World Wide Web

• 1. Introduction

  ›

  The World Wide Web, also known as WWW, W3 or simply the Web, was the target of a joke by the English writer Douglas Adams. In The Independent on Sunday in 1999, he observed that the Web's abbreviation takes three times longer to say than the name it abbreviates.

  The joke lands because the Web is so familiar. Billions of people now use it as their main way into the Internet, and it has become the dominant information platform on Earth. Yet it began as a modest idea from an English computer scientist named Tim Berners-Lee, who was working at CERN in 1989. He described his creation as a "universal linked information system".

  This documentary asks how a tool built to solve one laboratory's filing troubles ended up as a shared global habit. Why did its inventor reject the neat tree-shaped filing systems of his day? What happened when rivals offered a competing design? And why do so many web addresses begin with three letters that nobody ever required?

  Many people also treat the Internet and the Web as the same thing. They are not, and the difference explains a great deal about how this story unfolds.

  2461218

• 2. Design Against the Tree

  ›

  At CERN, a large organisation whose people and projects constantly shifted, Tim Berners-Lee faced a practical headache. He needed a way to store, update and find documents and data files. He also needed a way to share them with collaborators outside the laboratory.

  The existing CERNDOC documentation system arranged material as a tree, and so does the Unix filesystem. Berners-Lee dismissed that structure. He also passed over systems that labelled files with keywords, such as VAX/NOTES.

  His alternative grew out of ENQUIRE, a private system he had built at CERN in 1980. Later he learned of a hypertext model proposed by Ted Nelson in 1965. In that model, documents could be linked freely through "hot spots" embedded in the text. The discovery reassured Berners-Lee that his own instinct was sound. Apple's HyperCard system would go on to popularise the same family of ideas.

  But Berners-Lee asked for far more than HyperCard offered. His system was meant from the start to link multiple databases living on independent computers. It was also meant to let many people use it at once, from any computer on the Internet.

  He imagined it eventually carrying graphics, speech and video, not just text. Links might point to data files that kept changing, or even start programs on the server. He also sketched "gateways" so that documents stored in other ways, such as traditional file systems or Usenet, could be reached through the new system.

  Above all, he insisted that nobody should control it. No central authority would coordinate or approve the creation of links. In May 1989 he sent a proposal to CERN describing all this, though he gave the system no name.

  647121349

• 3. Releases and Rivals

  ›

  By the end of 1990, Berners-Lee had a working system. It included a browser called WorldWideWeb, which lent its name to the project and the network, and an HTTP server running at CERN. Along the way he wrote the first version of the HTTP protocol and the basic syntax of URLs. Without ever announcing it, he also made HTML the main document format. In November 1990 he first wrote the name in its original CamelCase form, which was later discarded.

  In January 1991 the technology went to other research institutions. On the 23rd of August 1991 it was opened to the whole Internet. It succeeded at CERN and then spread through scientific and academic circles, and within two years 50 websites existed.

  The decisive step came on the 30th of April 1993. CERN made the Web protocol and code available royalty-free, so anyone could build on it.

  That timing mattered, because a rival called Gopher was already in the field. The University of Minnesota ran Gopher, and in February 1993 it announced licensing fees for its own Gopher server software. Users feared that independent versions might also be charged. Gopher's growth stalled, to the benefit of the Web.

  Gopher also had a stricter design. Every document had a defined format and type, and users reached it through menus set by the server. The Web's HTML was far more free-form. In September 2000 the University of Minnesota re-licensed its Gopher software under the GNU General Public License, but by then the early Mosaic browser had absorbed Gopher's client functions along with its protocol.

  Mosaic came from the NCSA later in 1993. It was a graphical browser that could show images inline and send forms to be handled by the HTTPd server. Thousands of websites appeared in less than a year.

  Marc Andreessen and Jim Clark then founded Netscape. Its Navigator browser, released in 1994, brought Java and JavaScript to the Web and quickly took the lead. When Netscape became a public company in 1995, a frenzy for the Web followed, and it set off the dot-com bubble.

  Microsoft answered with its own browser, Internet Explorer, and the browser wars began. By bundling Explorer with Windows, Microsoft made it the dominant browser for 14 years.

  15202136

• 4. Anatomy of a Request

  ›

  Imagine a browser asked for the page /home.html on a server named example.org. Before anything else, the browser looks up that server name in the globally distributed Domain Name System. The lookup returns an Internet Protocol address, for instance 203.0.113.4 or 2001:db8:2e::7334.

  The browser then sends an HTTP request to the machine at that address. It aims at a well-known port number so the receiving computer can tell web traffic from its other services. Ordinary HTTP normally uses port 80, and HTTPS normally uses port 443. The request itself can be just two lines: a GET for /home.html, and a Host line naming example.org.

  If the server can comply, it replies with a status line reading 200 OK and a Content-Type of text/html, followed by the page. In the 1990s, moving through pages this way became known as browsing, web surfing after channel surfing, or navigating the Web. An early study of this new behaviour found five patterns: exploratory surfing, window surfing, evolved surfing, bounded navigation and targeted navigation.

  The page the server sends is written in HTML, which together with CSS and JavaScript forms a triad of cornerstone web technologies. HTML wraps text in tags written with angle brackets, marking headings, paragraphs, lists, links and quotes. Browsers hide the tags and use them to decide how to show the words. HTML originally carried some hints about appearance, but in 1997 the W3C began encouraging CSS for that job instead.

  Brendan Eich of Netscape developed JavaScript in 1995 for use inside web pages, and its standardised form is called ECMAScript. A page can be static, delivered exactly as stored, or dynamic, assembled on the fly by server software or by JavaScript in the browser. Dynamic pages suit things like banking sites and web email, where each user needs different information.

  Ajax, short for asynchronous JavaScript and XML, lets a page ask the server for small pieces of data without being replaced. The user sees one page while its contents keep changing underneath. Even Ajax updates neither add a page to the history nor cut off the forward history.

  On the server side, many generic web servers support scripting through Active Server Pages or PHP. Static pages are usually much faster and easier to cache, but only scripted ones can retrieve or modify database information. Web servers even sit inside printers, routers and webcams, where they let an owner manage the device with nothing more than a browser.

  Caching speeds up all of this. Site designers gather their CSS and JavaScript into a few site-wide files so that browsers and web caches can store them efficiently.

  444546253

• 5. The Accidental Prefix

  ›

  The first web server was named nxoc01.cern.ch, with no www in sight. The prefix came later, from a long-standing habit of naming Internet hosts after the service they provide. A file transfer machine might be called ftp, and a Usenet machine news or nntp.

  According to Paolo Palazzi, who worked at CERN alongside Berners-Lee, the popular use of www was an accident. The World Wide Web project page was meant to live at www.cern.ch, while info.cern.ch was meant to be the CERN home page. The DNS records were never switched, and other institutions copied the pattern.

  No technical or policy standard requires the prefix. Established sites still use it, or use other subdomains such as www2, secure or en for particular purposes. Many servers are set up so that the bare domain and the www version lead to the same site, while others insist on one form or treat them as different websites.

  There is a practical reason to keep it. A subdomain can point to a cluster of web servers through a CNAME record, which helps balance incoming traffic. Only a subdomain can be used this way, so the bare domain root cannot do the same job.

  Browsers have long quietly filled in missing pieces. Typing an incomplete name can make them add www at the start and possibly .com, .org or .net at the end. This feature appeared in early Firefox, when it still had the working title Firebird in early 2003, and it came from an earlier practice in browsers such as Lynx. Microsoft was reportedly granted a US patent for the same idea in 2008, though only for mobile devices. Browsers also add http:// when users leave it off, and the extra s in https:// marks the encryption that passwords and banking details require.

  The prefix is now fading. As the mobile web grew, services such as Gmail.com, Outlook.com, Myspace.com, Facebook.com and Twitter.com came to be named without the www, and often without the .com.

  Berners-Lee himself states that the official spelling is three separate capitalised words with no hyphens. Pronunciation varies. Some people in New Zealand say dub-dub-dub, and Stephen Fry, in his Podgrams podcast series, says wuh wuh wuh.

  4950515237

• 6. Glass Beneath the Web

  ›

  The Advanced Research Projects Agency Network, known as ARPANET, was created in 1969 in collaboration with universities and researchers. It was one of the first versions of the Internet, but only researchers could use it. In 1985 the National Science Foundation set up NSFNET, a program that gave researchers access to supercomputers.

  That is the distinction people often blur. The Internet is a global system of computer networks, joined through telecommunications and optical networking. The Web is a collection of documents and other resources, linked by hyperlinks and URIs, and fetched with HTTP or HTTPS.

  In 1993 the United States passed the National Information Infrastructure Act. Consumers and corporations had been pressing to privatise the network, because public access was limited. The law told the National Science Foundation to hand control of the optical capabilities to commercial operators.

  Privatisation and the Web's public release in 1993 together drove up demand. Developers looked for ways to cut the time and cost of laying new fibre. They also wanted to send more information down each strand. The large amount of optical fibre installed worldwide by the end of the twentieth century had already laid the foundation of today's Internet.

  In 1994 the optical components division of Pirelli S.p.A. introduced a wavelength-division multiplexing system, called WDM. Its four channels let a single fibre carry more information at once.

  Pirelli was not alone. Ciena Corporation was founded in 1992 by the optical networking engineer David Huber and the entrepreneur Kevin Kimberlin. It drew on laser technology from Gordon Gould and William Culver of Optelecom, Inc., and focused on optical amplifiers that move data as light.

  Under chief executive officer Pat Nettles, Ciena built a dual-stage optical amplifier for dense wavelength-division multiplexing. It was deployed on the Sprint network in 1996 and patented in 1997.

  6364656872

• 7. Search, Secrets and Surveillance

  ›

  In 1990 a tool named Archie was released as the world's first search engine. It indexed FTP sites, a method for moving files between a client and a server. Yahoo! in 1995 and Google in 1998 later superseded it.

  Modern search engines run algorithms on a web crawler, which lets them keep information up to date, unlike directories maintained only by human editors. Some content stays out of their reach. In 2001 the computer scientist Michael K. Bergman was credited with coining the term deep web for pages that standard search engines do not index. They include web mail, online banking and anything behind a paywall, such as video on demand.

  Much of the Web's everyday memory lives in cookies. An HTTP cookie is a small piece of data that a website sends and the browser stores. Cookies let a shop remember what is in a shopping cart, or let a site recall names and addresses typed into forms. Authentication cookies are the most common way a server learns whether a user is logged in, and which account is in use.

  Tracking cookies, especially third-party ones, build long-term records of people's browsing. That worry prompted lawmakers in Europe and the United States to act in 2011. European law now requires sites aimed at European Union states to obtain "informed consent" before storing non-essential cookies. Jann Horn, a Google Project Zero researcher, has described how intermediaries such as Wi-Fi hotspot providers can read cookies, and he recommends private browsing mode.

  Privacy leaks in other ways. Servers usually log the IP address of each request, and unless HTTPS is used, requests and responses cross the Internet in plain text. A virtual private network encrypts traffic and masks the original IP address.

  Once a user hands over a real name or email address, a site can tie later and earlier visits to that person. The profile might cover leisure activities, shopping interests and profession, and it may be sold or shared. For many people the result is merely some uncannily relevant advertising. Social networking sites push for real names because executives believe that makes the experience more engaging. Photographs add a face to a profile, and image caching, mirroring and copying make it hard to remove one.

  Criminals use the Web for malware, identity theft, fraud and espionage. Google has measured that about one in ten web pages may contain malicious code. Security firm Sophos found that most attacks happen on legitimate sites hosted in the United States, China and Russia. SQL injection is the most common malware threat. One 2007 estimate held that 70% of websites were open to cross-site scripting attacks on their users.

  In February 2013, RSA, the security division of EMC, estimated global losses from phishing at $1.5 billion for 2012. Jonathan Zittrain argued that users sharing responsibility for safety is far preferable to locking down the Internet.

  8992948588

• 8. Rules, Reach and Memory

  ›

  In 1996 the World Wide Web Consortium, founded by Berners-Lee, created XML and went on to recommend replacing HTML with the stricter XHTML. Developers meanwhile exploited an Internet Explorer feature called XMLHttpRequest to build Ajax applications, and the Web 2.0 revolution followed.

  Mozilla, Opera and Apple rejected XHTML and formed the WHATWG, which developed HTML5. The W3C conceded in 2009 and abandoned XHTML. In 2019 it handed control of the HTML specification to the WHATWG.

  The quarrel shows how web standards are made. They are not declared by one person or company. Standards organisations, groups of interested and often competing parties, produce them. The family includes W3C recommendations, the WHATWG Living Standard, IETF Requests for Comments, and standards from ISO and Ecma International. It also includes the Unicode Standard and the registries kept by the Internet Assigned Numbers Authority.

  Standards also serve people who might otherwise be shut out. The W3C holds that the Web must be accessible to people with visual, auditory, physical, speech-related, cognitive or neurological disabilities. Accessibility helps people with temporary problems like a broken arm, and ageing users. Berners-Lee once said, "The power of the Web is in its universality. Access by everyone regardless of disability is an essential aspect." Many countries now regulate web accessibility, and the W3C Web Accessibility Initiative produced simple guidelines for authors and software developers.

  The W3C Internationalisation Activity works to make web technology function in all languages, scripts and cultures. Unicode gained ground from 2004 or 2005 and, in December 2007, overtook both ASCII and Western European as the Web's most frequently used character set. Resources were first identified by URIs in a subset of US-ASCII, but IRIs now allow any character in the Universal Character Set.

  The Web's structure is described by the webgraph, in which pages are nodes and hyperlinks are directed edges. Those edges decay. Pages move or vanish, leaving dead links, a problem called link rot. The Internet Archive, active since 1996, is the best known effort to save websites before they disappear.

  28293031