Interesting People mailing list archives
IP: a tin can and a wire [see note]
From: David Farber <farber () central cis upenn edu>
Date: Sun, 15 Oct 1995 20:12:25 -0400
**** Why is it the Economist "almost" gets it right so often yet almost always misses the real point. They did it yet again djf *** Date: Mon, 16 Oct 1995 00:45:07 +0100 To: www-buyinfo () allegra att com From: anon-remailer () utopia hacktic nl (Anonymous) The Economist, October 14, 1995, pp. 75-76. Will your next computer be a tin can and a wire? This week's fall in technology stocks was bad enough. But what if the Internet destroyed the personal-computer industry ...? For nearly a decade, Sun Microsystems, a Californian computer maker, has mystified punters with its cryptic slogan: "The Network is the Computer." Surely computers are the boxes that do all the work, and the network is just the wires that connect them? Sun's riposte: just wait. Now a growing camp within the computer industry thinks the wait is nearly over, for two reasons. The first is the inexorable rise of the Internet, the open computer network already connecting some 30m-40m users worldwide; the second is the emergence of Java, a new kind of program that runs over it. As well as Sun, firms such as Oracle, a software giant, and Netscape, the darling of the booming Internet-software industry, are spending tens of millions of dollars on a bet that the Internet can do a lot more than pass around e-mail and transmit data. Along with computer visionaries such as George Gilder, they think it can also do much of the work of today's computers, holding not just information but software, from word processors and spreadsheets to games and entertainment programs. Most radically, they go on to argue that this could end the reign of the personal computer. If small, efficient programs can arrive speedily from the Internet, who needs a $2,000 PC on their desk? Perhaps a glorified terminal -- just a screen and a processor -- would work just as well, at a quarter of the price. This is a tempting idea. PCs are not just expensive, they are also caught in a vicious circle. Each time Intel, a chip maker, launches a faster processor, Microsoft, a program writer, releases fancier software that sucks up most of the new power. Diligent users trade up every other year, but their word processor probably runs more slowly than it did in 1985 (although it does a lot more). They have little choice: Microsoft and Intel may be benevolent dictators, but they are still dictators. It is this stranglehold that Sun and others believe they can break. If they are right, it would reshape the whole computer industry, returning it to its early days of buccaneering competition. Neither Intel nor Microsoft has any special advantage on the Internet (other than having more money than anyone else to throw at it). That has not hurt them much so far because, in its slow and stumbling infancy, the Internet is still a mere adjunct to the PC. But some, at least, of the elements required to turn it into a rival are starting to fall into place. The first is the all-important fast connection. Many businesses now have fast Internet connections over leased lines; in a few years, many homes could get the same service through upgraded cable-TV networks and digital satellites. Today, many home users struggle with an Internet connection at 14,400 bits per second. Tomorrow's networks will offer 10m bits or more. But even using such big pipes, today's Internet is still a static medium, hardly more interactive than a telephone. Most of the sites on its multimedia World Wide Web simply send pictures and text in response to requests. What it lacks is software: working, useful on-line tools that users can control -- just the sort of thing that spurred the PC'S success in the first place. That is where the second new element comes in. Earlier this year, Sun launched Java, a new programming language (the code in which software is written). Unlike other computer tongues, it is designed especially to run on a network, by using "applets" (small programs designed for specific jobs). Sun reckons that programs written in Java could let users do everything (and more) they do on their PC, all from the network. Already there are nearly 400 Java applications, including spreadsheets, wordprocessors and games. Software on demand Java teasingly suggests the holy grail of computing: programs that do exactly what you want, when you want it -- but no more. Today's multimegabyte PC-software packages include every power-hungry feature their designers can think of, on the off-chance that a user might occasionally want a few of them. No wonder Oracle's chief executive, Larry Ellison, calls the PC "a ridiculously over-engineered device -- a mainframe on the desktop". Sun's vision for Java is that its compact applets, many taking up less than 100,000 bytes, will do a single job well. If a user wants another feature -- say, spell-checking on a word-processor, or a graphic chart -- he simply clicks to fetch another applet, which arrives in a few seconds. Java thus offers the user the tempting prospect of a virtually infinite supply of just-in-time software -- passing the burden of storing it to the network. Today the Internet's content is like that of a TV: a mostly static information and entertainment store. Users move around it with a "Web browser" programme. Add Java (or its equivalent: Microsoft and others are working on similar languages) and such browsers become workshops too. Both Netscape and Sun have piloted versions of their browsers with Java "interpreters" built-in. Oracle is developing a similar product. Mr Gilder thinks such efforts will shake the computer industry: "To the extent that Java or a similar language prevails, software becomes truly open for the first time." Forget Windows 95; some people are starting to wonder if they need Microsoft at all. They may soon get their chance to find out. Sometime next year, predict both Mr Ellison and Sun's chief executive, Scott McNealy, companies will release machines (dubbed "Internet appliances" or "network computers") that will do nothing more than run Internet software and perhaps a simple word processor. Sans Microsoft, Intel, hard drive and almost everything else, and equipped with a network connection and perhaps one megabyte of memory (compared with eight or more on a typical PC), such Internet computers will probably sell for less than $500, aimed initially at the vast majority of homes that do not yet have a PC. Some who might make such devices are Oracle (perhaps working with Korean electronics giants such as Samsung or the LG Group), and game-machine companies such as Sega or Philips (whose CD-i game box can already access the Internet). Apple is developing its own stripped-down computers and Toshiba is working with Sun on a mobile Internet machine. A steaming cup of reality No wonder Sun's share price has tripled in the past 16 months. Specialised network computers may indeed catch on: after all, dedicated game machines still outsell PCs in America, even though the bigger computers can run many of the same games. But those who predict that such machines will kill the PC are ignoring computing history, and glitch-prone real life. The PC beat the mainframe because users wanted the whole computer on their desktop, not in the basement. That makes Java terminals look like a step backwards: by putting program storage far away down a shared network, it makes it vulnerable to delays, congestion, and all the unpredictability of anything out of a user's control. "Anything that happens in a box on your desk is always going to be faster than something that happens down a wire," says Craig Mundie, head of Microsoft's consumer-systems division. As network capacity increases, user demand is sure to increase just as fast, reinventing the vicious power-sapping circle of the PC -- every growth in "bandwidth" consumed by more ambitious multimedia data and programs. Moreover, many homes will not get fat data pipes for years, if ever. Without them Java and the Internet in general are reduced to simple communications, lightweight jobs, and a few pictures. Intel's chief executive, Andy Grove, thinks that makes his company safe: he calls the notion of high- bandwidth communications reaching every home "a fantasy". Even without network delays, Java programs will by nature tend to run more slowly than traditional software because they must, for security reasons, be "interpreted" line-by-line by software in the terminal. Finally, there is the question of where all these splendid Java applications will come from. Mr McNealy thinks that "three computer-science students from Berkeley hacking code late at night" will create a powerful word-processor, which they will "give away for free because they want fame first, knowing that will lead to fortune." Perhaps, but perfectly good word processors are already free for the taking all over the Internet, and yet people still pay more than $100 for the features and standards of Microsoft's Word. "It's not likely that someone is going to build the functionality of Word one component at a time," says Mr Mundie. If dedicated network computers cannot do what people want their PC to do today, and if they are slower in what they can do, the PC is safe. Being cheap is not enough: although you can buy a computer for only $900 today, you probably won't. When it comes to computing, performance is everything. That does not mean that there is no market for millions of Internet appliances, only that it is not the PC market. Such machines could well be the TV to today's PC, bringing information and entertainment into the home, rather than bumping the PC from the working desktop. This could be a huge market -- and one that Microsoft and Intel could indeed miss out on -- but it is an addition, not a replacement. [End] $0$AD
Current thread:
- IP: a tin can and a wire [see note] David Farber (Oct 15)