The public understanding of computer networks: The Danish case 1979-1999

Today, I presented a sub-part of my study of the Danish web in the 1990s at the 2025 ECREA Communication History Workshop at CERN. For those who are interested I have made the presentation downloadable at the buttom of this blog post.

The presentation is entitled ‘The public understanding of computer networks: The Danish case 1979-1999’, and the overall idea is to map how ‘publicly available computer networks’ have been understood in the Danish public sphere since the first ideas in the late 1970s to 1999 where my study’s period ends.

The reason why this question is relevant is that how computer networks are understood may have an impact on how actual existing computer networks are appropriated and used by their users.

Asking this question has implications for what is in- or excluded from the study.

First, I have to establish a dataset in which to analyse trendings of understandings of computer networks in the public. I selected three national Danish newspapers with geographic spread, different leanings (political, business friendly, etc.), and that have existed in the entire period 1979-99, and that were digitally available:

• Berlingske Tidende
• Information
• Jyllands-Posten

This choice begs a brief reflection on why the national newspaper Politiken was not included, and the reason for discarding Politiken is that they do not want to be included in the national Danish digitised newspaper collection, Mediastream (cf. below), and therefore they are only available on microfilm which is not searchable. Politiken can also be found in the media database Infomedia (initially owned by Politiken), but not for the entire period 1979-99, and only as text, not with the full newspaper layout, including images and illustrations. Apparently Politiken does not want to be studied so they’re out.

I could, of course, have chosen all available Danish newspapers, but this would demand more ressources than I have, so I ended up with a sample. However, random checks of other newspapers have shown that the ways of understanding computer networks are by and large identical with the ones found in the three selected newspapers. Radio, television, and video commercials could also have been included, but checks showed that there was not much to be found compared to newspapers.

I decided to not include more specialist magazines and newspapers such as Computerworld or Ingeniøren, because they are not considered as conveyers of the wider public’s understandings of computer networks, on the contrary, they are a bit more ‘nerdy’ first-movers, and what they write about usually finds its way to mainstream newspapers.

I also excluded articles about computer networks that were not aimed at the public, in particular networks within academia and in use by companies and corporations.

For this study I have searched the national Danish digitised newspaper collection, Mediastream, at the Royal Library — this is an invaluable source, and I can only recommend using it. However, just setting up the search proved to be more complicated than anticipated. I started with the term ‘computer’, but then it all exploded, there were simply too many articles containing that word. I also knew that the term ‘computer network’ would not be useful, because this is not the term that has been used by journalists and others writing in Danish newspapers. Therefore, I had to use synonyms, but they change over time, as the actual available or imagined computer networks evolve. For instance, in the beginning of the 1980 ‘Teledata’ or ‘datamat’ were relevant terms often used to identify publicly available computer networks, later I had to use BBS (Bulletin Board Systems, in use from app. 1986), later again ‘surf*’, and ‘Diatel’, until I finally reached the web in 1993. I ended up using 56 different search words, composing quite complicated search strings with NOT, OR, etc., and I even reached the limit of how many characters a search string could be in Mediastream — luckily the nice people at the Royal Library fixed this by extending this limit, thanks 😉

But there was constantly a lot of noice. A few examples:

• job ads, and ads for cars and housing were full of mentions of all synonyms, but I didn’t want to include them, however, studying the changing understandings of computer networks in job ads would be an interesting project in itself,
• and the same goes for general ads, but they were also discarded, but I kept some, just because they were relevant for my study as such,
• some of you may remember the French actor Brigitte Bardot, commonly abbreviated BB, but when writing about something belonging to her, it becomes BBs which is not what I was looking for when searching for ‘BBS’,
• FIDO was searched for, which is a famous BBS, but, of course, this is also a name often used for dogs, so I did get a lot of stuff related to dogs,
• the term ‘surf’ produced a lot of noice, I had to use it with an asterisk, like ‘surf*’, but then I did get ’surface’ and similar, and, of course, also anything related to surfing on water, so I had to include a long row of NOTs (NOT vacation, sail, strand, wind, etc.),
• web* also prooved to give a lot of noice, including mentions related to Andrew Lloyd Webber,
• also, throughout the period there were great differences related to spelling, you wouldn’t imagine how many ways one could spell online: ‘online’, ‘on-line’, ‘On-Line’, etc. — and I noticed a general point related to this, namely that in a period when a new media form emerged the spelling was characterised by uncertainty and different spelling forms,
• also, in the years following the advent of the web, app. 1994 onwards, the search terms ‘’www’ and web’ were used all over the place in newspaper articles, for instance adding simply a web address after an article or a small note, that was in itself not relevant for finding out what people understodd by ‘the web’.

Finally, I found out that the newspaper collection in Mediastream which is scanned from microfilm, does a rather bad job with OCR, Optical Character Recognition. All search terms returned a large number of articles of no relevance because of misspellings originating from bad OCR. I know making good OCR is difficult, but I didn’t know it was that bad.

After several weeks of work I ended with a corpus of around 4,000 articles covering the term ‘computer network’ and relevant synonyms 1979-99.

I then started reading and coding the articles, that is trying to identify various characteristics of the articles in a systematic and structured way. I coded for a number of things:

• formal¹ (placing in newspaper, article size, illustrations, journalistic genre, communicative actor),
• computer network in time/space (past/present/fu-ture; here/there),
• use forms and place (shopping, banking, information seeking/distributing, etc.; home, work, etc.),
• themes related to how the computer network affects individual life, specific groups, companies, society (positive/negative).

Unfortunately, it has taken forever to collect and code all these data, so I have only managed to study the first two years, from 1979 to 1981, but to give an impression of the developments I also included the fourth quarter of 1993 and 1999. Therefore, this is what it was possible to present at CERN. However, although this is a ‘pilot study’ it yielded some interesting and thought-provoking initial results.

First, the overall picture. The figure below shows the number of newspaper articles in percent, quarter by quarter. It can be noticed that the last quarter of 1979 is standing out, whereas computer networks are not talked much about in 1980. More articles are published in 1981, probably because this is the year before the launch of the first Danish computer network, Teledata, which ran as a pilot project 1982-84. A bit surprising that the number of published articles is relatively low in 1993 compared to te previous years, but in 1999, when the web has been around for app. 6 years the number of articles has exploded.

Second, I looked at the use forms of computer networks that I could identify in the newspaper articles, and I presented this as a heat map (see below). To the right you have the different use forms that were found in the newspaper articles, at the bottom time, quarter by quarter, and to the left the heat map: the darker the color, the more the specific use form is present in articles in the quarter in question.

As can be seen in the early years the number of different use forms is relatively limited, but five use forms are almost constantly present:

• shopping
• information
• advertising
• news
• email

At least to me it was somewhat surprising that the five use forms above where there from the very first writings about computer networks, and that they continue to be present throughout 20 years.

In the presentation there are some more findings, so I have uploaded the entire presentation (link below) in case you want to dig more into the results.

Presentation ‘The public understanding of computer networks: The Danish case 1979-1999’, presented 6 Feb at the 2025 ECREA Communication History Workshop at CERN.