When Pull Comes To Push

Los Angeles Times, November 25, 1996. By Steve G. Steinberg

When Marc Andreesen and his team at the University of Illinois launched Mosaic into the world in 1993, thereby setting off the World Wide Web boom that continues today, the response from Internet old-timers was predictable. "It's going to turn the Net into TV," they would remonstrate into their beards.

It was a prediction that was based on two fears. First, that the Web would make the Net so easy to use that the great unwashed masses would invade what had traditionally been the province of elite techies. Second, that the Web's support of sound and graphics would lead to an emphasis on style rather than substance.

Both these fears have come true, but the Net is still not TV. The fundamental difference remains: A user must go to the Net, but television comes to the viewer. It is the difference between "push" and "pull" media. The Net, whether the Web or some other service, is like a giant library--the user can browse the stacks and pick up whatever interests her. But TV is a push medium: Once it's turned on, it controls what you see next.

I'm describing this distinction because it is about to disappear. A number of products have been announced in the last few weeks--including Pointcast 2.0, Marimba's Castanet and Netscape's Constellation--that seek to turn the Net into a push medium. These new systems allow content providers to send information and images directly to users instead of waiting for users to come to them.

The reason for this development is, quite simply, advertising. The last two years have shown that the Web is not terribly fertile ground for advertisers. After all, when the users are in control of what they see, chances are they aren't going to steer toward ads. Turning the Net into a push medium puts the content providers--and hence the advertisers--back into the driver's seat.

This, by itself, doesn't mean that products like Pointcast and Marimba are necessarily bad. (I'd much rather have advertisers pay for what I read than have to pay for it myself.) Even the fact that these products will, essentially, "turn the Net into TV" doesn't worry me. But when you closely examine this new breed of Net services, a host of dangers pop up. And these do worry me.

The granddaddy of these push media services is the Pointcast Network, which has been available to PC users since February. Anyone can download the Pointcast application for free and install it as their screen saver. Then, whenever the computer is idle, Pointcast will display news stories, weather reports and--most conspicuously--animated advertisements on the screen. A continually updated stream of these tidbits comes over the Net from Pointcast's server in Cupertino, Calif.

Pointcast claims that 1.5 million people have signed up for their service, but it's not clear how many of them really watch it. After all, there is something decidedly counterintuitive about Pointcast's approach. As Tim Barkow, a section editor at Wired magazine jokes, "Pointcast waits until it is absolutely sure nobody is near the computer, and only *then* does it start putting stuff on the screen!"

A more serious issue is how much weight people give to news that appears as little more than a scrolling headline. Watching Pointcast, I find myself more interested in the strange poetry created by the rapid succession of headlines than in any particular story. The headlines blur together, and I'm left convinced that with so much happening nowadays, it's a good thing that I have a computer to keep track of it all.

Contrast this with the process of getting the news off the Web. Here, I first decide which news site I want to go to, and then as I scroll through the stories at my own pace, I follow the hyperlinks that interest me until I find myself deep in a history of Yemen. I may end up feeling as if I'm lost in an infinite labyrinth, but I learned something from the journey.

Pointcast also brings to the fore a long-prophesied danger of computer-delivered news: the loss of serendipity. Because Pointcast allows me to customize what kind of news I want to see--only articles about technology and the Green Bay Packers, for example--it's easy to become insulated to what's happening with the rest of the world. This kind of tunnel vision is in sharp contrast to Web surfing, where random-seeming connections and unexpected juxtapositions are what keeps things interesting.

I consider both these problems to be serious shortcomings for Pointcast. But from a purely business viewpoint, what has held Pointcast back from mass market success is its closed architecture. Pointcast only provides Pointcast content. It would be as if Zenith made TVs that only carried the Zenith channel.

This is why Silicon Valley venture capitalists are currently so abuzz about Marimba, a start-up company that has developed an "open" version of Pointcast.

Marimba was founded by four members of Sun Microsystem's original Java team, and their first product--introduced last month--is called Castanet. This cleverly named piece of software is a platform that allows developers to quickly and easily broadcast content to users.

Marimba describes how the scheme works by comparing it with broadcast radio. A content developer uses the Castanet transmitter software to send text, images or software. At the other end, users run the Castanet tuner to select the desired channel, which appears on their desktop.

This method offers some clear advantages over Pointcast. By opening up an infinite number of channels, Marimba's scheme restores the diversity and heterogeneity that makes the Web so interesting. But Castanet still retains the flaws of all push media: It encourages rapid pacing instead of thought-provoking depth, and it encourages tightly focused "narrowcasting."

Perhaps it should come as no surprise that one of the first developers for Castanet, MGM Interactive, is working on a multimedia soap--the Net really is becoming like TV.

Marimba's openness also exacerbates another danger with push media: consumer privacy. Because Marimba channels are sent to identifiable addresses and can transmit active Java code, it becomes the perfect vehicle for tracking and monitoring users. Forget about the anonymity of Web surfing--now advertisers can keep track of exactly who is looking.

These dangers of push media are worrisome. They destroy the depth and serendipity of the Web for the sake of advertisers' needs. And the trend toward push media is showing no sign of abating. If you need further proof, just look at the two industry giants: Netscape and Microsoft. Just last week Netscape announced a new product called Constellation, which will interoperate with Pointcast and Castanet in order to bring content directly to subscribing users. And Microsoft has been making very similar noises, hinting about an "active desktop" extension to Windows.

Sure, pull media has problems of its own. It's often cumbersome to search through the morass of data available, and it can be a horrible time-sink. But we need to be very aware of what the trade-offs are between these two kinds of media so that, when push comes to shove, we choose the one that meets our needs.