Date: Wed, 09 Apr 1997 02:42:58 +0200
From: Herbert A Meyer

A Pull Media Story: From Memex to the Web

Herbert A Meyer

In the thirties, the meaning of the term computer has altered from the description of a person who carried out calculations by hand, to the description of a machine, which mostly performs the tasks in an independent manner. The program of pull media - the technical support of user controlled access to recorded documents - was formulated at this time, too.

Vannevar Bush, scientific advisor to US President Roosevelt during the Sec-ond World War, exposed this program with the draft of an integrated library work-place as early as at the beginning of the thirties. The aim of his concept was to ensure a more direct and quicker access to large volumes of stored scientific material than was possible with traditional library indices. Bush’s concept centred around the creation of an external memory analogous to human thoughts paths, furnished with ‘associative’ processing functions. The result of his research efforts was a fictitious machine, the Memex (memory extender). Thousands of documents were to be recorded onto micro-film and by choice projected onto outlined areas marked onto the tops of writing-tables. To this extent, the documents were planned as being variable, in that cross-links could be applied by the reader, and Bush called this process ‘the user building a trail’. The path developed as a result would be used later during further reading sessions, where it would more or less present itself automatically. The idea for what is characterized today as a ‘hyperlink’ was born with this concept. The technical implementation of ‘selection through association’ remained a utopian idea, but the drafting of the Memex had a centrally thematic influence on later to be developed actual constructed systems.

The formulation of the fundamental concept for pull media is regularly quoted as taking place in 1945, this being the year in which the by now famous paper ‘As we may think’ was first published. However, in his autobiography, Bush himself dated the first project concept as 1932. The technological utopia of Memex was first referred to by name and subsequently described in detail in an unpublished research paper from 1939.

The reason for the lack of feasibility of translating ‘associative’ micro-film archives into a practical reality was because Bush favoring a non-digital process as part of the technical realization. His imagination was ‘analogue’. According to him, analogue computers retained the advantage, in that they are capable of directly reflecting things, even when only in a simplified manner. User controlled flexible access however, as seen in the experiences gathered in follow-up projects, would only succeed if the units of memory were made available in digitized form. First with the introduction of the mathematically formulated Turing-Machine which made possible main-frames capable of storing, transferring and calculating program loops simultaneously was it possible to device a digital filing system from an analogue micro-film archive which could be accessed flexibly via algorithmic calculation processes. This observation implies - from the perspective of information technology - an important difference between analogue and digitally represented documents: in the case of regular documents, storage and representation go together by force of requirement, but in the case of digitized documents these functions have to occur separately.

Inspired by ‘As we may think’ three decades later, Douglas Engelbart and Theodor Nelson began research work in an attempt to fulfil Vannevar Bush’s technical utopia. Their efforts were essentially assisted by the now advanced state of computer technology - Turing’s paper machine had developed into digital-electronically functioning main-frame computers according von Neumann by the 1950’s. The projection areas built into normal writing tables of the Memex era were transformed into virtual monitor windows in Engelbart’s NLS/Augment-System. Here documents could be filtered out of a database inventory using a devise similar to a mouse. In 1965, Nelson devised the term ‘hypertext’ to relate to a document administrated in this way. Two years after Nelson’s introduction of this term, and following his succinct comment ‘the hardware is ready’ the corresponding software was made available. Andries van Dam from Brown University however reserved the right to demonstrate the first functioning hypertext system in 1967 in the form of The Hypertext Editing System. This system was financed by IBM and ran with 128K of working memory on a /360 main-frame computer. It was utilized as a documentation system for the Apollo Space Program. Engelbart’s system reached completion one year later.

Hypertext research, still characterized by individual representations until the middle of the eighties became increasingly institutionalized. First in 1987 - the year of the first international conference and the introduction of the hypertext system HyperCard, supplied free of charge with the purchase of a Macintosh - could evidence of organized scientific interest in the subject be seen. As a result of intensive research effort, the development of a variety of hypertext systems followed, with technical performances increasing steadily in quality from version to version. First of all, these systems were bound-in to fixed base computers, and later into decentrally networked computer systems (distributed hypertext systems).

1992 is the year in which the European Centre for Particle Physics (CERN) first publicly presented the hypertext system World Wide Web, subsequently distributing it via the Internet as freeware. As such, this year is to be counted as an important turning-point in a still very young history. A standard for the programming of hypertext documents was established with this system and it spread like wild fire because it could be used on any computer irrespective of its type or construction. From 1993 onwards, it was possible to see the web indexed on a graphics oriented user interface which could be serviced with the click of a mouse. The broad usage and enormous popularity of the web and the large volume of globally distributed and stored hypertext documents have developed directly from this.

As demonstrated in the sketches above, hypertext is made particularly viable through the dissociation of the storage and representation functions. The technical foundation for this is digitization; the reduction of print, sound, image and moving image as a result of the electrical conditions of ‘on’ and ‘off’. In both local and open hypertext systems this enables the storage, calculation and representation of data formats from various media on a single platform. In parallel to this, digitization guarantees the compatibility of the infrastructure, enabling hyperlinks to transport programmed flows of data.

Generally speaking, electronic documents can be defined as documents which are published and distributed in digitized form, through which three processing functions are guaranteed within a computer program: storage, representation and transfer. Hypertext documents can in reference to this, be defined as a particular mix of electronic documents, being exactly those which dispose of digitally-electronic supported references. Hypertext systems or pull media, these expressions being used interchangeably, are correspondingly descriptive of electronic databases, over which a graph of possible gateways to chosen parts of the bases are stretched. As a result, defined digits are thus functionally linked and an interactive convolution with individual parts of the bases is made viable. This is exactly the mode of interactivity that makes hypertext so interesting. While in the case of multimedia, where the synchronous coupling of text, sound, image or moving image is exclusively important, pull media goes one step further, with the point of emphasis being placed on the synchronous networking of electronic documents. The is new and distinguishes pull media from all other media forms.

Books, magazines, newspapers, television, radio and film as a rule provide their material diachronically in the form of fixed sequences. Narratives are related which imply a fixed time structure. Exactly this linear orientation to the time axis can - but does not have to - be circumnavigated by pull media. This means for example that we are not required to tune in anymore when a program is broadcast, but can search for the program ourselves. And more - we can even modify the fine structure of the chosen program. These attributes; the search for stored material (selection) and the dynamic contact with stored material in dialogic form (compilation), characterize pull media. One should not fail to recognize however, that real-time access to digital databases is not unrestricted, but pre-programmed through hyperlinks.

Pull media is understood absolutely as a technical function, but not as an organizational principle. With hypertextual structures the organizational principle can be many-sided; simple or complex, linear or non-linear, hierarchical or networked. In a computer system, a hypertextual form already exists when two documents are linked to one another via a single hyperlink. Condition is that the documents have to exist in digitized form, otherwise the link will not function. At first sight, this view may appear trivial; what is important here however is to highlight the fact that a recourse to the term text does not provide an acceptable expression for an understanding of hypertext. Hypertext systems are administrative systems which organize the access - within local or networked systems - to inventories.