IP: HALLOW DREAMS

[Dave Farber <farber () cis upenn edu>]

Sorry for the bad formatting, no time to fix djf


(Sent with permission)
SAN JOSE MERCURY NEWS
Copyright 1995, San Jose Mercury News
 
 DATE: Sunday, November 12, 1995
 PAGE: 1D                                EDITION: MORNING FINAL
 SECTION: BUSINESS
 SOURCE: Story by Lee Gomes 
 MEMO: ANALYSIS
 
HOLLOW DREAMS
THE INDUSTRY IS ABUZZ WITH TALK OF A LEAN, MEAN 'HOLLOW' PC. BUT DOES THE
TALK REFLECT STRATEGIC THINKING...OR WISHFUL THINKING?
 
    AFTER 15 years of bulking up, technologically speaking, are personal
 computers about to go on a diet? And not just lose a few pounds, but slim
down
 so much that they become altogether ""hollow?'' 
    That's the premise of a new and currently fashionable school of thought
in
 the computer industry, one that's adhered to by some of the industry's
biggest
 names and by a legion of Wall Street investors who, whether they realize it
or
 not, now have billions of dollars riding on the idea that the future of
 technology is a hollow one.
     This view, which has not persuaded everybody, holds that a new chapter
has
 begun in the PC saga, one in which the devices will lose their dependence on
 sinewy microprocessors like Intel's Pentium and large operating systems like
 Microsoft's Windows. In their place will be ""network computers'' costing
$500
 or so, that will connect to a new, improved Internet - which itself will
grow
 to become the source of our computer power and our computer programs.
    In its purest form, this vision holds that applications such as word
 processors won't reside on a local hard drive - which, in fact, we may not
 even have at all. Instead, programs will be stored on some distant server
that
 we'll link to with a high-speed connection.
 PCs, the argument goes, will become ""hollowed out,'' serving not as the
 foundation for all computing, but instead as a simple  entry point to an on-
 line world stuffed with all the resources anyone could want.
    Many people consider this ""hollow talk'' to be but wishful thinking by
 companies shut out of, and increasingly threatened by, the dominant
""Wintel''
 computer architecture. And indeed, the notion of hollow computing seems to
be
 in conflict with most of the trends in technology during the last few
decades
 - like the one toward having individuals, rather than remote machines,
control
 their own computing destiny.
    Others point out that the recent history of personal computers is full of
 still-unfulfilled predictions of the imminent collapse of Intel and
Microsoft
 - such as the notion that Unix workstations would one day replace PCs, or
that
 desktop computers would be eclipsed by handheld ""personal digital
assistants'' or super-smart ""high definition'' televisions.
    Still, the talk abounds.
    For example, Oracle Corp. President Larry Ellison, says PCs have gotten
too
 complex and hard to use and are overdue for a replacement. Ellison promises
a
 $500 ""network computer'' next year.
     Another big booster is Sun Microsystems Inc., developer of ""Java,'' a
new
 version of the C++ computer language that Sun has reworked for use on the
 World Wide Web, allowing a browser program to execute small programs known
as
 Java ""applets.''
    While Java is one possible approach to livening up Web pages - for
example,
 by animating a company's logo - Sun is making much bolder claims for it.
Java,
 says Sun, could become nothing less than the future of computing; the lingua
 franca that programmers everywhere will use for nearly all computer
 applications, all of which will run on-line.
    In recent months, various industry analysts have warmed to the hollow
idea,
 giving talks with titles like, ""Why Microsoft and Intel don't matter any
 more,'' or calling PCs the ""horse and buggy of the Information Age.''
   The futurist George Gilder, in a widely read essay this fall, called Marc
 Andreessen of Netscape Communications Corp. ""the new Bill Gates'' and
 predicted that soon no longer will the programs in your machine determine
the
 functions you can perform. The computer becomes a peripheral to the
Internet.
   On Wall Street, the theory of the hollow PC provides the theoretical
 foundation, such as it is, for the current investor mania for anything and
 everything connected with the Internet.
    Shares in Netscape, for example, closed at $97.50 Friday.  A common
 investor refrain is that Netscape is ""the Microsoft of the on-line world,''
a
 view that borrows many of the assumptions of the hollow computer camp.
 On-line overwhelms
    Those assumptions begin with several undisputed observations: that
 communications-related tasks, as opposed to those involving solitary
 computing, have become an increasingly important component of what PCs do;
and
 that PC users are spending more and more time on-line.
    The disagreement involves what happens next.
  *Will those communications functions eventually become the entire
 computer, as those in the hollow camp contend? Or will they simply be
another
 in the PC's increasingly impressive repertoire of skills, a repertoire that
 includes activities as diverse as sending faxes and playing CD-ROMs?
 * Will a Web browser become the operating system? Or instead, will the
 operating system simply expand, as it has in the past, but this time to
 swallow browser functions as well?
 * Will a minimalist computer really be easy to use? After all, computing's
 user-friendly features, such as graphical interfaces, require enormous
amounts
 of hardware and software, albeit carefully hidden from users.
    It should also be noted that history is not on the side of those
advocating
 hollow computers.
    For one thing, special purpose computing devices have always lost out to
 general purpose machines like PCs, - especially considering the constantly
 declining prices of PCs. Given a choice, who would pick one machine that
does
 less than another one?
   For example, five years ago, people were making many of the same
predictions
 for X terminals, a kind of slimmed down but network-aware workstation, that
 they are today for network computers.
    But Eileen O'Brien of International Data Corp., the research group, said
X
 terminals never became anything more than a niche product, largely because
 they couldn't match PCs.
    And in recent years in the consumer marketplace, numerous companies have
 brought out what were essentially souped-up game machines with limited
PC-like
 functions, such as the ability to play some CD games. Despite elaborate
 marketing campaigns, these products all fizzled - even while millions of
 consumers continued buying PCs three or four times as expensive.
    Another problem for hollow computers is that they seem to go against one
of
  the cardinal tenets of computing: that no matter how much microprocessor
power
 or memory or storage space you have, it is never enough.
    Every year or so, some new PC technology comes along that causes the
 machines to swell up. A few years ago, it was CD-ROM-based multimedia; next
 year, it is likely to be the simultaneous moves to 3-D computing and full-
 screen video - both of which depend on ever-more powerful hardware and
 software.
    Since those sorts of capabilities would be on the wish list of even an
 entry-level PC buyer, it is unclear who would be attracted to a hollow 
 computer. Especially since the mass market economics of the PC industry
allow
 companies to continue to add features to machines without markedly
increasing
 their cost.
    Finally, a computer that depends on a network like the Internet must cope
 with the fact that the Internet has some major problems of its own. For all
 its riches, the system is also quirky and obtuse; it is often slow,
something
 likely to get worse as usage increases; and has far more breaking points
that
 can cause things to shut down than any individual PC ever will.
    But Oracle is undaunted.
   While it is careful to note that it won't actually be selling the machines
 itself, the company says it will have a network computer to demonstrate by
 early next year, and that other companies will be selling them by year's
end.
    For $500 or even less, the company says, you could get a machine with a
 RISC processor, 4 megabytes of memory, a keyboard and some sort of network
 interface, such as a Ethernet connection or a built-in modem.
    But why, precisely, would anyone want to buy one? Especially when they
 could get a more powerful, if used, PC for the same price?
    Because, said Andy Laursen, Oracle's vice president of network computing,
 of their utter simplicity.
    ""The simplest PC is still beyond the skills of most of the population,''
 he said. ""A network computer will be as easy to use as the telephone.''
    ""People will still buy PCs, but at some point, the growth rate for
network
 computers will surpass the growth rate for PCs. People might have one PC,
but
 they will have several network computers,'' Laursen said.
    The machines will perform specialty functions, he said. One might be a
set-
 top box for a TV set; another will be dedicated to Web browsing.
 Good for anything?
    But while the sort of machine that Oracle is describing may be easy to
use,
 can it be used for anything that's really useful?
    For starters, the most rudimentary ""network computer'' will use a home
 television set as a monitor, meaning a low resolution display that can't do
 justice to most Web pages.
    Another major issue is whether a stripped-down machine can keep up the
 rapidly changing ways in which people actually are using the Internet. In
 fact, it's likely that rather than making computers slimmer, the growth of
the
 on-line world will be yet another reason that PCs will become even more
 corpulent.
    At the start of the year, for example, Web pages were relatively simple
 collections of text and pictures. But they are slowly adding sound, video,
 telephony - and now with Java, full-blown computer programs of their own.
 Those all place added demands on the computer that's doing the Web surfing.
    Laursen maintains that without the burden of having to run a large
 operating system, a 4 megabyte machine can handle everything. But that
notion
 is disputed even within the ""hollow PC'' camp itself.
    For example, Netscape founder James Clark has professed skepticism about
at
 least the hardware portion of the hollow computer idea, saying that
Netscape's
 Navigator browser program needs all of the performance ""oomph'' that
today's
 PCs can muster. As evidence, consider the fact that running Java on the
Wintel
 version of Netscape currently requires the more powerful Windows 95
operating
 system - although the company promises a version for Windows 3.1 as well.
 $500 barrier
    Another challenge to the hollow computer involves the economic assumption
 behind it: that to reach mass market status, a product can't cost much more
 than $500. That rule applies to TVs and VCRs; it doesn't, though, apply to
 some other well-known product categories.
    ""If you believed that, then it would be true that only 30 percent of
homes
 would have cars,'' said Martin Reynolds, a Dataquest analyst. ""The problem
 with a $500 computer is that it just doesn't do enough. The people who
 understand what a PC does will turn their nose up at them. And anyone who
 knows what a PC can do, and who wants one, will find a way to buy one.''
    Reynolds, a student of PC component expenses, said there is very little
 cost to be squeezed out of a typical Wintel PC, meaning that a computer
 couldn't get much cheaper without huge sacrifices being made.
    An entry-level system with a $2,000 retail price tag might have cost its
 manufacturer $1,700 or so to make. Of that, the $350 spent for a
 microprocessor and the $240 for 8 megabytes of RAM memory are the only two
 expenses that are ""unnaturally'' high, owing to the prices charged by both
 Intel and the world's RAM memory ""cartel.''
    Even with more vigorous CPU competition for Intel, Reynolds said, the
price
 of a microprocessor would come down by only about 20 percent.
    Reynolds said the other costs - $300 for a monitor; $200 for a full-
 function motherboard; $250 for storage; $300 for various packaging items
 all were produced in low-profit commodity businesses, meaning there was no
 ""fat'' that could be cut.
    And while Microsoft evokes a special enmity in many people, it adds only
 $50 or so to a unit's cost - the discounted price that computer companies
pay
 to license Windows.
    The $500 machine is just the hardware half of the hollow PC idea; the
other
 half involves a software system that can liberate users from the PCs and
make
 them instead full-fledged citizens of the Net. This is the role  Sun has in
 mind for Java, as demonstrated by the way it is trumpeting the language as
 something much more than a way to add gimmicks to World Wide Web pages.
    But just as with Oracle's stripped-down hardware, it is unclear how
 pleasant it would be to live in a Java-only world, if a visit to Sun's Java
 home page is any indication.
   Even without considering the time involved to transmit them over the
 network, Sun's sample programs are strikingly slow - at least by comparison
to
 those written specifically to run on a PC. That's one reason Microsoft says
 claims about Java's capabilities as a general-purpose software solution are
 wildly exaggerated.
   Those programs are all first passes at using the software, meaning they
will
 improve with time. In addition, Sun's Eric Schmidt said Java will get a big
 boost next year with the availability of a Java ""compiler,'' which will
take
 a Java program and translate it to run on a specific machine.
    ""There is a clear space where Java will be the best choice - in network-
 centric, multiplatform computing. There doesn't seem to be any disagreement
 about that,'' he said.
    ""But we believe the language can go well beyond that. For example, we
 think that a majority of programs for e-mail or for browsing - all the sorts
 of things that knowledge workers do - will be written in Java. Whether it
does
  or not is a matter of your view of the future.''
    Actually, one company does dispute the statement that Java is the best
 choice for network computing: Microsoft, which has its own versions of ""on-
 line languages'' waiting in the wings.




-------