4.4.3 Principles of the design

But what sort of system? After criticising the types of linking which were then in vogue – for example, the hierarchical tree-structures exemplified by the Help systems of mainframe computers, or those which relied upon indexed keywords – Berners-Lee went on to propose a solution to CERN's problem. It was a relatively old idea called ‘hypertext’ – i.e. non-linear text in which one part of a document contained links that could take one to other parts of the same document or to other documents.

The special requirements of a hypertext system for CERN were, he believed, that it should: allow remote access across networks; be heterogeneous (i.e. allow access to the same information from different types of computer system); be non-centralised; allow access to existing data; enable users to add their own private links to and from public information, and to annotate links as well as nodes privately; and enable ‘live’ links to be made between dynamically changing data.

The proposal was accepted by CERN management. In November 1990, Berners-Lee wrote a program called a browser which provided a virtual window through which the user could see the resources held on various internet computers as a ‘web’ of linked information sources. It refracted, as it were, a world of disparate information sources in such a way that they appeared as a uniform whole.

He then wrote another program – a hypertext ‘server’ which would hold documents and dispense them as requested by browsers.