The Power of Nine
Overview of Information Navigation Options
2. Principles of Organisation
2.1 The traditional modes for the organisation of information have been:
- Hierarchies
- Matrices
- Alphabetical or Numerical sequences
- Degrees of Magnitude
- Degrees of Centrality or "Relevance"
All of these systems are entirely arbitrary or have arbitrary elements.
2.2 Hierarchies
2.2.1 The most traditional method of arranging information in Western culture is the hierarchy, still seen widely in the use of menu systems. This depends upon the author (originally the Greek philosopher, Aristotle) arranging information according to taxonomical postulates. The hierarchical system usually involves all information being subdivided as the hierarchy descends. Such systems are rigid and do not allow for structural cross referencing though users obviously retain information at nodes in different parts of the system and make mental connections. Information on different branches or 'twigs' of the bifurcating hierarchy can only been reached through ascent to the lowest common junction.
2.2.2 The taxonomy of hierarchies is postulate dependent; a Christian and an agnostic, for example, would derive very different taxonomical hierarchies.
2.2.3 In addition to hierarchies of information arranged according to substance or content, there are separate kinds of information hierarchies dependent upon source, publisher and structure; these are dealt with in Section 3 below.
2.3 Matrices
2.3.1 Matrices are geometrically formalised sets of cross references whereby any piece of information receives two or more taxonomical classifications. A hotel, for example, might be classified on one axis of a matrix according to its geographical location and on another according to its price. It might be classified on a third according to some measure of "value for money". Matrices depend on taxonomy but it is less hierarchical than that in a bifurcating hierarchy because any 'cell' in a two-dimensional rectangular matrix, with the exception of top left, bottom right and, where it exists, centre, may be accorded different degrees of importance on its two axes.
2.3.2 The more axes in a matrix the less hierarchical it is.
2.4 Alphabetical or Numerical Sequences
2.4.1 Alphabetical sequences, other than the editor's assignment through inclusion and exclusion which is often unstated, do not assign hierarchical value to information; the linguistic designation of a term dictates its place in an order of delivery. Thus, an English and a French musical dictionary will place "Bach" above "Couperin" and "Couperin" above "Mahler". When translatable terms are introduced the order in each dictionary alters, thus "key" in English will be inserted between "Couperin" and "Mahler" but "clef" will be inserted in the French dictionary between "Bach" and "Couperin". It should be noted that some proper names, such as those for countries and cities, can be translatable, e.g. "Germany" and "Allemagne", "London" and "Londres".
2.4.2 Numerical classification follows a variety of principles. Thus, elements might be listed alphabetically but are usually listed according to their molecular structure as expressed in the Periodic Table. The most familiar form of numerical expression is chronology; here the order of a set of events will not vary between calendars but the numeric allocation may be different, e.g. a series of events will follow the same order in an Islamic and Christian account of the Crusades but each event will have a different year according to the tradition. Unlike alphabetical listings which depend to a large extent on authorial selection, unless a yardstick is stated and adhered to - e.g. all towns with more than 100,000 inhabitants - numerical listings may be arbitrary or rigid, e.g. the Periodic Table is rigid but a chronology of the Crusades is arbitrary. Some information, such as the listing of towns over 100,00 inhabitants, may be listed alphabetically or in ascending or descending numeric order of size.
2.5 Degrees of Magnitude
2.5.1 The most intuitive and familiar form of measurement of magnitude is the numeric quantifier. Thus, from a previous example, towns may be ranked according to the number of inhabitants but many scales of magnitude are less clearly defined and the same set of factors may receive different rankings according to the time or the culture, thus the centrality in historical studies of historiography. Through time a series of events may lose value in their relative magnitude to more modern events but their relative magnitude to each other can alter.
2.6 Degrees of Centrality or "Relevance"
2.6.1 The notions of centrality and relevance in ordering information are familiar through the presentation of news in print and in the electronic media. Unlike the construction of a classical essay which begins with an "introduction" and ends with a "conclusion", much current affairs information is presented in the author's "descending order of importance" so that it can be "cut from the bottom" or abandoned at will by the recipient. This is not to be confused with "magnitude" ordering. In current affairs a trivial additional to a long running story will appear at the top of the order and the background, including a summary of the events of most magnitude, will come later.
3. Value Assignments
3.1 There are three major characteristics of value assignment in information:
- Source
- Publisher
- Structure
3.2 Source
3.2.1 Until the wide publication of anonymous or unsigned information, source was a primary indicator of the value of information. Now that much information is assembled collectively and presented anonymously the designation of source as an indicator of value is declining.
3.2.2 The explosion of sources and the current postmodernist theory of deconstruction have both reduced the importance of source as an indicator of value.
3.3 Publisher
3.3.1 As with source, the designation of publisher as an indicator of value is declining but it is still the case that different values will be allocated to identical information from an identical source with a different publisher. Thus, economic statistics will be allocated decreasing value when published respectively in: a peer reviewed journal; a book; a broadsheet newspaper; a tabloid newspaper. The modern equivalent of the notion of "publisher" is "brand".
3.3.2 The loss of the traditional information value markers - periodical, book, newspaper - imposes a difficult task upon users who wish to assess the value of information before they spend time absorbing it.
3.4 Citation of source and publisher is as important in information navigation as its physical design.
4. Structure
4.1 Structure transforms data into information. We have already discussed structural elements in Section 2 but the way in which information is presented is a structural element which assists the recipient to assign value to information. the four major aspects of information structure are:
- Sequence
- Size/intensity
- Layout
- Colour
4.2 Sequence
4.2.1 We intuitively understand some major features of information sequencing such as: chronology in electronic media presentations and live drama (the author arranges information in a linear order for our reception); from top to bottom and left to right; from point to point of identical characteristics (chapter heading to chapter heading).
4.2.2 Increasingly, information navigation is being made more difficult either through the deliberate abandoning of intuitive sequence (chronologies presented out of order in novels and films) or because the visual pleasantness of a design makes the sequence ambiguous (in an array of five vertical panels is the left or the centre to be examined first?).
4.3 Size/Intensity
4.3.1 In the presentation of information, particularly text and static graphics, object size and intensity play a central role in allocating value; we tend to pay more attention to larger type or more densely inked type; we are attracted by large objects and vivid colour.
4.3.2 In the presentation of information the distinction between foreground and background is an important value indicator.
4.4 Layout uses spatial relationships to reinforce sequence and size/intensity.
4.5 Colour can be used both to establish taxonomy and allocate value to information.
4.6 Problems are created for recipients when rules of significance are not established in design or when design 'conventions' well understood by the recipient contradict, e.g. for the purpose of symmetry a top left and bottom right element are given equal weight; foreground is indistinguishable from background; collage makes sequencing unclear.
4.7 Whereas aesthetics should assist in the establishment of the significance of content, an over emphasis on aesthetics, particularly in visual design, may retard the assignment of intended value to content.
5. 3-Dimensional Navigation on a 2-Dimensional Screen
5.1 Interpreting 3-dimensional constructs in a 2-dimensional environment depends upon the establishment of conventions, e.g. the convergence in the far distance of apparently parallel lines, the use of the depiction of light to establish shadow, the use of 'depth of field' to differentiate foreground from background. There are also conventions for 3-dimensional data arrays such as notional cube matrices. Without such rules the information is not accurately conveyed.
5.2 Navigating without conventions through 3-dimensional hypertext presents users with severe problems in allocating significance to information and particularly in establishing structural relationships between pieces of information. Thus, the use of 'eye candy' in a presentation militates against a ready grasp of the central message; moving from link to link in a random 3-dimensional fashion dictated by individual authors can easily lead to disorientation where the only strategy is to retrace a route link by link.
5.3 As there is obviously no unity of convention between designers, colour coding for significance may be completely counter-productive; an ecology site might allocate information of prime importance to a green zone with red being purely decorative whereas a health and safety site might consign vital data to red zones, using green purely for decoration.
6. Search Strategies
6.1 Major new strategies for making searching and navigating more effective are detailed in Part 3, "Factors in Deriving an Optimal Information Navigation System".
6.2 Clearly, in a pluralist, unregulated 3-dimensional environment the traditional methods of navigation are no longer satisfactory. The bifurcating hierarchy, though widely used in menu systems, clearly lacks the richness and flexibility to take advantage of hypertext; the first benefit of the concept of hypertext, the cross-reference, is helpful, particularly when it is established through a link but an extended sequence of cross references leads to disorientation; and 3-dimensional rectangular matrices are limited because of the explicit relativities which their common axes dictate. On this last point it is easy to misunderstand the conventions. A matrix which shows geography and price at right angles is not capable of producing a graph to plot the relationship between the two axes unless the geographical axis itself is an ordered sequence, say, towns ranked in order of distance from the capital city. On the other hand, price at right angles to a weighted estimate of services provided will produce a graph of value for money. Thus it is possible to put three axes into a rectangular cube where not all three can give aggregate significance to a cell.
6.3 Any strategy for establishing any kind of relationship must be careful to distinguish between spatial relationships and value relationships; a circle whose circumference contains points symbolising the countries of the European Union gives them all equal spatial relationships but this does not mean that Luxembourg is as big or important as France, the spatial relationship is just an aid to navigation.
6.4 In any information search and navigation system there is an optimal point for the opposite values of simplicity and richness. There is also an optimal point between the length of a sequence and the ability to retain the information if most of the sequence is off screen.
6.5 3-dimensional environments offer immense opportunities to improve data searching and navigation.
6.5.1 In arrays with a large number of nodes - in a matrix, as points where two sides of a polygon touch, as points on the circumference of a circle - the orientation of the information can be transformed from the neutral presentation to one seen from the point of view of a node identified by the user. Any node identified in one plane can be the starting point for a search in a plane at an angle. As a new plane is presented, retracing to previous planes can be executed simply. In a series of highly complex arrays the establishment of a single point where all planes 'touch' can establish an easy route back to a starting point.
6.5.2 'Conventions' can be established which customise presentations for significance; thus, the user might identify the red zone as that which should contain the information to which the author attaches most significance regardless of the author's initial colouration. In this context the author should designate value but leave the user to define representation. To use our example, the use of red and green to define significance and decoration is not much use to somebody who is colour blind.
6.5.3 Date can be reorganised according to user choice at least according to:
- Chronology
- Alphabet listing
- Access by key word
- Presentation according to use
6.5.4 The strategy for defining and narrowing searches is not part of this project but it is important to note the significance of intelligent agents to sort information. Such devices can present information according to use, i.e. in any subject the information most often used is presented first. An alternative use of intelligent agents is to establish a method for defining a search by reference to levels of richness and complexity. This would be a complex measure but would help, for instance, to produce very different results for a 7 year old and a Doctoral student using identical key words.
7. Rules
7.1 The most important aspect of rules is that they should be clear. They do not necessarily have to be simple - there are complex rules for the way pieces may be moved in chess - but they must be readily understood.
7.2 Rules should not be contradictory, e.g. if "top left" means most important and heavy inking signifies greater importance than light inking then "top left" elements should not be inked more lightly than other elements.
7.3 Conventions used to facilitate easier understanding of rules should be subject to user manipulation.
7.4 Because different information designers use different rules their rules should be explicitly stated and where the user moves between regulatory systems - i.e. by migrating from site to site through links - a warning should be issued automatically if the user will be subject to significantly different rules on the immigrant site.
7.5 To keep such complexity to a minimum without damaging author integrity, standard rules for navigation, data searching and the allocation of value should be drawn up from which authors can depart, stating their exceptions.
8. Disability, Searching and Navigation
8.1 In terms of access to information, disability is best considered as a series of functional limitations which require coping strategies rather than as a number of clusters of syndromes described under general headings such as "blindness", "deafness", "physically disabled" or "having learning difficulties".
8.2 In considering the way that people with functional limitations of various sorts handle information it is important not to create an artificial apartheid between fully able and disabled people. Functional limitations, from very mild to extremely severe, can be plotted along a horizontal axis with a general rule of thumb that intensity is in inverse proportion to the number of people affected. Thus, to take a very simple but highly relevant example, people with profound learning difficulties may experience extreme difficulty in navigating a simple menu system but many people who would never be classified as having a learning difficulty may not be able to navigate using a street map in a moving vehicle. On this basis, what is good for people with functional limitations is generally of assistance to the population as a whole; this is what is known as the principle of "design for all".
8.3 Everything that has been said in the first seven Sections of this Deliverable applies to those people who are classified or who classify themselves as "disabled". However, as functional limitations become more severe the need to design specific coping strategies increases.
8.4 The following are some basic principles which should underlie navigation and search systems so that they can be more readily used by people with functional limitations:
- Synchronicity and Integrity
- Degrees of Simplicity and Complexity
- Multiplicity
- Over-ride
8.5 Synchronicity and Integrity
8.5.1 All multimedia content should present its strands synchronously. This may mean, in the case of subtitling, for instance, allowing for a varying degree of density in one strand.
8.5.2 At the same time, each strand in a multimedia presentation should be self-sufficient in its own right, carrying the main burden of the author's intentions. Clearly this is a statement which cannot universally apply - a text commentary on a symphony, no matter how detailed and learned cannot represent the content itself - but in information systems, particularly those concerned with citizenship (including access to content provided through libraries), optimal access should be attempted through considering each strand separately as well as considering them together.
8.6 Degrees of Simplicity and Complexity
8.6.1 The user should have control over the rate at which information is delivered and the level of complexity with which it is delivered. The design of systems should include provision for variations in the rate of delivery, allowing the user to arrest and re-commence information flows.
8.6.2 Users should also be able to define the number of levels of information they require from a basic headline level to the most detailed, "academic" consideration. For this purpose, systems designers need to define the level of complexity of information just as tags are used to define structural hierarchies in HTML. Style sheets could be used to allocate complexity values to information.
8.7 Multiplicity
8.7.1 Even within one cluster of functional limitations one strategy will not meet all needs. In visual impairment, for example, centring information may suit people with narrow fields of vision but it will slow down the delivery rate of screen readers whereas the opposite strategy of setting all information to the left may be good for screen reader users but not helpful to people with narrow visual fields. On the other hand, those with narrow fields may have to balance the need for large amounts of information to be squeezed into a small space against the need to keep the screen filled so that they do not become disoriented in extensive blank areas. This, in turn, may involve 'decorating' the blank areas which might, in turn, obstruct the effective use of screen readers. It might also make it difficult for people with poor visual perception to sort out essential from non-essential information.
8.7.2 In order to preserve diversity of access it is important to establish a system for the customisation of hardware, software and information systems. (See "Enhancing Access to Library Based ICT Services for Visually Impaired People").
8.8 Over-ride
8.8.1 In order to preserve a proper balance between the integrity of the author's intentions and the ability of users to obtain maximum access, it may be necessary to develop facilities for the creation of temporary files of data and metadata where the user can over-ride the intentions of the other on a temporary basis without creating a file which could be mistaken for the original.