even if the new release that you want to see is out of stock, it's not that big a deal. Most people can wait a week or two. If it wasn't important to see something in the theaters, it isn't so important to see it the minute it hits the video stores.
The inhabitants of Littleton seem to agree with this. Even during a particularly harsh winter, even with Fran-on-demand delivered right to their living rooms, many of the 300 families in the VCTV trial continued to go to video stores. The market research showed only a slight decline in video store rentals among these families. Cable companies may yet build some sort of interactive infotainment superpipe. But they will probably have to come up with some other killer app, some other reason for people to subscribe to a new generation of cable services.
"Everybody is talking big," says one telecommunications analyst. "But no one has proof that there will be enough demand to justify the costs."
Never before has so much investment and hype and ingenuity gone into such a trivial task as replacing the video store. No one knows this better than Fran. When told she will be replaced by computers one day soon, Fran just shrugs and smiles, her eyes trained on a screen that, at the moment, shows not a single request.
