Paper versus on-line

O'Hara and Sellen's paper

Kenton O'Hara and Abigail Sellen's paper "A comparison of reading paper and on-line documents" [O'Hara and Sellen (1997)] reports on a study they carried out in the Rank Xerox Research Centre in Cambridge. They gave ten volunteers from the staff at the Centre the task of summarising an article. Five of the subjects were given a printed copy of the article, some blank sheets paper to make notes on, and a sheet of paper to write the summary on. The other five were given a computer running Microsoft Word with three open documents displayed: one containing the article to be summarised and two with the heading "Notes" and "Summary" respectively. After the subjects had written a 200-300 word summary the experimenters questioned them about how they had gone about the task.

The findings that they consider to be most applicable to systems design were grouped under three headings: "Annotation while reading," "Movement within and between documents" and "Spatial layout".

Annotation while reading: The subjects reported that annotating while reading the article was important for developing an understanding of it. Those working with paper found it easy to annotate the article by writing on the printed copy, while those working on-line either found it difficult to add their own comments or did not try (saying later that they would have made notes on a printed version). The subjects working with paper also found it easy to make frequent notes on a separate sheet whilst reading, while the subjects working on-line made notes in a separate document only after long periods of reading.

Movement within and between documents: Even when they were writing, the subjects working with paper moved automatically between the article, their notes and their summary by moving the sheets with their hand. In contrast, the subjects working on-line found navigating the documents to be slow and distracting.

Spatial layout: The subjects working with paper were able to lay out the pages of their documents to get an overview of them; they were also able to arrange the documents so that they could read one while writing on another. The subjects working on-line found the screen sizes limiting; they couldn't fit all the information that they wanted on the screen at once and they had difficulty working with tasks that involved using more than one window.

O'Hara and Sellen conclude from their study that, although on-line tools provide valuable support for writing, "in the support of reading, and more specifically reading for the purpose of writing, this study has shown that the benefits of paper far outweigh those of on-line tools" [O'Hara and Sellen (1997) p. 6]. They then go on to look at ways in which the design of on-line information could be developed to offer the same types of support for reading displayed by the paper-based information, giving three examples of lessons that on-line information could learn from paper:

  • Recognise that annotation can be an integral part of reading and build support for these processes;**
  • The need to support quicker, more effortless navigation techniques;
  • The need to support more flexibility and control in spatial layout.

[**It is worth noting that most paper documents are not specifically designed to support annotation and only do so by accident. In their books on document design Schriver doesn't consider annotation at all [Schriver (1997)] and Burnett just mentions that sometimes "outside margins are even wider to provide space for note taking" [Burnett (2001) p. 250]. Should we also be looking at how best to support annotation in paper-based information?]

O'Hara and Sellen give details about their choice of paper-based and on-line set-ups and they acknowledge that "better systems and better interfaces may significantly alter the reading process" [O'Hara and Sellen (1997) p.2]. In contrast, the only background information that they give about their test subjects is that they were "members of staff from [the Rank Xerox Research Centre] laboratory who participated voluntarily" [O'Hara and Sellen (1997) p.2] and they seem to miss the point that a better understanding of the interface and system might similarly alter the reading process. This is a serious flaw that limits they extent to which their results can be generalised. O'Hara and Sellen's arguments only support the conclusion that in supporting the reading of people with the same sort of educational and IT background as the test subjects the advantages of paper far outweigh those of typical word processors. Given either a better on-line reading system or subject more skilled at reading on-line the results might have gone the other way.

The netdecisions Intranet site

I'll turn now to my work from June 2000 to August 2001 as an Information Analyst developing and maintaining the Intranet of netdecisions, a global e-commerce solutions provider. The netdecisions Intranet is a website that can only be accessed from the company's offices around the world and it covers: news; company policies and procedures; background information about netdecisions; contact information for netdecisions employees; and local information about the area around each office. Although this is an on-line system all employees have easy access to high quality printers so they can choose whether to read the information on-line or print it.

There is a strong on-line culture at netdecisions. The company builds e-commerce systems so the employees tend to come from IT backgrounds, and they spend most of their time working collaboratively over the computer network. As a result they get plenty of opportunity to learn how to take advantage of the properties of on-line information. Paper is well established as a medium for reading and writing in the wider culture, so they also get plenty of opportunity to learn how to take advantage of the properties of paper. This results in a wide variety of reading practices that reflects the wide variety of personal attitudes and working styles.

Do the practices of employee of netdecisions using the Intranet (in particular what they print and what they read on-line) show the same advantages and drawbacks of paper and on-line systems as found by O'Hara and Sellen?

Annotation while reading

Although there has been a lot of research into developing annotations systems (such as Monroe's work in on-line maths text books [Munroe (1998)]; Golovchinsky and Marshall's work on using annotation as a note taking and a navigational tool in e-books [Golovchinsky and Marshall (2000)]; and work by Hirotsu et al (1999) on annotating content from the World Wide Web) the netdecisions Intranet uses established Web technologies which have no support for annotating its content.

Printed documents do not change; the sentence you underline stays the same and the paragraph next to your marginal note doesn't move, and this is one of the properties of paper-based information that makes it easy to annotate. Employees expect the Intranet to be kept up to date; they expect it to change. So, although employees do occasionally print out information to annotate it (in much the same way as O'Hara et al (1998) found that graduate students photocopied library documents to annotate them), the printouts (and their annotations) are limited in their usefulness by their potential to go out of date and are soon discarded.

Even if the information was published in a paper-based form, such as a staff handbook, and staff could make notes on the pages, it would need to be updated daily and those notes would be lost when the pages the notes were written on were revised. So it is not just the fact that the Intranet is an on-line system that gets in the way of annotation, but the dynamic nature of the information means that there are no fixed points of reference for annotations to be associated with.

Also, the text on the Intranet is written to be easy to read and employees usually dip into it to get a specific piece of information, so the annotation to support understanding isn't required in the way that it is for the reading to summarise in the study.

Unlike the subjects in the study, the employees of netdecisions have few problems writing notes in one window while reading an on-line document in another. When I (as an example of a not particularly IT adept netdecisions employee) was given the job of analysing the content of the Intranet I had a Web browser displaying the Intranet open in one window and the analysis I was writing in another. I was able to switch between the Intranet and the analysis document effortlessly and I was able to automatically copy text verbatim from the Intranet to the analysis (which would have been laborious with a paper-based system).

Movement within and between documents

The computers at netdecisions run Microsoft Windows NT 4 which has a Taskbar running along the bottom of the screen containing a button for each open application. Moving between documents just involves clicking the appropriate buttons, and the speed of the computers and the network means that there is no noticeable delay in navigation. The staff are skilled at using this system so they find navigation between documents effortless.

The benefits of paper-based versus on-line information when navigating within documents depends on the format of the documents. In netdecisions the four dominant formats for on-line information are: websites (such as the Intranet), Microsoft PowerPoint documents, Microsoft Word documents and Adobe PDF files. Employees almost always read websites (including the Intranet) on-line, their non linear structure make them difficult to print onto a linear series of sheets of paper, and once printed they lose the interactivity that their navigation is based on. Microsoft PowerPoint documents are normally read on-line but they are sometimes printed. Microsoft Word documents are normally printed if they are going to be read in any detail but many employees are happy to read them on line. PDF documents are almost always printed out to be read because they can be awkward to read on-line.

The general principle seems to be that for information that is designed to be read on-line (such as Websites and PowerPoint presentations) the benefits of navigating them on-line outweigh the benefits of printing them to read them, and for information designed to be read on paper (such as MS Word and PDF documents) the benefits of navigating them on paper outweigh the benefits of navigating them on-line.

Spatial Layout

Being familiar with Microsoft Windows and the World Wide Web is a basic recruitment requirement at netdecisions and all employees are comfortable in the virtual space inhabited by on-line documents. They are also comfortable in the physical space of paper documents. Every employee has a networked PC and a desk in a crowded office. It is relatively easy to navigate between up to about fifteen open documents on the PC and difficult to work with more than about four paper documents on a desk that already has a monitor, keyboard and mouse mat on it.


To sum up, we use the Intranet because it makes it easy to publish constantly updated information to a worldwide audience. Although this means that the information is delivered online, all employees have easy access to fast, high quality printers. Given O'Hara and Sellen's results we might expect the employees to print out information from the Intranet to read them. This doesn't happen because:

  • The Intranet has a better interface for reading than the word processor in the study;
  • The staff at netdecisions find navigating within and between on-line documents effortless;
  • The way that the text is read and the way it is constantly updated greatly reduces the value of direct annotation on the text;
  • The employees' ability to switch easily between the Intranet and other applications makes on-line annotation in a separate document effortless.

So, in the support of reading the information in the netdecisions Intranet the benefits of on-line tools outweigh those of paper.


Burnett, R E (2001) Technical Communication, Harcourt College Publishers

Golovchinsky, G and Marshall C (2000) ‘Hypertext interactivity: from choice to participation', New Review of Hypermedia and Multimedia, 6 (2000), pp.169-196

Hirotsu, T, Takada, T, Aoyagi S, Sato, K and Sugawara, T (1999) ‘Cmew/U - A multimedia Web annotation sharing system', Proceeding of the IEEE Region 10 Conference (TENCON'99), pp. 356 - 359

Munro, N (1998) ‘Interactive textbooks for control', Measurement and Control, 31(5), p.146-150

O'Hara, K and Sellen, A (1997) ‘A comparison of reading paper and on-line documents', Proceedings of CHI'97, Human Factors in Computer Systems, Atlanta, Georgia, pp. 335-342

O'Hara, K, Smith, F, Newman, F and Sellen, A (1998) ‘Student Readers' Use of Library Documents: Implications for Library Technologies', Proceedings of CHI'98, Human Factors in Computing Systems, Los Angeles, California, pp 233-240

Schriver, K A (1997) Dynamics in Document Design, John Wiley & Sons