快猫短视频

NAPSTER鈥檚 vision of free music for the masses may be dead and buried, but its
spirit lives on, and not just in copycat song swapping services. When the dust
settles, Napster will likely be remembered not so much for enabling music piracy
as for starting a revolution that changed the way the Internet worked.

Napster is the pioneer of a technology known as peer-to-peer networking, or
P2P for short. The core idea of P2P is to allow individual computers to
communicate directly over the Internet. By bypassing central servers, the
technology promises to transform the way people use the Net. In the process, it
could destroy the ability of anyone鈥攊ncluding corporations and
governments鈥攖o control what happens in cyberspace.

鈥淭he only reason the Internet until now has been relatively censorship-free
is that people who would censor the Internet haven鈥檛 considered it worthwhile,鈥�
says Ian Clarke, the inventor of a P2P system called Freenet. He and a group of
like-minded programmers envision networks that are totally decentralised,
impossible to censor and completely anonymous. In other words,
cyber-anarchies.

In technical terms, P2P networks are nothing new. The Internet itself started
life as a peer-to-peer system in which university and government mainframes
swapped information as equals. Only when the masses began to demand access did
the P2P ethos crumble. Private companies started hooking in their big computers
and offering connections and online services to modest little PCs. Thus was born
the client-server model. Big servers with fast connections and lots of memory
hosted the information. Little computers accessed it.

Napster鈥檚 winning idea was to give P2P to the masses. It figured out that it
didn鈥檛 have to store everything itself. Instead, it acted like a dating agency,
bringing music fans鈥攁nd their MP3 collections鈥攖ogether. Napster
provided members with an index of all the music stored on other members鈥�
computers, and software that enabled them to hook into each other鈥檚 hard drives.
Members could then swap files without the direct involvement of Napster.

Napster was thus able to give its members access to massive amounts of music
without having to store a single note itself. That turned out to be hugely
popular鈥攁t the last count Napster had 61 million users鈥攁nd was also
a big legal advantage. It鈥檚 clear that most of the recordings were being
distributed in violation of copyright laws. If Napster had been storing pirated
music on its site, it would have been shut down in days. The reason it lasted so
long was that it could quite credibly argue that it was an innocent
intermediary. If users happened to be trading pirated music it was no more
Napster鈥檚 fault than it鈥檚 the fault of the postal service if people mail
home-taped cassettes to one another.

Napster hadn鈥檛 just found a way of dodging the copyright lawyers, it had
solved a problem plaguing many large networks, especially the Internet. The
client-server models they are built on are hierarchies, and like all hierarchies
they鈥檙e great as long as you are near the top. But most small-time users are
near the bottom, shackled to an Internet service provider and its rules.

Achilles鈥� heel

Among the most irksome are those rules imposed by third parties, often backed
by lawsuits. Consider the successful campaign the Church of Scientology has
waged against its online critics. The tactic is to accuse critics鈥� ISPs of
hosting materials copyrighted by the Church. Scientologists have taken ISPs to
court and managed to have many of the critical websites removed.

Napster鈥檚 Achilles鈥� heel was that it retained a trace of the client-server
model. Because members were dependent on Napster for software and indexes,
record companies had a target to go after. And go after it they did. In December
1999, EMI, BMG, Sony, Warner, Universal and the Recording Industry Association
of America sued Napster for copyright infringement. Although the suit is not yet
settled, Napster suffered a terminal blow last month when a US court of appeal
ordered it to stop enabling the exchange of copyrighted material. Napster has
effectively thrown in the towel and is now trying to find a way of charging for
its services so it can pay royalties.

But the P2P pirates aren鈥檛 about to go away. Napster鈥檚 success has inspired
others, and they鈥檙e determined to learn from its mistakes.

One such system is Gnutella. Originally developed by a company called
Nullsoft, the software was released in March 2000 only to be withdrawn the same
day under pressure from Nullsoft鈥檚 parent, America Online, which was in the
process of merging with music and media giant Time Warner. But the cat was out
of the bag. Enthusiastic hackers unpicked Nullsoft鈥檚 code and used it to write
versions of their own. Within weeks there were several different but mutually
compatible Gnutella knock-offs on the Web. The Gnutella commonwealth was
born.

Unlike Napster, Gnutella has no central authority. No one keeps track of
users and nobody indexes the files they exchange. Anyone can write software to
access the network, and most of what has been written is open source, so anyone
can add to it and improve on it. There are now more than a dozen versions
available for free, with names like Gnotella, Newtella, Gnut, LimeWire and
ToadNode.

To join the network, you simply download one of these software packages from
the Web. This turns your computer into a 鈥渟ervent鈥濃�攂oth a client and a
server. Once you鈥檝e done that you鈥檙e ready to find some other
servents鈥攖heir locations are widely publicised on websites and chat
rooms鈥攁nd make contact with them. The connections are made over the
Internet, and all the computers are identified by their Internet Protocol (IP)
addresses, the basic numeric addresses that identify computers on the Internet.
But Gnutella is not the World Wide Web. Your computer communicates directly with
the servents it knows about, and those servents pass messages back and forth to
yet more servents, which do the same in an ever-expanding net.

To search for a file, you type in keywords and send them to your immediate
neighbours. They search the contents of their hard drives, return the hits to
you and forward your request to yet more servents, which repeat the process. A
single request can quickly reach thousands of computers.

Gnutella is designed to share any kind of file: images, text and software, as
well as MP3s. Each user decides which files to make available. A lot of pirated
material gets passed around, but the decentralised nature of the network means
that there鈥檚 no obvious legal target.

In any case, Gnutella鈥檚 difficulties right now aren鈥檛 legal, they鈥檙e
technical. Napster鈥檚 courtroom problems sent a wave of new file sharers onto the
Gnutella network鈥攕uddenly upping the number of people logging on each day
by 50,000, according to some estimates鈥攁nd the surge in traffic brought
the whole thing grinding to a halt. It turned out the network simply didn鈥檛
鈥渟cale鈥� well. The number of requests increased exponentially until it exceeded
the capacity of a standard 56K modem. Suddenly, thousands of servents turned
into dead ends. The network fragmented into dozens of small, disconnected
networks, none of them bigger than about 1200 computers.

True to its roots, Gnutella is dealing with the problem in a decentralised
way. Each group of programmers has its own favoured solution. Some push the
slowest servents to the edge of the network. Others abandon ideological purity
and revive the client-server model, making slow servents connect via a faster
node. Whichever fix works best will out.

But even for Gnutella, legal problems are looming. It鈥檚 true that the system
is less centralised than Napster, but that doesn鈥檛 mean there鈥檚 nobody to go
after. About half of all Gnutella files are provided by just 1 per cent of
users, and that 1 per cent present a big, fat target to anyone who wants to
start suing for copyright infringement.

See you in court

The organisation most likely to start filing lawsuits is the Recording
Industry Association of America. 鈥淲e have not done any enforcement against
Gnutella at this point. But that鈥檚 not going to last long,鈥� says Frank
Creighton, director of the RIAA鈥檚 anti-piracy initiative. When the RIAA decides
to move, he says, it will probably target that active 1 per cent. Finding out
who they are shouldn鈥檛 be hard because Gnutella servents need to know one
another鈥檚 IP addresses to communicate. Anyone can find out which ISP hosts a
particular IP address, and after that a threatening letter or writ can have the
user kicked off or force the ISP to reveal a name that can be pursued through
the courts.

But there is a P2P network that looks capable of evading the lawyers. Called
Freenet, it鈥檚 a radical system created from the ground up to be anonymous and
censorship-proof. Its creator, Ian Clarke, is a free-speech absolutist who feels
that today鈥檚 Internet, despite its freewheeling image, is vulnerable to
censorship. And that鈥檚 dangerous, he says. 鈥淚f we look back through history we
can see repeated examples where censorship and propaganda have been used to
manipulate people into permitting, and even participating in, the most terrible
acts of barbarism.鈥�

Like Gnutella, Freenet uses the Internet as a backbone to send and receive
information, and identifies each computer by its IP address. But unlike
Gnutella, it covers its tracks whenever information is transferred.

Hooking your computer up to Freenet is similar to joining Gnutella. First you
download the software from the Web. Then you contact other Freenet computers,
whereupon your computer becomes a Freenet 鈥渘ode鈥�. Freenet is made up of
thousands of these nodes, and each one can make files available. When you
鈥渋nsert鈥� a file鈥攕ay an MP3鈥攊nto Freenet it is encrypted and then
copied to several other nodes. Each node knows which documents it holds and also
has information about documents stored on a few other nodes. Neighbouring nodes
communicate routinely, updating one another on additions to the network. But no
single node knows about more than a fraction of the entire network.

How do you get information out of a system like this? As Clarke explains it,
the strategy is similar to the way people navigated before maps. Starting out, a
group of travellers might have known only to go north. But the closer they got
to their goal, the more detailed was the information they got from people they
asked, until finally they found someone able to tell them that yes, the minstrel
they were looking for lived right around the corner, second hovel on the
right.

Before you start a Freenet search, you must know the title of the document
you鈥檙e looking for. How users will do this is still up in the air. One obvious
possibility is an index within Freenet itself鈥攖hough that raises the
question of how to find the index in the first place. Another idea is to post it
on the Web, though this may create a juicy legal target. Each document also has
a numeric key that is cryptographically linked to the title, and it鈥檚 this
you鈥檙e actually looking for during a Freenet search.

Let鈥檚 say you know the key is 123鈥攖hough of course real keys will be a
lot more complex than that. Each node, including yours, knows what documents it
holds, and also has a list of documents held by a few other nodes. Your computer
will look to see if it has document 123. If not, it will look up to see if it
knows a node that has document 123. If it doesn鈥檛, it contacts the node with the
document that comes closest鈥攎aybe document 135. That node might not know
where 123 is either, but it knows which node has document 119, so it sends the
request there. The idea is that with each request you get closer to the document
you really want. When the document is found, it鈥檚 returned along the request
chain (see Diagram). As the document
is returned, each node along the chain makes a copy of it and stores it.

One consequence of this is that the more requests come in for a piece of
information, the more copies there will be on the network, and the easier it
will be to find. It also means there鈥檚 no way of telling where the document
originally came from. All you know is that you asked a neighbouring node for it,
and it fetched the document from somewhere. Conversely, if you receive a request
for a file, you have no idea who made it.

The result is a censorship-proof network. If the powers that be request a
file from a node they鈥檒l get a copy. If they seize that node they鈥檒l definitely
find a copy. But it would be impossible for them to prove that the file was
there before they requested it, so the exercise amounts to entrapment, Clarke
says. And because documents are stored in encrypted form, the node鈥檚 owners can
argue truthfully that they had no idea any particular document is held there.
What鈥檚 more, as the act of requesting a document generates new copies,
censorship is self-defeating.

Freenet鈥檚 developers insist that their network won鈥檛 suffer the same scaling
problems as Gnutella. Although they鈥檙e not sure how many users there
are鈥攖he network, after all, is designed not to give away too much about
itself鈥攁bout 20,000 people have downloaded the software. Simulations show
that Freenet won鈥檛 grind to a halt as it grows bigger.

Not everyone accepts that Freenet is as censorship-proof as Clarke thinks.
Creighton reckons he can bring it down by getting the IP addresses of individual
nodes, sending letters to ISPs, and taking some users to court, just as he wants
to do with Gnutella.

But if Clarke turns out to be correct, Freenet will usher in a different
world. No one will be able to stop you downloading free music files from the
Internet. You鈥檒l be able to criticise the rich and powerful without fear of
being silenced or punished. And you鈥檒l be able to read whichever spy memoir your
government is trying to suppress at the moment.

By the same token, you鈥檒l be powerless to stop people from plagiarising your
copyrighted work or telling lies about you. Nobody will be able to take down
child pornography or stolen nuclear secrets.

Napster set out to give us free music, but it seems to have put us on the
road to absolute freedom of speech. If so, the real challenge hasn鈥檛 even
begun.