2007-03-04
It is sometimes claimed that the World Wide Web was supposed to be text-only or at least primarily text-based. This claim could be related to an anecdote purportedly originating with Marc Andreessen. Here’s how USA Today told the story of how the web went graphical:
Kevin Maney, ‘10 Years Ago, Who Knew What His Code Would Do’, USA Today (2003-09-03, updated 2003-10-03). Accessed 2007-02-03.In 1993, the Internet was almost solely used by academic research scientists and the military. Navigating it required memorizing arcane text commands. Only a few years before, in a research lab in Switzerland, Tim Berners-Lee created the hypertext links that formed the basis for the World Wide Web, but that was still text-only and meant for research.
No one had created a visual way to navigate the Net. There was no way to put up images.…
Andreessen, Totic [sic for Aleks Totiç], Jon Mittelhauser and a cabal of students worked part-time at the university’s famed computer lab, the National Center for Supercomputing Applications (NCSA). There, the idea of a visual browser bubbled to the surface. Andreessen and fellow NCSA worker Eric Bina grabbed it. The concept, Andreessen says,
was just there, waiting for somebody to actually do it.The two slammed together the code for the first graphical browser. On March 14, 1993, Andreessen put it on NCSA’s Internet site. He introduced it:
NCSA Mosaic provides a consistent and easy-to-use hypermedia-based interface into a wide variety of information sources.…For Andreessen, the thrill faded as Mosaic caught on. NCSA exerted more control over its development, and the existing Internet community lashed out at Mosaic. At a conference, Berners-Lee yelled at Andreessen, telling him that adding images to the Web was going to bring in a flood of new users who would do things like post photos of nude women.
He was right, Andreessen now says with a shrug.
But consider this anecdote from Silvano de Gennaro, who runs the CERN Music Club:
Back in 1992, after their show at the CERN Hardronic Festival, my colleague Tim Berners-Lee asked me for a few scanned photos of
the CERN girlsto publish them on some sort of information system he had just invented, called theWorld Wide Web. I had only a vague idea of what that was, but I scanned some photos on my Mac and FTPed them to Tim’s now famous info.cern.ch. How was I to know that I was passing an historical milestone, as the one above was the first picture ever to be clicked on in a web browser!
It’s difficult to reconcile the Berners-Lees recollected in these two anecdotes. But regardless of whether Berners-Lee did blame Andreessen for the coming onslaught of pornography, the notion espoused by the article that the Web was intended to be hypertext, never hypermedia, is demonstrably an urban legend.
The World Wide Web Project was conceived on the back of decades of hypertext speculation, research, and experimentation that recognized that hyperdocument systems would not be restricted to text. Ted Nelson coined the term ‘hypermedia’ in the same article as he coined ‘hypertext’, giving ‘hyperfilm’ as an example:
T. H. Nelson, ‘A File Structure for the Complex, the Changing, and the Indeterminate, Proceedings of the ACM 20th National Conference (1965), 96.Films, sound recordings, and video recordings are also linear strings, basically for mechanical reasons. But these, too, can now be arranged as non-linear systems — for instance, lattices — for editing purposes, or for display with different emphasis. …The hyperfilm — a browsable or vari-sequenced movie — is only one of the possible hypermedia that require our attention.
Some existing hypertext systems, such as Apple’s HyperCard and Pei-Yuan Wei’s Viola, already supported multimedia.
In his early design notes on document formats for the World Wide Web project, Berners-Lee envisaged that his universal hypertext system
would support graphics, video and sound clips, object-oriented graphics definitions
across a wide range of user agents through content negotiation and transformation.
In 1990, at a time when many computer users were stuck with text consoles, Berners-Lee had the luxury of developing the first ever web browser with NEXTSTEP. WorldWideWeb, the result, was a essentially a WYSIWIG editor to which Berners-Lee added support for hyperlinks. Right from the start, WorldWideWeb supported images, but at first they were displayed in separate windows not inline.
Most of the public technical discussion of the World Wide Web Project took place on tthe www-talk mailing list. Sadly, a great many of the early messages have been lost. Fragmentary selections are preserved at the W3C’s archive and has collected the World Wide Web History Project.
But the messages that do survive show that Andreessen was by no means the first to propose inline graphics in HTML, never mind a lone voice advocating multimedia on the web. In fact, hypermedia was a topic of regular discussion. On 1992-06-03, Jim Davis wrote the list about his desire to include non-text data (e.g. bitmaps, sound) in my hyperdocuments
, proposing a new BITMAP tag and suggesting that browsers could handle images by either launching a viewer or rendering the content inline. Dan Connolly’s proposed DTD of 1992-06-24 actually included an IMAGE tag. Yet another request for embedded images came from Dave Raggett on 1992-11-25:
We want to have documents with pictures in them, that appear and are formatted as part of the document. A reasonable way of doing this would be to include references in html documents to objects which should be displayed as part of the document, but which aren't part of html itself. It seems sensible to support a variety of formats: TIFF, GIFF, DIBs Fax group III/IV, CGM etc. The idea of embedded objects which return their size and can display themselves at a given location seems appropriate.
In so far as there was a debate about hypermedia, it was seemingly never about whether the web would include hypermedia, but always about how, even though some people, including Wei and Robert Raisch, worried about interoperability with console-bound HTML clients. There were repeated discussions about the suitability of MIME types to faciliate multimedia content negotiation. Berners-Lee’s preferred solution was for multimedia to be referenced or included via ordinary links with special link types rather than dedicated tags.
As of 1992-11-03, when a snapshot of the Project’s own website was taken, it defined the World Wide Web as a wide-area hypermedia information retrieval initiative aiming to give universal access to a large universe of documents
, and clarified that hypermedia was a term used for hypertext which is not constrained to be text: it can include graphics, video and sound, for example
.
When Andreessen proposed IMG as a new HTML tag for inline images on 1993-02-25, the ensuing debate was not about the scandalous uses to which IMG might be put, but a triumphantly geeky rerun of earlier discussions about MIME types and markup, with Berners-Lee briefly holding out for link types rather than media-specific tags. Andreessen’s trump card was that while the Wei and Tony Johnson (developer of Midas) were also working on inline images, the Mosaic team had already implemented IMG. So in the event Berners-Lee and Dan Connolly included the IMG element in their HTML draft in June 1993. The draft described HTML as a language for representing, among other documents, collorative hypermedia
and simple structured documents with inlined graphics
.