JOVANA
Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
Back to the library
信息论 / 计算机科学 1989

信息管理:一份提案

蒂姆·伯纳斯-李

他把超文本接上互联网,给每份文档一个地址——万维网。

Choose your version
In depth · the introduction

1989 年,一座物理实验室里,一份不动声色的提案,只求一件不大的事,却无意间,描述出了今天大半个世界赖以栖居的媒介。

把核心想法拆开看

蒂姆·伯纳斯-李,在日内瓦近郊那座巨大的粒子物理实验室 CERN 工作。那里头绪太多、人太多、变得太快,以至于整个地方,老是对自己的知识失去掌控——谁懂什么,某样东西写在了哪里。当时的归档系统太僵硬了:一棵严格的树,把每份文档塞进唯一一个抽屉;而关键词清单又会失败,因为没有两个人,会给同一样东西,贴同一个标签。

他的答案,是给信息换一种形状。别用树,用一张网:让任意一份文档,都能带着链接,指向任意另一份文档,落在任意一台计算机上,没有哪个中心,来裁定什么才可以相连。要阅读,你便跟随链接。这一张相互链接的文档之网,就是我们今天所说的万维网。

它从哪里来

早在 1980 年,伯纳斯-李就在 CERN 写过一个私用的小程序,叫 ENQUIRE,能把关于人和项目的笔记链接起来——那是这想法的一缕端倪。1989 年 3 月,他把它正经写成了《信息管理:一份提案》。他的主管迈克·森德尔,在顶上潦草地写下「含糊,但令人兴奋……」,而最要紧的是——给了他放手一试的余地。

他和一位同事,罗伯特·卡约,在 1990 年重订了提案,并动手去造:第一个浏览器、第一个网页服务器,都在一台乌黑利落的 NeXT 计算机上。第一个网页,1991 年上了线。接着,1993 年,CERN 做了那件决定性的事——它把万维网送了出去,任何人都能免费使用、在其上建造。正是那份开放,让它铺满了整个世界,而非困守为一座实验室的工具。

它为何重要

万维网胜出,不是因为它的设计最丰富——别人有过更宏大的构想——而是因为它简单、开放。任何人都能挂一个页面;任何人都能链向任何人;无需许可、不要执照、不付费用。那道极低的门槛,让千百万人都能往里添东西;而一样人人都能添的东西,长得比任何一家公司所能掌控的,都要快。不出十年,它便成了人类发布与阅读的寻常方式。

一个类比

把互联网想成道路与邮政的网络:线缆,以及给每一栋楼(每一台计算机)的一个地址。光靠它自己,不过是在机器间运包裹。而万维网,是跑在其上的邮政服务——一套共用的方式,去为一封「信」(一个页面)编址,无论它栖身谁的楼里,都能把它取来,还能在信里,塞进指向别处任何信件的指针。伯纳斯-李没有去铺路(那是互联网,出自 1974 年的瑟夫与卡恩);他发明的,是信,和信里的链接。在下方,亲手跟着几条链接走走看。

一幅可交互的节点—链接图,取自伯纳斯-李提案中的十份文档,由他亲笔标注的链接相连。你从「这份提案」出发;只有当前页面射出的链接可点。点一下,便走过去。目的地带着圆环;抵达时,部件把你的点击数与最短的可能路线相比。

它在知识长河中的位置

万维网,立在本馆中两个更早的想法之上。万尼瓦尔·布什(1945)最先构想出链接——把任意两份文档系在一起,沿着轨迹走下去——而泰德·尼尔森,给它取名「超文本」。它又跑在瑟夫与卡恩(1974)所造的互联网之上:后者给了每台机器一个地址,与一套传递数据包的办法。伯纳斯-李把这两者——链接,与网络——接成了一套简单、开放的系统。他造出的那张链接之图,后来喂养了为它排序的搜索引擎,也喂养了在它所承载的文本上训练出来的 AI 语言模型。他也曾警告:开放的万维网,已漂向少数几个巨头平台;他在努力,要把它推回那非中心化的根上。

The original document
Original source text
Tim Berners-Lee · CERN (ref. DD/OC) · March 1989; reissued May 1990
Abstract
This proposal concerns the management of general information about accelerators and experiments at CERN. It discusses the problems of loss of information about complex evolving systems and derives a solution based on a distributed hypertext system.
Losing information at CERN
If a CERN experiment were a static once-only development, all the information could be written in a big book. As it is, CERN is constantly changing as new ideas are produced, as new technology becomes available, and in order to get around unforeseen technical problems.
Over the two-year scale of a typical project, Berners-Lee argues, people arrive and leave, and the informal mesh of who-knows-what decays with them; the details end up either lost or locked in a single person's head. A laboratory of thousands needs its records to outlive the people who made them.
Why hierarchies and keywords fail
He examines the two filing methods then in use — tree-like hierarchies (as in CERN's own CERNDOC, or a computer's filesystem) and keyword indexes (as in VAX/NOTES). A tree forces every item into a single branch, so a fact that belongs in two places can be filed in only one; keyword systems fail because, as he puts it, two people scarcely ever choose the same word for the same thing.
A solution: linked information
A 'web' of notes with links (like references) between them is far more useful than a fixed hierarchical system.
Information becomes a graph of nodes — documents — joined by typed links: 'describes', 'includes', 'refers to', 'wrote'. No node is the centre; a link need not respect any hierarchy; anyone may add a link from anything to anything. The proposal's hand-drawn diagram shows exactly this tangle of boxes and labelled arrows, with ENQUIRE — his own 1980 program — sitting among them.
What was proposed
Berners-Lee asks CERN to build a distributed hypertext system: documents held on many networked machines, a program (a browser) to display a document and follow its links to wherever the target lives, and a universal scheme to address any document on any computer. It must be non-centralised and able to wrap existing systems, so it can be adopted piece by piece. Reissued in May 1990 — co-written with Robert Cailliau as 'WorldWideWeb: Proposal for a HyperText Project' — it led to the first browser and server on a NeXT machine; his manager Mike Sendall famously pencilled 'Vague but exciting…' on the cover and let it proceed.
[ … ]
CERN · March 1989