This time 12 months ago, I was writing a column of high tech predictions for the new year. That's just one in a long list of formulaic columns I've used again and again over the years — Christmas, election day, the Burning Man, St. Swithen's Day. You know, the usuals. But last year's effort had the unintended consequence of some reader actually paying attention. Months later, he wrote me an extensive analysis of my predictions and their outcome, declaring me to have been 60 percent correct. I'd say it was more like 70 percent, but even 60 percent is good enough to win an NFL head coaching job. Still, the experience chastened me a bit, so this year, I'm avoiding lists of any kind. I'll still make predictions, but in the cagier form of an essay.

Some predictions are easy, like fewer computers being sold in the first quarter of 1999 than in the last quarter of 1998, or like the further consolidation of the networking market now that Lucent is free to buy companies. Look for the Lucent-Cisco rivalry to heat up with the greatest effect being on 3Com, which might be driven into the arms of a suitor.

Apple will continue to improve with the introduction of System 8.6 and MacOS-X Server. This latter product, if it is relatively bug-free, could make some major waves in the office networking arena as an easier-to-use alternative to both Netware and Windows NT, though I doubt that it will have the impact of Linux.

The trend toward ever cheaper PCs, which accelerated this week with the introduction of eMachines' $399 boxes, will continue, moderated only by Microsoft's unwillingness to drop the price of Windows 98. There is a really significant trend here — more significant than a lot of people realize. I am convinced that the price of an entry-level PC will within five years converge on zero. That is, PCs will be free. This is exactly analogous to the free cellphone. Cellphones can be given away because their providers make money on the airtime. So too, as the Internet and online entertainment become more and more central to the lives of the greater mass of Americans, it will become worth it to give away PCs. This only makes sense in the era of the Internet.

And of course, the government still won't have a clue how to regulate, tax, or generally control the Net, though that won't stop them from trying.

But the most important prediction I have for 1999 is likely to be a bit of a disappointment at first. It's what truly ought to be a multi-year prediction — an important trend that will start in 1999 and roll like a juggernaut through succeeding years until it has completely changed both computing and entertainment forever.

I wrote awhile back about how the rise in importance of the Internet has led to what's essentially an acceleration of Moore's Law. Not only does the steady and predictable increase in processor power mean that computers can do more year-to-year, but the application of that same trend to network equipment means we can have ever faster communication channels for the same price. Add the widening pipes to the growing ability of our computers to compress and decompress data to flow through those pipes, and we come up with total throughput numbers that look like Moore's Law squared, which is scary. But I couldn't really think of what we'd do with all that communicating power until now.

One of the most significant computing advances of the 1980s was the development by Adobe Systems of the PostScript page description language. PostScript made really good desktop publishing possible, and revolutionized magazines and garage sale fliers in the process. But PostScript also gave us an archetype for what is about to start happening in home entertainment.

Before PostScript, the only way to have a computer print attractive fonts was to define those fonts as bitmaps and shoot them over the printer cable one pixel at a time. That's a lot of bits, and literally required the invention of Ethernet to make even remotely practical in higher resolutions. Print could be beautiful, but it couldn't be easily scaled up or down. PostScript changed all that by separating the information, the words, from their representation, the font. PostScript is an interpreted language that runs in the computer we call a printer. First you download the chosen font family descriptions to the printer, then you download the words to be written in that font. Each character is downloaded to the printer only once, and even then, it is downloaded not as a bitmap but as a very simple description of how to draw (render) the character. With PostScript, your computer literally tells the printer to render particular text in a particular style. It takes very little bandwidth, but a lot of processing power. Whether the printer prints at 300 dots-per-inch or 1,200 dots-per-inch, it does so from the same input file, which doesn't grow. This is called resolution independence.

Jump now to the world of electronic home entertainment, which is horribly bandwidth limited. Watch some video over the Internet and you'll see: It sucks. Those tiny little jerky pictures aren't worth watching if you ask me. But then they look a lot like experimental TV pictures from the late 1920s, so I guess you have to start somewhere.

The trend I see follows from what I've already described, and has to do with the product of rendering speed and bandwidth. We all know that network bandwidth is improving daily, but how much does it have to improve for Internet video broadcasts to look like television broadcasts? It has to improve a LOT. ISDN (128 kilobits-per-second) bandwidth seems to be good for a jerky version of quarter screen video at 20 frames-per-second. If we want full-screen video at 30 frames-per-second that requires eight times as much bandwidth, or one megabit-per-second. Hey, ADSL will do that! But remember, this is for a VHS quality picture, not full broadcast quality. Still, if we can get enough ADSL circuits installed, it looks like the Internet could give cable TV a run for its money.

Yes and no. The Internet probably will become a good alternative to cable for people who are willing to use it like cable — to accept a single broadcast that goes out at a specific time and is received by millions of viewers. This IS cable TV, just cable over IP. But that's not the strength of the Internet, which ought to allow every viewer to choose his own content, to define his own cable channel. I'll be watching the Perry Mason Channel, myself. Alas, there won't be enough bandwidth on the Internet to allow a channel for every viewer for years and years.

Do the calculation. Let's say we have 100 million Internet users in a year or so and they all have ADSL connections, allowing downloads of one megabit-per-second. Internet Service Providers all install one modem for every eight customers right now, so let's assume that an eighth of these users are online at any one time. That's 12.5 million megabits-per-second of downloading, which is no sweat if everyone is watching "Seinfeld" at the same time, but a pain in the butt if everyone wants to watch something else — and at their own time, too. Start talking about high definition video signals, and it gets even worse.

But there is an answer to this problem, and it goes straight back to that mysterious prediction I started to make awhile back about something that, this year, would start to change the future of home entertainment. It has to do with an impending revolution in rendering. Today, the gold standard in rendering price/performance is found in the video game business, where $200 boxes from Nintendo, Sega, and Sony can draw around five million 3-D polygons per second. We care about 3-D because it is a physical world we are generally rendering. A year from now, the consumer rendering rate will have grown to a Moore's Law-busting 50 million polygons-per-second. And that will change everything.

Think of how digital video is stored on your DVD drive or transmitted to a DSS satellite receiver. The base technology is some version of MPEG compression. MPEG compression divides the screen into blocks that are eight pixels on a side, uses a discrete cosine transform to compress each block into as few bits as possible, then uses buffering and motion estimation to transmit only the blocks that change from video frame to video frame. It's complicated, but the truth is that MPEG is just a bit map with an attitude. It still takes a LOT of bandwidth to carry a broadcast-quality MPEG signal.

The ability to render 50 million polygons per second instantly means that such consumer devices can render whole video frames in real time. Just as PostScript used rendering to eliminate the bandwidth requirements of laser printing, rendering can eventually remove the bandwidth limitations of MPEG. We can use this rendering power to go from describing a frame to describing a scene. Instead of describing the placement and movement of a million pixels, we can say, "Render the 'Today Show,' with Katie Couric on the sofa. Here are her lines."

Where PostScript is a page description language, this future I am describing requires a multimedia description language. The result, like Web browsers that each render HTML slightly differently, will be some variation in the way scenes look. But maybe that's an advantage. Maybe I want MY Captain Kirk to be black or Chinese. Taking it even further, maybe I can define my own "Star Trek," "Seinfeld" or "Jerry Springer" episodes. This is the future, and it doesn't in any way negate the work that is being done right now on the Internet or in digital TV. Bits are bits, and anything we do to improve the pipes will help us all.

From a business standpoint, who will benefit most from this revolution that will take five to 10 years to complete? Put your money on the vertically integrated players. Anyone who has content, does distribution, develops software, and makes rendering hardware will win in four ways. That looks like Sony to me.