Introduction
Recent advances in display technology include the development of both large and flat displays: CRT technology is approaching laser printer resolution; flat screen technologies are making rapid strides, mostly due to the pop ularity of portable displays and the perceived attractiveness of a large flat screen television. Such technologies enable one to rethink the way in which electronic information is accessed.
Figure 1: The large monitor as it's being used today.
A broadsheet-sized color monitor has been used as the primary display on one of our workstations for more than two years. The display has four times more pixels than an average X11 workstation, and it is used differently. The difference cannot be measured in square centimeters only; when using the large display, screen space management shifts emphasis from screen area conservation to screen overview. Scale becomes available as a design element. Mechanisms such as peripheral vision become available as a communication channel [Mollitor 90]. See figure 2. At the same time, increased display size and density can be problematic, for example, increasing the possibility of competition for the user's attention.
Figure 2: The large screen leaves more of the screen surface in the peripheral vision--a challenge and opportunity. The large screen is compared with a standard workstation and PC display.
Through the X11 window system, the screen gives access to all standard computer applications, e.g., electronic mail, word processors, drawing programs, and games-lots of them! Of special interest is the use of the monitor to present news in electronic form-the screen's similarities with a newspaper front page makes the newspaper format a natural starting point.
Digital news
Most people prefer to receive their news on printed paper instead of using electronic sources and computer displays. The paper-based version is cheap er, it's easier to fold over a breakfast table, and it's illustrated. The front page gives readers an instant overview over the most important stories, and the headlines make it possible to scan large amounts of information quickly. Paper-based news distribution has a long tradition and centuries of experience stand behind today's formats. Pages, headlines, columns, and fonts have been tuned in form and function. They all are a part of a user-friendly and universally accepted product [Gurtler 84].
Current screen-based news presentation lacks many of these properties. Limitations in display technology have barred the presentation of a full-sized page with attractive text and images of acceptable quality. The single biggest obstacle for computer displays to compete with paper is the physical proper ties of the media. While the resolution of computer displays begin to rival paper for the first time, no known display technology approaches all properties of paper-even if we extrapolate ten or twenty years ahead. Therefore, in or der to compete with paper, a screen-based electronic newspaper must offer functionality not available in traditional newsprint, e.g. personalization, dynamics, user participation, and navigation.
Context
Over the past years the Electronic Publishing Group in the MIT Media Lab oratory has been conducting a series of media experiments that explore personalized newscasts. These projects, which address issues of content analysis, user modeling, data representation and distribution, and presentation, are described in [Lippman 86], [Lippman, Bender 87], and [Bender, Chesnais 88] and are collectively referred to as "NewsPeek". The most recent project, "Newspace" is an attempt to take advantage of advances in display technology to create an electronic newspaper that is scalable onto a wide range of display devices [Bender et al. 91]. The project addresses news gathering, manipulation, and presentation; The Electronic Broadsheet is the presentation module for a large display under the Newspace project.
Other projects [Hoffert, Gretsch 91] [Erickson, Salomon 91], in addition to the Newspeek projects, have also taken news in digital form and used computer screens for presentation. The innovation in this project is the size and nature of the screen; we explore the impact large screens have on electronic information presentation.
Organization of this paper
Section 1 discusses traditional interfaces to newspapers and computers, and how an electronic newspaper can take advantage of both. The rest of the paper describes resources currently available and how the Electronic Broadsheet was implemented under current constraints. We conclude with observations about the potential for success and failure of very high resolution displays. The discussion is based upon the authors' experiences with both the electronic newspaper application, and the scaled up version of the X11 window system in which it is couched.
Information interfaces
The user interface of newspapers has been developed and standardized throughout centuries. Despite sociological differences, publishers and editors from different parts of the world can meet to discuss the content, role, and technology of newspapers-just as readers from different parts of the world can pick up a local paper and immediately know how to read it if the written language is known. The different elements of the newspaper interface are collectively known as the "newspaper metaphor".
The newspaper interface
The basic building blocks of the newspaper (headlines, columns, the name plate, front page etc.) turn into an advanced user interface when skilled editors and typographers collaborate on the product. When preparing the news paper, and especially the front page, editors process the information to accommodate all readers. By scanning the front page, the reader can get an overview of the most important issues in a matter of seconds. Large quantities of news can be searched with little or no predetermined focus. The process of scanning helps the user to uncover and read articles of interest, with out incurring significant overhead on the part of the reader either in terms of time or effort. Minimum effort is required to digest 10% of the print, and there is little penalty in skipping the remaining 90%.
Simply by switching from scanning to reading, the user is able to change modality from overview to detail. The front page is your menu in which the selections are available immediately. If the reader finds that the wrong selection was made, i.e., the wrong article is being read, the menu is still there to choose from.
The newspaper is a static medium for one way communication only. Still, by turning to a section page a reader can drastically change the content of the page being looked at. The stories on the page become more specialized, and in one sense the reader has interacted with the editors of the newspaper. By turning to the section, the reader requests more information, and the editor provides more coverage. Of course, the interaction is predefined and limited to the content of the newspaper edition.
Newspapers provide a facile and forgiving interface by not demanding anything from the reader, and they update themselves without requests. It's a forgiving interface; even if a subscriber ignores the paper for a week, the journalists and editors will still produce new editions.
The computer interface
There are fundamental differences between the newspaper interface and that of the usual computer information retrieval system. Computers offer largely sequential access under direct control of the user. No assumptions are made by such systems as to the user's intent, therefore the presentation of the data is not tightly coupled to its retrieval.
Computer interfaces also make use of menus of various kinds where the user chooses a specific element from a list, after which the computer executes the selection. Typographical cues are rarely used to differentiate items. If the computer performs a successful search, the result is displayed on a small screen with few typographical clues. Then the computer requires the user to issue several commands, like pan or scroll, to show the full result of the retrieval. In comparison with a newspaper front page, the computer menu falls short.
Modern computer programs are event-driven, i.e., nothing will happen unless the user issues commands, e.g., a mouse click or button press. The interface keeps the user active and in control, and if the user is inactive, most computer programs will remain passive. Editions of a newspaper, in comparison, will keep coming without user feedback.
While newspapers are a universally accepted and understood product, computer code requires specific hardware and specific versions of the system software to run properly. The user interface of computer programs varies enormously from system to system. Even programs based on the popular "desktop metaphor" cannot be interchanged freely-neither by computers nor humans.
Benefits from using dynamic displays
As we have seen, traditional newspapers often compare favorably with hi- tech information systems. Trying to replicate the best parts of the newspaper metaphor while taking advantage of dynamic displays is an important goal in The Electronic Broadsheet.
The newspaper format has its shortcomings. Newsprint is a static medium. After the ink is put on paper it doesn't move-except onto the fingers of the reader. This has two important implications: articles don't get updated and the medium can't show moving pictures.
By entering the electronic domain, newspapers can add continuous updates and video sequences to the presentation. Digital technology also opens for other enhancements, e.g., personalization of content, two-way communication between information producer and consumer and better navigational clues. These benefits apply not only to the presentation of news, but also to other forms of screen-based electronic publishing. See [Lie 91] for a more thorough discussion.
From pixels to pages
The Electronic Broadsheet is an attempt to transcode the newspaper metaphor into an electronic medium using state-of-the-art presentation technology. The following chapter describes and evaluates the hardware and software components involved in building pages from pixels. Emphasis is put on issues where displays differ from paper.
The monitor
The continuously advancing technology has provided us with "paper-like" displays that we claim can start competing with newsprint. The monitor used in the project is a color CRT showing 2048 * 2048 pixels on a 20" * 20" (508 * 508 mm) viewing area. This is almost as high as a broadsheet newspaper and a little wider. The image consists of 4 million pixels and is refreshed at 60Hz.
When seeing the display for the first time most people are struck by its size and resolution. The monitor frame measures 694 * 673 * 760 mm (w/h/d) and weights ca 98 kg [SONY 89]. To make this huge piece of glass and metal as flexible and portable as paper is impossible; at least three people are required to move it. It is a unique device, but its physical properties leave much to be desired. It may be suitable for a Chinese wall newspaper, but it will not fit on your desktop.
Merely considering the technical specifications, it may not be clear that the display really is able to take on newsprint with regard to legibility. A resolution of 2k over 20" yields around 100 dots per inch (dpi). A typical laser print er outputs 300dpi. To close this gap a technology known as soft fonts is used to render all text.
Soft fonts
Soft fonts, also known as fuzzy fonts, antialiased fonts or grayscale fonts in troduce a new way of thinking about text on computer displays [Negroponte 80] [Warnock 80] [Schmandt 80] [Bigelow, Day 83]. The monitor is no longer considered a discrete device with a fixed matrix driving it. Instead, the characters are scaled onto a continuous space; any partly covered pixel by the edge of a character is quantized into a grayscale value. Soft fonts don't improve resolution, but rather, improve addressability of the existing resolu tion. This is important to properly render the letter forms, as well as position the letter forms on the display. Ergonomic studies show that they are easier to read [Bender et al. 87], and without the use of soft fonts on the display it would be much harder to claim competitiveness with paper.
Legibility vs. text density
A dilemma typographers face when laying out text is legibility vs. text density. Newspapers opt for high text density while legibility suffers. A good example is the front page of NYT [Merill 80]. There is a minimum of white space, and headline fonts are often condensed. Margins are minimal, and the overall impression is "dark". When presenting news on a dynamic display, space comes cheaper than on paper. The Electronic Broadsheet can afford to emphasize legibility instead of maximizing text density. One example of this is how paragraphs are displayed. While newspapers break paragraphs to align columns and minimize white space, our electronic version always display paragraphs in one piece. This idea, used in the "SuperBook" project [Egan et al. 89], sacrifices real-estate for legibility. Although readers always have accepted split paragraphs, we believe the assumed improvement in legibility is worth the wasted space. It also gives pages a "lighter" look.
Automating the layout process
The layout of a newspaper is designed to attract readership and to optimize the newspaper's effectiveness in presenting information. Rules and conventions have evolved over the years and almost all newspapers share well-established layout principles. The large screen allows for the use of newspaper layout techniques on a computer display.
Newspaper layout was one of the first newspaper processes automated with the help of computers. The application is obvious and the market is large. The problem is reasonably constrained; the program is given a set of news articles and advertisements. Advertisements are placed according to one set of rules, while news articles are placed in the remaining space (the "newshole") according to another set of rules. Since The Electronic Broadsheet does not contain advertisements, we do not discuss them. Without the ads the problem is surprisingly similar to the computer game Tetris; blocks are to be placed to minimize open holes.
The Electronic Broadsheet tries to use elements from traditional page layout, but the process of laying out the pages is very different. Paper-based newspapers are issued in discrete editions, while the electronic page continuously adds new articles. Accordingly, old or less important articles have to be removed and this complicates the shape of the newshole.
The use of color
One property of paper, and especially newspaper pulp, is that it yellows when exposed to light. This characteristic feature is used to indicate to the reader how long an article has been on display; a map of the pages represent each article with an icon faded relative to its age. The map also conveys the given importance of each article. Jacobson and Bender [Jacobson, Bender 90] report that when contrast of value is low between foreground and background colors (dyads), energy is controlled by hue alignment. Complementary dyads were found useful as highlights. Monochrome dyads did not get the user's attention. On the map, important articles are given a border using a hue complementary to the icon color in order to draw the user's attention. Less important articles are given a border using the same hue as the icon color. For a more thorough discussion on the map and the use of colors see [Lie 91].
The rendered pages
Having composed the pages, the Electronic Broadsheet is now ready to offer the reader pages that hopefully are visually and intellectually fulfilling. The reader is offered a range of sections, i.e. pages devoted to a particular subject, in addition to the front page. As the reader navigates through the different pages, the computer will try to track how much time is spent on each article. This information is passed over to the article selection module, which dynamically updates each reader's selection criteria. It is beyond the scope of this paper to describe the news selection mechanisms, see [Bender et al. 91] [Orwant 91] for a more thorough discussion.
Are metaphors scalable?
Running 1k software on 2k displays is suboptimal. This section provides some guidelines for making applications that take advantage of the large medium.
One of the stated goals of the X window system is to avoid making user interface policy decisions. Instead, X provides a rich set of mechanisms for implementing a variety of user interfaces-in our case the newspaper meta phor. Also, using X made a variety of software available for immediate use- editors, terminal emulators, window managers and games to name some of the categories. However, most of these programs has been written for and on 1k screens.
One example is programs that put a window in the center of the screen on purpose to catch the attention of the user. They will probably succeed on a 1k display, but the same is not true for a larger screen. The window is more like ly to be missed due to its relatively smaller size. Instead of using absolute locations for new windows, they should be positioned relative to related windows.
Some applications use sound to signal errors or special events. The infamous "bell" sound (ASCII character 7) may have worked well when termi nals were physical devices and not a window among several other. The non- spatial bell sound is hard to trace, and the more applications, the harder it gets for the user.
Applications should always be aware of the environment they are a part of. They should know the size and type of screen they are being displayed on, and thy should expect a fair competition for the available resources. Grabbing the entire screen is antisocial behavior and a programs that fail to use a reasonable font size is less than optimal.
The user's attention is also a scarce resource that applications should com pete for. Programs should never insist or expect getting the full attention of the user, but should indicate, perhaps through visual cues in the peripheral vision, that new data is available or input is requested. Correspondingly, programs should never make assumptions about what the user is looking at. In the context of newspapers, the word "tabloid" carries two meanings: condensed size and a journalistic style. The close link between the two indicate that metaphors don't scale easily. The same is true for computer screens. E.g., the desktop metaphor does not scale easily onto a large screen: finding the cursor proves difficult and dragging icons across the screen suddenly requires planning ahead. Making the cursor larger and faster are inadequate solution. While it is terrific to have an Emacs buffer which can display 200 lines of text, it is annoying to have to hunt for the cursor.
Conclusion
At 2000 lines of resolution, computer displays start to compete with news print in size and legibility. Since the newspaper has had centuries to develop effective mechanisms for navigating large databases displayed on dense, high resolution displays, it is not surprising that many techniques, such as "front page" and "sections" are effective when directly translated into elec tronic form. We can further enhance the utility of the display by adding dy namic use of color contrast, motion and sound. On the other hand, a window system like X11 needs serious rethinking when scaled up by a factor of 4. The typical window managers neither take advantage of the additional resource, e.g. avoiding overlapping windows when possible, nor do they have any framework for the consistent application of color, screen position or motion as means of helping the user navigate.
Augmented by dynamic screen updates, the Electronic Broadsheet takes on paper-based news distribution; it handles all aspects of screen-based news presentation from low-level typesetting to multi-page layout and user interactivity.
The ultimate evaluation of a newspaper is done by the readers. For the Electronic Broadsheet, the number of potential subscribers is strictly limited by the number of 2k monitors in the world-a number that is still very low.
Acknowledgment
This work was supported in part by IBM.