The Case of NRC Handelsblad
Peter Vasterman, Peter Verweij (ed.), Computer Assisted Research and Reporting. Phaedon, 1994. ISBN 90-72456-89-0. pp. 24 – 33.
Among all the cheering stories about the electronic superhighway and all its supposed blessings for journalism I would like to take a more down to earth approach: what can the computer do for the journalist working at a Dutch newspaper? What kind of online resources are available? What’s the advantage of the electronic version of the morgue? I’ll discuss the application of computer graphics, the use of spreadsheets, database programs and of course: Internet and CompuServe. At the end I’ll give my view on the future of computer assisted research and reporting in the Netherlands.15
The editorial system
NRC Handelsblad was quite late in introducing computers in the newsroom. In the mid-eighties we started with a limited number of terminals. After two years a System Integrators system was installed, still operating at the end of 1994. The heart of that system is an article database. Of every article about 150 attributes are kept in a separate file, which makes it not only very easy to select articles, but also to manage access levels. Much has been done to develop a Dutch translation for the user interface metaphors to make the system as user-friendly as possible.
The editorial system is both a communication system and a production system. For intemal communication there is an intemal messaging option, allowing messages up to 256 characters to be sent to other users. Longer messages have to be prepared in the form of an article. Users can give others access to articles they wrote by changing one of the attributes, or, to use our metaphors, they can put the article in another basket, e.g. in the home desk basket, the sports basket or the foreign desk basket.
Wire services (ANP, AP, AFP, DPA, Reuters and UPI) and some other organizations send their takes through an X. 25 port. The routing process offers very versatile possibilities for selecting and filtering incoming messages. Apart from a set of filters for the whole staff, every user can set his own to trim the system to his needs.
The terminals we use contain a Motorola 68000 processor and local memory. They are configured in such a way that they can only load the dedicated word processor. Four years ago it was decided not to implement the possibility to load a terminal emulation program, which would facilitate’external communication from the same terminal. This decision not only prevented access to external resources, but also blocked access to our own electronic morgue, which was not yet in use at the time of the introduction of the system.
The editorial system can be accessed from outside through personal computers. After having used a system with very limited capabilities for a couple of years, we developed a program that emulates the dedicated terminals. It was written in the script language of the communication program Telix. There is also a Mac version, written in White Knight’s script language. These applications allow users to do anything at home or on the road that they can do in the newsroom. Exceptions are typesetting and some system management operations. The program opens Word Perfect as a Standard word processor, but any other word processor can be used as long as it is able to export plain ascii files with IBM extended ascii for accents. Through Tymnet and Infonet, users have world wide access to the editorial system with a local or national phone call.
There are próbably few newspapers that went this far in opening up their editorial system. Of course there are risks involved. The main obstacles for hackers are the secret phone number to access the system and the need to have a name and password of an authorized user. In little over three years we never had an unauthorized user inside. But even if someone would break in, it would be almost impossible to disturb the production process. Using Telix’ host mode we also developed an electronic P.O. box for non-regular contributors to the paper. This runs on a PC that acts as a firewall. Through this box it is not possible to access the editorial system itself.
The electronic morgue
Until 1991 clippings from our own newspaper were kept on paper and, for older ones, on microfilm. For articles older than 1990 microfilm is still the medium. Numerical content codes show the way to the desired clippings. But since 1991 we have an electronic morgue, which contains all articles since 1990. The articles are described with a Standard description language to mark headers, bylines, et cetera (Standard Graphic Markup Language), which makes it easy to convert a retrieved article to a Standard layout for printing.
The electronic morgue cannot be accessed from our terminals. When a reporter wants one or more articles published in his own newspaper, he has to go to a librarian and ask him to retrieve the articles. The librarians access the electronic morgue using a personal computer as a terminal. There are several ways to search the morgue. The one used most by our librarians is through the old numerical subject codes: that is what they were familiar with. But the morgue can also be searched full text. Apart from our four librarians the only person that has direct access to the morgue is the assistant of our political desk, situated in The Hague. She can log on over a phone line.
It is clear that this situation is far from ideal. When reporters cannot access the morgue themselves, they are not able to browse, or follow hunches. Having to ask the librarians puts up a threshold: one only asks if one is really looking for something. It also takes a lot of time: there are four librarians and over a hundred reporters, so they cannot always do your searches immediately. Another disadvantage of this way of working is that the librarians spend much of their time doing simple searches in our own archives and labeling articles with numerical subject codes. That is a tremendous waste of skills and searching capacity: they could have spent that time on more complex searches in extemal databases, which require more specifïc knowledge about query languages and database contents. The future lies in direct access to our own archives for all reporters. However, it is not yet clear how and when this will be realized. Finally, we do not have a central archive for clippings from other media, and reports from companies and organizations. Every reporter keeps his own archive (or not).
Before our librarians could do online searches in our own archive, they already had access to several online resources. We started to use external databases around 1988, primarily for the economics desk. Datastream was among the first suppliers. Others, like Lexis/Nexis, Profile and Dialogue, followed soon. Through Nexis our librarians had access to articles from NRC Handelsblad before we had our own electronic archive.
Now our librarians have access to all major international databases. The initiative for a search has to come from a reporter. The librarians generally do not do any searches without explicit request. Together they answer about 1,200 questions a month. About 15 percent can be answered by using traditional reference books or the old clippings archives, 50 percent by using our electronic morgue, while the remaining 35 percent requires the use of external electronic sources. Lexis/Nexis is the most important one (60 percent, mostly international news), then Profile (20 percent, mainly news) and the Dutch Press Database (15 percent), which contains material from almost all Dutch national newspapers and the ANP, the national press agency. The database of the Dutch parliament Parac is a small contributor. Questions that require an Internet search are growing in number.
Apart from our librarians only a few people have direct access to extemal databases: some business reporters, the assistant of the political desk and some political reporters.
Business reporters have a direct connection to the Reuters fïnancial database. For this purpose a Reuters terminal is running at the business desk in the newsroom in Rotterdam and at the stock exchange desk in our sub office in Amsterdam. We have had access to the Reuters system since 1991 and are among the few general news media in the Netherlands that use this service. The main advantage is that fïnancial news is available faster through this service than through the general wire services, and it shows any kind of stock quotes from all over the world. Four people at the business desk know how to use the service and use it on a daily basis.
At the same terminal they have access to Reuters Business Briefings, a Canadian database with clippings from newspapers all over the world translated into English. It also contains clippings from Dutch newspapers. However, our business reporters use this service mainly for fïnding background information on Dutch companies’ activities abroad and for getting background information on a company when this is needed really fast. For example, the Akzo Nobel merger was announced 90 minutes before deadline. Reuters Business Briefings made it possible to prepare an extensive profile of the Swedish company Nobel in time for the paper of that day.
In our sub office in The Hague, home of the political desk, we have a personal computer with access through a phone line to our own archive and to the clearinghouse of govemment databases RCC (Rijks Computer Centrum). The assistant was trained to work with Parac, the database of the Dutch parliament, which is one of the databases accessible through RCC. Some political reporters have trained themselves to do queries in Parac. The other databases of the RCC are hardly being used, mainly because hardly any reporter knows of their existence and their contents.
It may seem rather queer that reporters in sub offices can do more with external databases than their colleagues in the main newsroom. There are two reasons for this. It is the systems editors’ policy not to have any other terminals next to our dedicated System Integrators terminals. And so far it is the editor’s policy to keep the distinction between searching databases and writing articles.
There is one online source that more reporters use (from home): the phone book. It is accessible through a public Videotex system. The advantage of this system is that it allows searching on address.
Home-made graphics started to appear in the paper around 1988, at fïrst only in the business section. A plotter was used to make -the word ‘design’ is not appropriate here- the graphs from data supplied by Datastream. The graphs were thought to be useful, but ugly, so a Mac was bought to make them look better. Through the years this was extended to a fully equipped graphics studio with four high end Macs, scanner and a color Linotype typesetter and a 600 dpi A3 laser printer for proofïng. Since about a year we use postscript typesetters for output. In the electronic page design system that is now being implemented, graphics are an integral part of the page output, just like photographs. Graphics, sometimes in full color, appear in the paper almost every day now.
The people working in the studio are artists. However, they also gather information themselves if they need it to make illustrations. Through the years they formed a substantial collection of maps, line drawings and clip art. They usually work with Adobe Illustrator and Photoshop. Graphical representation of information can make things clear at a glance that are difficult to describe in text. To avoid mistakes a close cooperation between reporter and artist is absolutely necessary, especially in producing graphs representing numbers. None of the people in the studio has a background in statistics and they do not know how to work with a spreadsheet. Reporters cannot make their own draft graphs, because they do not have access to the right computers, and they are not supposed to touch the Macs in the studio.
As far as I know there are only two reporters (one at the business desk and one at the political desk) that use spreadsheet software on a regular basis. They developed the know how on their previous jobs. One of them worked for a business magazine that was made on Macintosh computers, so the step to experiment with Excel was not a very big one. Because our terminals cannot run spreadsheets, he uses the application at home. From there he has online access to Datastream, from which he downloads stock quotes and other economic data. Through the years he has also built his own database with relevant historical fïnancial data. This allows him for instance to check on inflation tendencies himself instead of relying on some spokesman. It brings him closer to the sources of the news, he says. Programs used are Excel on a Macintosh and Supercalc for DOS on a PC.
The electronic availability of data on Dutch companies, e.g. from their annual reports, is limited. There is a cd-rom that contains such data, but we do not have it. This cd-rom with fïnancial data is not considered ideal, because the publisher involved converts all data to a Standard format. That way one is never sure whether the data shown are exactly the same as in the annual report, or were converted to make them comparable with other companies’.
We have a with a cd-rom drive in an experimental setting in the system editors’ room, but we do not have a production unit yet, neither in the newsroom, nor in the librarians’ room.
Because our terminals cannot run database software and we do not have any other computers in the newsroom (apart from the Reuters terminal at the business desk), reporters that want to use a database program have to do this on their home computer. Of course this creates a severe threshold, that limits the use of these programs to a couple of people with more than average computer literacy. As far as I know only two reporters use a database program: one uses DBase III for Dos, the other one (i.e. me) Lotus Approach for Windows and previously PC-File Plus for Dos.
In PC-File Plus for example we made a simple flat file database with people in boards and steering committees of social security organizations. The more than 2000 names were taken from annual reports of some thirty organizations. After a couple of days of reading and typing it was a matter of minutes to get overviews of the people that held two or more positions and the people that were in more than one committee with the same other person. In half an hour we could draw a complete network of the probably most influential people in this field. Many of them were interviewed, which led to a short series of articles about the world of social security executives. All in all, the database was not mentioned in the paper, and only used as a tooi to find the right people and to confirm the existence of the network we supposed there would be.
A big step forward was taken in spring 1994. After the municipal election in March we got the idea that it should be possible to show election results in a map, percentages or gains and losses of a particular party corresponding with a gray tone or color. We already got the results in a special format to be printed in the paper, so they had to be available in a more suitable digital form. The national press agency ANP, that collects the results immediately after the closing of the ballots, was willing to give us a floppy with the results of all municipalities as’soon as they had collected them, which was somewhere in the middle of the night. This meant we had about ten hours to convert the text file to a database format, analyze the results, find out what aspects of the results were interesting to show in maps, convert those tables to a format readable for a mapping program. And that is what we did at the elections for parliament in May.
The text file was tidied with a word processor until it was comma delimited ascii. In Lotus Approach we developed a relational database application to read, convert, sort and analyze the contents of that file. The reason to use Approach was simple: it was the program I happened to have on my home PC and was familiar with. Developing the whole setup and testing it with dummy data took about nine days. After that I had learned very much about Approach and could develop the whole thing from scratch in two days.
The whole procedure in the night after the parliamentary elections took four hours -Approach is easier to use than e.g. Foxpro or Paradox, but definitely slower- before we could export the most interesting data to Maplnfo. For an evening paper like ours this left enough time to make the appropriate maps.
The maps were considered a good illustration of the geographical aspects of the election results. That’s why we used the same setup again with the elections for the European Parliament in June. Because fewer parties participated in these elections the database files were smaller. And as the application was made a bit more efficient as well, the analysis took only two hours this time.
Later Approach was also used to analyzé data extracted from Central Bureau for Statistics (CBS) files. The CBS sells some of its data on floppy to a wider audience. We bought the files with statistical information per municipality, containing anything from demographic, economic and housing data to area covered with forest or lakes. With the data (15 megabytes) comes a dedicated browser, which can be used to select areas and data and put them in a .dbf, ascii or other format. These files are then analyzed in Approach, the results of which can be put in a map. So far only a map of the geographical distribution of divorce was published.
Maplnfo was acquired for the elections project. It runs on one of the Macs in the graphics studio. One of the (freelance) graphics designers leamed to work with it. With the program we bought a map of the Netherlands per municipality. So far we have only used maps in gray scales in the paper. Usually the graphics designer and the reporter that did the data analyses watch the screen images together to decide which boundary values are most suited to illustrate the results.
Public access to Internet is relatively new in the Netherlands. In May 1993 some reporters got an account at the first public access provider, Xs4all, rooted in the Dutch hackers’ scene. A science editor was the first who was able to write an article mainly on the basis of Internet material about the alleged proof of Fermat’s last theorem. A foreign editor and an international affairs commentator appreciated the immediate access to texts of Safety Council resolutions and U.S. presidential statements. For the home desk there was very little of use on the net, because there are very few Dutch resources available. Two home desk reporters roamed the net more to get an idea of the future perspectives than for actual daily use.
It is obvious that the net has important perspectives. That’s why in August 1994 we got our own host on the net, directly with the main commercial access provider. One of the fïrst things we did was give e-mail access to all reporters and editors. They can send and receive e-mail on, their terminals. The messages are being sent through a fïrewall PC in the same way we can send faxes or telex messages from the terminals. Apart from this e- mail access, about forty reporters have interactive access to the net, but maybe only five of them use it intensively. The unfriendly interface is a severe barrier for anyone with no particular computer affïnity.
At the moment, November 1994, we are preparing a switch to client server applications over a pseudo slip connection. Our own Web server is to be set up before the end of the year.
We have been using the net for getting information, but also for publishing. During a three months experiment a daily selection of articles was put into the electronic newsstand of De Digitale Stad, the Amsterdam freenet system opened in January 1994. The articles could, and still can, be read after selecting date and article in a Gopher menu, but can also be searched full text.
Since CompuServe has a local access point in Amsterdam it has become affordable for an individual to subscribe. A few reporters have done so. They work with CompuServe from their home computers, because the terminals in the newsroom cannot run the required software. Our medical reporter uses CompuServe to be able to access several medical databases without having to bother the librarians. I use CompuServe myself mostly for the software support forums. The Lotus support forum proved invaluable in developing the elections application.
Computer assisted reporting is a wide concept. I hope to have shown that NRC Handelsblad is developing this kind of reporting in various ways. However, there is no central policy, nothing is being coordinated. Part of the developments take place outside the newsroom -in the library department, which is not part of the editorial organization, or in the studio- and out of reporters’ direct reach. Another part is done by the systems editors. And yet another part is being developed by individual reporters following their own interests. To judge the perspectives of computer assisted reporting one has to look at several aspects: resources, hardware, editorial organization, knowledge embedding and attitude.
A very important difference between the United States and the Netherlands, or Europe, is the impact of the Freedom of Information Act. In the Netherlands there are very few public records. And the new privacy law restricts the access to such records for anyone but government institutions even further. There is a law that grants access to government policy documents, but there are many exceptions, and the procedures to get the documents requested are often long, complicated and expensive due to lawyer costs. News media make relatively little use of the possibilities of this law.
The govemment itself has not shown any tendency so far to make use of modem media like computer networks to make documents available to the public. The only exception is the city of Amsterdam, that gave access to its government information through the freenet mentioned earlier. But while it takes only minutes to get the text of a speech by president Clinton, it takes at least half an hour, a fax machine and good working relations with his spokesman to get the text of a speech by the Dutch prime minister. While it takes minutes to get U.S. Supreme Court decisions, it may take hours, or even days, to get their Dutch equivalents. As of yet there is no effective lobby for public access to govemment documents, to say nothing of other organizations’ information.
Another condition for a flourishing tradition in computer assisted reporting is suitable hardware in the newsroom. It is clear that the future is to open systems, but what this means for NRC Handelsblad has not yet been decided. I can only hope that the new system we’ll get in couple of years will allow every reporter to do searches in external databases, to send e-mail all over the world, to run spreadsheets, make their own graphs and use database programs. Only if people have access to these facilities, if they can play with it in a spare hour, if they see what colleagues do with it, one can hope that they will want to develop the skills and the knowledge required themselves.
But it is not only a matter of hardware and programs. At least as important is the editorial organization. For me it is obvious that the watershed between librarians and reporters will vanish. There is no principal difference between asking a database and asking a spokesman. Searching databases will be part of a reporter’s elementary skills in the future. Librarians do not need to fear for unemployment, because a lot remains for them to be done: complex searches in expensive databases, advising reporters on where to fïnd specific material, negotiating with suppliers, coordinating in house databases. But the difference with reporters skilled in database searching will be a gradual one. It is an open matter when and how NRC Handelsblad will facilitate the organizational changes necessary to allow such a development.
A further question is how to embed knowledge about all these new tools and resources within the editorial organization. What kinds of things must everyone know? Do you try to get all the knowledge with the users, or do you keep a team of whiz kids acting as consultants for reporters with complicated problems? How do you guarantee that knowledge developed by one reporter is made available for the rest of the organization? These are problems we have hardly addressed.
We hardly addressed these are problems until now. For now they seem luxury problems. But we definitely need to think about them on the road to more developed applications for computer assisted reporting.
A final point I want to address here is attitude, or if you want to consider it on a collective level, tradition. Dutch journalism is deeply rooted in a political tradition, in which opinions were valued higher than the facts. During the past two or three decades researching those facts became more important, but still, the attitude of Dutch journalists remains quite different from that of their American colleagues. There is still so much we do not want to know.The author works as a reporter and editor at the home desk of NRC Handelsblad. Opinions expressed in this article are his own, and not necessarily his employer’s.