快猫短视频

鈥淪ure, we鈥檙e worried.鈥�

Steve Ballmer hardly seems the worrying kind. A large and ebullient man, he
is Bill Gates鈥檚 number two, his likely successor at Microsoft, and worth more
than $10 billion. But answering a question after a speech in September,
Ballmer admitted that something was getting to him.

The question was not about the latest US Department of Justice antitrust
action, nor the growing success of Sun Microsystem鈥檚 Java programming language.
And it certainly didn鈥檛 mention a rival outfit: with a price tag of $270
billion, Microsoft dwarfs them all. What Ballmer, Gates and the authors of
leaked Microsoft memos are all worried about is a bunch of idealistic hackers
scattered round the world, the software they鈥檝e created and the revolutionary
method they have devised for writing it. This new software poses a 鈥渄irect鈥�
threat to Microsoft鈥檚 revenue, says one memo, leaked on 31 October.

Their programs are already running most of the Internet. According to the
monthly survey by the British consultancy Netcraft, the Web server software
Apache is used by more than half of all websites. Sendmail moves nearly every
e-mail message across the Net, while the BIND program acts as a traffic cop for
most of the global network, directing messages down the right connections to
their final destinations.

The proven robustness of these programs is bad enough for Microsoft. Even
worse is that all of them鈥攁nd many others鈥攁re completely free.
That鈥檚 鈥渇ree鈥� in two senses: you do not have to pay for them, and the 鈥渟ource
code鈥� in which they are written is openly available. You can modify the programs
in any way you like, and even sell the result.

The 鈥渙pen source鈥� movement is Microsoft鈥檚 worst nightmare: a group of
programmers that it cannot outcompete because its members are not motivated by
profit, and which it cannot buy because they do not exist as a formal company.
And because the results of their work are so good, more and more of Microsoft鈥檚
potential customers are turning to them.

In the vanguard of the open source movement is Linux
(http://www.linux.org.uk/), which was started in 1991 by a 21-year-old Finn,
Linus Torvalds, who wanted to write a free alternative to the popular but costly
operating system Unix. Today Linux is used by an estimated 7 million people, and
the number is growing rapidly. One of Linux鈥檚 advantages is that it runs on just
about any hardware, from multi-processor supercomputers down to Palm Pilots. It
is compact (it can fit on a floppy), highly efficient (even PCs based on Intel鈥檚
old 386 chip can handle it) and very fast.

Darwinian dynamic

Torvalds did not invent the idea of software that is doubly free: that honour
goes to Richard Stallman, the leader of another group of software rebels called
the Free Software Foundation (http://www.fsf.org/). But Torvalds stumbled
upon, and has since developed, a crucially important Darwinian dynamic.

In a commercial software company, every program is carefully planned, and
writing tasks are allotted unilaterally by the project leader. Linux is
different. It is designed as a series of modules, and anyone can work on any of
these interlocking elements. Whether your work gets included in the final
release depends on the consensus view of how good it is鈥攏atural selection
in action. Torvalds has the final say on what goes in, but he listens to top
programmers in the relevant areas鈥攊n networking, say鈥攚ho in turn
canvass opinion in their fields. The only reward anyone, even Torvalds, gets for
this work is kudos from fellow hackers. That鈥檚 enough, it seems, to attract a
flow of keen recruits, typically computer science students or software engineers
who code Linux on the side.

Such purposive anarchy is made possible by the Internet. Trial versions of
programs can be downloaded, and comments sent back to the authors, wherever they
are. Programs frequently evolve on a daily basis. With the help of the Net, the
Linux model exploits the ingenuity of typically hundreds of programmers and
hundreds of thousands of testers. It鈥檚 a pool of creativity that Microsoft, for
all its huge resources, will never be able to match.

In particular, the Linux approach has proved far better than traditional
testing methods at winkling out bugs in programs. This is partly thanks to the
huge numbers of people involved, but it鈥檚 mostly because, with the source code
freely available, testers offer solutions to the problems they discover. This
just cannot happen with commercial software. Successful debugging is no small
advantage: as programs get more complex, bugs are becoming a colossal
problem.

If Torvalds invented the new open source approach, its leading theorist is
another hacker, Eric Raymond. In his 1997 essay The Cathedral and the
Bazaar, he analysed why Linux worked and suggested that its approach might
be applied more generally (http://sagan.earthspace.net/~esr/writings/cathedral-bazaar/).

Surprisingly, the Net software company Netscape got the message, and in
January announced that it would release the latest version (version 5) of its
flagship Web browser suite Communicator, code-named Mozilla, as open source
(http://www.mozilla.org/). It was unprecedented for a commercial company to
hand over what amounted to its crown jewels鈥攖he source code for its
leading program. In a sense, though, Netscape had little to lose. Communicator鈥檚
main rival, Microsoft鈥檚 Internet Explorer, had always been free (though its
source code is still secret). So giving Communicator away became the only
option.

Netscape benefited immediately. For example, restrictive American export laws
meant that Netscape had to release Mozilla with a watered-down encryption
system. But a group of hackers called the Mozilla Crypto Group put back what
Netscape had taken out 鈥渓ess than a day after the code was released鈥�, as Eric
Young, one of the team, says with pride.

Once free, always free

The Mozilla experiment might seem a dangerously high-profile test for the
open source concept. But Raymond is unconcerned. 鈥淲e鈥檝e already passed the point
where Netscape鈥檚 [failure] would wreck us,鈥� he says. Fellow open source hacker
Havoc Pennington stresses that the whole point of releasing Mozilla as open
source is that 鈥渋f Netscape goes down, someone else just takes over鈥�. (Mozilla
is not an experiment about free software, he says. 鈥淭he experiment is whether
Netscape can make money on it.鈥�) Similarly, whatever the Internet service
company AOL does, now that it has bought Netscape, it cannot rescind
Mozilla鈥攐nce free, always free.

Partly as a result of Netscape鈥檚 trailblazing move, several other companies
have followed suit, suggesting that open source is entering the business
mainstream. This summer, for example, IBM announced that it would include Apache
as an option with its otherwise traditional WebSphere server software. While
this may be small beer for IBM, the implications for open source are profound.
Not only will a major software company be working with the open source movement
just as it would with conventional commercial partners. More importantly, IBM
will for the first time be providing after-sales support for an open source
program.

This removes one of the big objections that commercial users have had to open
source software. 鈥淚f a company is going to use some technology for a
mission-critical part of their work they normally like the idea of a supplier
they can sign a contract with, and threaten to sue when things break and they
don鈥檛 get fixed,鈥� says Young. Not being able to get this sort of assured support
鈥渕akes management twitchy鈥�, he says.

Until now, people who use open source programs have obtained their support
online, through Usenet newsgroups and, even more directly, by e-mailing the
relevant author. This leads to a swift resolution of problems in times that are
unheard-of in the commercial sector. But all this happens on an informal basis,
hardly something you can run an IT department on.

But here was IBM offering precisely the guaranteed and dependable support
that was missing: 鈥淭his is one of the areas where IBM is helping out the Apache
organisation,鈥� says Tony Occleshaw of IBM. Suddenly, open source is beginning to
look very attractive for businesses. 鈥淚 think we鈥檒l see other companies than
Netscape and IBM that endorse open source projects,鈥� says Torvalds.

Alongside lack of support, another major criticism of Linux was that no major
business programs ran on it. In particular, heavyweight databases鈥攖hose
digital treasuries entrusted with key corporate information鈥攚ere
conspicuous by their absence. So a clutch of announcements this summer that all
bar one of the top vendors of database software鈥擟omputer Associates, IBM,
Informix, Oracle and Sybase鈥攚ould be coming out with Linux versions of
their products was a real coup. At a stroke, a major objection to the deployment
of Linux in commercial companies had been removed.

The absentee is Microsoft.

It is not only software companies that have suddenly seen the open source
light. The chip maker Intel will support Linux on its forthcoming 64-bit
high-performance processor, code-named Merced. This is particularly significant
because Windows NT鈥攚hich Microsoft is banking on as the future operating
system for business and personal users alike鈥攊s also being recast for
Merced. But Linux will get there first, and when NT does finally arrive, it will
find a formidable rival already in place. That Linux has become a force in
supercomputing was demonstrated in April, when scientists at the Los Alamos
National Laboratory in New Mexico, built a supercomputer out of 68 off-the-shelf
processors running Linux. It cost about a tenth of what they would have had to
pay for a comparable conventional machine.

Of course, this backing for Linux and open source is all very well, the
cynics will say, but where鈥檚 the money? If the open source model is to become
anything more than a curiosity, then software companies using this approach need
a viable business model. That鈥檚 precisely what several open source pioneers are
hoping to come up with. For example, alongside Sendmail the open source program,
there is now a commercial version from Sendmail the for-profit company. Set up
by the original author of the program, it will sell enhanced varieties while
continuing to support the public-domain effort.

Similarly, John Ousterhout, who invented a popular open source programming
language Tcl, has set up Scriptics to sell advanced Tcl tools alongside the open
source core that he will continue to support for free. He expects such hybrid
companies to become an essential part of the open source world: 鈥淚 think that
you鈥檒l see more and more open source companies over the next several years, as
more and more interesting open source packages arise and need the resources that
only a company can provide.鈥� Ousterhout also sees a big advantage in this mixed
approach. 鈥淭he open source version will keep us honest,鈥� he says. 鈥淲e鈥檒l have to
make sure that our version is always at least as good as the free one.鈥�

These fledgling ventures can draw comfort from the fact that in some areas
this combined approach is already well established. California-based companies
such as C2Net and Tenon have thrived by offering commercial versions of the
Apache Web server. And in the Linux world, many sell what are called Linux
distributions. Linux itself is just a kernel鈥攖he central
functions鈥攐f an operating system. Distributions are huge collections of
programs that complement it to create a complete and powerful operating system.
Some of these companies, for example Caldera and Red Hat, also provide support
for customers who use their software.

Given these advantages, and the manifest goodwill that the open source
approach generates, software companies are starting to ask themselves 鈥淲hy
shouldn鈥檛 it be open source?鈥� rather than 鈥淲hy should it be?鈥� Corel, for
example, sells a Linux-based PC system called NetWinder, and makes the code
available as open source. Even Microsoft seems to recognise the shifting mood.
鈥淧erhaps,鈥� Ballmer has conceded, 鈥渨e need to open things up.鈥�

For Microsoft, opening things up may be too much of a wrench, but it has
clear benefits for software with a small but enthusiastic following. It allows
that enthusiasm to be channelled into continuing development that a product鈥檚
owners could not otherwise afford. In this vein, the American consumer rights
advocate Ralph Nader wrote an open letter to IBM鈥檚 boss, Lou Gerstner,
suggesting that Big Blue鈥檚 OS/2 operating system would flourish if released as
an open source program. IBM demurred.

Apple was similarly unmoved when Macintosh activist Don Yacktman suggested
placing portions of the Mac OS X operating system in the public domain to
encourage faster development. If it doesn鈥檛, Yacktman warns, Linux distributions
could overtake Macintosh as the world鈥檚 number two operating system after
Windows. 鈥淭his is very possible,鈥� he says.

While the open source idea is clearly gaining ground, there remains a major
concern. 鈥淭he need is for more applications,鈥� says Dean Taylor, Caldera鈥檚
director of marketing. There aren鈥檛 many open source applications around that
have achieved the maturity or popularity of software such as Linux, Sendmail or
BIND. One notable exception is the GIMP, an excellent image manipulation program
that does a job similar to that of the glamorous Photoshop. But what about more
mundane applications? 鈥淚鈥檓 fairly pessimistic about open source software being
able to succeed for the non-sexy applications,鈥� says Peter Mattis, one of the
GIMP鈥檚 original authors. 鈥淎 word processor isn鈥檛 very sexy and I think that鈥檚
why we haven鈥檛 seen a good one yet.鈥�

Another reason applications have been slow to appear, says Pennington, is
that it鈥檚 taken until now to build the necessary software
infrastructure鈥攖hings like the graphical user interface. Unlike Windows,
Linux has no standard screen layout. Or rather it has two, KDE and Gnome, which
are battling it out for supremacy. This schism weakens the image of open source
among potential commercial users. It seems to confirm the jibes that such
software is written by fanatics for fanatics, and that its programmers would
rather engage in a good ideological scrap than sit down together to produce
something useful.

The main Linux distributions, such as those from Red Hat and Caldera, suffer
from similar inconsistencies: they are sufficiently different that application
programs written for one distribution will not always function with another.
This makes it particularly hard for a company to risk making one particular
distribution the cornerstone of its IT policy.

To address this problem there is a move to create something called the Linux
Standard Base, a kind of minimum agreed standard that will allow applications to
run whichever distribution is chosen. 鈥淭he good thing about the Linux Standard
Base is that the community is working together,鈥� notes Taylor.

Raymond is optimistic: 鈥淎pplications and end-user interfaces are the
happening thing now, and that鈥檚 only going to accelerate.鈥� Torvalds goes
further. In his view, the battles over user interfaces and distributions are not
a weakness, but a reflection of open source鈥檚 underlying strength. He has
encouraged a system where contention flourishes, because for natural selection
to operate there has to be competition. This may lead to short-term problems,
but ultimately, he says, the result will be better.

So, does anything else stand in the way of open source鈥檚 passage to the
computer mainstream? Perhaps just one thing鈥攐r rather a person. For one of
the most implacable critics of some aspects of the open source movement is none
other than its spiritual godfather, Stallman, who is deeply suspicious of the
commercial hybrids. 鈥淭he open source movement is being abused by a number of
companies that seek to be regarded as part of the free software community, even
though their main activity is writing proprietary material that works with free
software,鈥� he says.

But it is unlikely that Stallman鈥檚 moral exhortations will halt open source鈥檚
rapprochement with commerce. 鈥淚t鈥檚 an ideal time for alternative visions of
computing to assert themselves,鈥� says Raymond. He argues that Microsoft has
become a victim of the paralysis that afflicted IBM in the 1970s,
when鈥攍ike Microsoft today鈥攊t was charged with abusing its dominant
position. The next release of Windows NT is taking so long to complete that 鈥渋t
is beginning to look like it might be a full-fledged disaster鈥�, says Raymond.
(It鈥檚 now been renamed Windows 2000 to reflect its ever-receding launch date.)
By contrast, Linux is forging ahead. 鈥淲e in the open source community are well
positioned to win,鈥� says Raymond.

One thing seems certain: Steve Ballmer and Bill Gates are not going to be
able to stop worrying any time soon.