DSV CMC home page

USER ADJUSTABILITY -
More flexible computer conferencing systems?

by Tarja Lintunen
Department of Computer- and Systems Sciences

Stockholm University and KTH
Electrum 230, 164 40 Kista
March 1995

See also: Lotus Notes - An Enabler for the Giants to Learn to Dance?

1. Introduction

Asyncronous computer conferencing systems are used to accomplish specific goals: to facilitate and ease communication and co-operation among people time- and place- independently. (källa 1) Communication can be regarded as a process which is included in a constantly changing cultural, social and cognitive context. Communication or co-operation is usually on one-to-one, one-to-many or many-to-many bases. Group communication activities (computer conferences) could be defined as a set of members (a group of individuals) and a set of interlinked items (objects) containing information that is exchanged by the members. Activities are usually organised along categories (topics), that provide a common object with a common purpose to all members who participate.

The environment(s), the user(s), the task(s) and the asyncronous computer conferencing system(s) are important factors while regarding the fit of a particular system into specific settings. The properties of the environments and the tasks are beyond the scope of this paper because they are context specific. The perception of the technical properties of a asyncronous conferencing system, such as adjustability, can not be isolated from the social context of use. The properties of the users, regarding to some general human information processing aspects, and the properties of asyncronous computer conferencing systems, as far as general adjustability goes, are the focus of this paper.

Some factors that are important for achieving the acceptance of the users of traditional (= text-based) asyncronous computer conferencing system are proposed by Oliver Vallée. The following summary is based on his work (källa 2). A system must communicate the group communication concept with clarity. It must allow many different user approaches, from an occasional browser to a structured learner. Standard user frendliness concepts are very important. On-line dictionaries, proper set of functionality (a set of tools for a mix of media's), adequate metaphors and helps are examples of other important parameters.

Very interesting, although not yet fully understood, issues are raised by the emerging global, multicultural and multimodal conferencing systems that support the seemless communication. Since the main focus of this paper is on the asyncronous computer conferencing systems these more complex issues should be investigated during the future studies.

Since the computer mediated information has to be understood by the users before it can be elaborated, it is valuable to understand how the human information processing functions. In the chapter 2 some basic aspects of human information processing is described. The next chapter deals with some general user friendliness aspects and the forth chapter deals with adjustability in regard of asyncronous computer conferencing systems. In the fifth chapter the adjustability is related to one specific computer conferencing system (Lotus Notes). Since the future development of the in-house asyncronous computer conferencing system is one of the implementation areas of this study, the next chapter deals with the adjustability of that system and the last chapter elaborates some possible future directions.

2 Some human information processing aspects

Cognitive and perceptual psychologists (amongst others Anderson) have studied how human beings observe and understand their environment, how they communicate and how they use their cognitive skills. At least two different cognitive levels can be distinguished: the high cognitive level and the low cognitive level. High level meaning conscious and analytical processing, which has a low capacity and handles one thing at a time. Low level meaning processing many parallel activities and being near to sensor-motor activities. This level has a high capacity and processing is almost automatic.

The high level is used for reading and understanding numeric and semantic information. On the lower level the pattern recognition, for example in terms of shape, colour and frequency in time, happens. To perform daily work people are simultaneously using both cognitive levels. People have a tendency to organise the work environment so that they can concentrate on the "core" of their tasks. This core requires (mostly) the use of the high level cognitive activities and the supportive activities for performing the tasks can be automated. Handling any computerised conferencing system requires high level cognitive activities and interferes thus with the core activity i.e. creating the content of the information to be communicated.

Distinction between primary and secondary information is another important factor while humans process information. The primary information is read to be understood and interpreted and requires thus high cognitive level skills. The appearance of the information (formed by colour, pattern and spatiality, just to mention some aspects) could be regarded as secondary information and is perceived on a low cognitive level.

Third important aspect of human information processing concerns the memory, which can be divided into short term and long term memory. While working with and through asyncronous conferencing systems, the characteristics of the short term memory are important. Only a few (5+/-2) information elements can be stored simultaneously, decay time is short and short term memory is easily disturbed by other cognitive activities. The conferencing systems should therefore be designed so that the load on the short term memory is minimised and leaves the maximum capacity to the core work task.

It is also important that the user has all information needed simultaneously in order to perform a task effectively. Having to switch between different screens and to integrate different information leads to an extensive cognitive load. An uniform design, where the elements build up a pattern, is essential in order to create a common reference among users (colleagues).

Humans think in non-linear chunks. People try to associate these chunks with each other and build a network of concepts. Human cognition is thus essentially organised as a semantic network in which concepts are linked together by associations. Human associations are primarily evoked by sensory-motor stimuli. This means that the more sensory dimensions are involved the more associations will be evoked. The most important human sensory modalities are vision, hearing and muscular sensation. Another medium for evoking human associations is language. This implies that human-computer interfaces should propose a multimedia integration, containing at least a graphical display, an audio output and a device for direct muscular manipulation, thus making it possible to evoke associative meanings in a maximum of different modalities.

Writing models involve following three phases: exploring, organising and encoding. Exploring means that the writer retrieves potential context from long-term memory or external sources, considers potential relationships, groups ideas and constructs structures creating a network of ideas. Organising means putting ideas in order leading to a structure of related concepts. Encoding means translating this structure into a sequence of words, sentences, paragraphs etc. The structure of the encoded text is linear. Reading employs these phases in reverse order, which means that the linear sequence of words is transferred into a structure that is integrated into a network in long-term memory. Reading models point out that understanding of information takes place at four level: lexical, syntactic, semantic and pragmatic. These levels interact continuously and they can not be truly separated.

3 Some general user frendliness aspects

Many sets of usability heuristics have been published (amongst others Nielsen). Following summary of a set of nine heuristics is based on his work. It has to be pointed out that there is some degree of individual subjectivity in this kind of heuristics and it is important not to rely blindly on them.

The first factor is visibility of system status. This means that users should be kept informed of what is going on. This could for example be done by providing timely and accurate feedback.

The second factor is the match between system and real world. This means that the system (at least on the human-computer interface level) should apply user's language containing familiar terms and metaphors from the real world. User's and system's conceptual models ought to be near each other.

The third factor is user control and freedom. This means that possibilities to undo and redo should be supported, user actions should be reversible, interaction should be modeless and so on.

The fourth factor is consistency and standards. This means that same things should be expressed and look the same way, the system should conform to platform interface conventions and so on.

The fifth factor is error prevention. This means that the system should prevent errors from occurring in the first place.

The sixth factor is recognition rather than recall. This means that users should be able to see-and-point instead of remember-and-type. The repertoire of available actions should be salient. The lists of choices should be provided and so on.

The seventh factor is flexibility and efficiency of use. This means that accelerators should be provided. Users should be able to tailor their interface in order to speed up dialogues and so on.

The eighth factor is aesthetic and minimalist design.

The last factor is helping users recognise, diagnose and recover from errors.

The highlighted words are Jacob Nielsen's "subjective attempt to abstract the main usability thrust of each factor". (Nielsen)

The above mention general rules are applicable in any software and more so while we are focusing on asyncronous computer conferencing systems, since these systems are used by people without extensive training in computer sciences. Computer conferencing systems should not only be studied as such, while they are tools used by individuals and groups of individuals within more or less complex settings.

4 Adjustability

To adjust is defined as " to modify to meet new conditions" (källa 10 Websters New Dictionary of Synonyms). Adjustability means thus that something can be modified to meet new conditions. Tailor ability and customisation are two other terms used to describe this ability.

4.1 Traditional adjustability approach

The basic philosophy according to this approach is that the asyncronous computer conferencing system should attempt to fulfil a (pre)specified set of user requirements. This approach supports mainly the standard operating procedures for which the system was/is designed. So called "parameter-driven" adjustability have been one way of adding flexibility to this kind of systems.

Commercially successful computer software (word processing packages, spreadsheet packages etc.) has usually features that provide extensive capabilities for adjusting the software according to new requirements e.g. by allowing users to modify the interface according to her/his needs or by building new data structures. Many of the capabilities for adjusting are though purely cosmetic e.g. they allow users to modify the visual properties of the interface according to their own preferences. Those systems that allow structural adjustability could be classified as open systems. Open is here used in the sense that these systems impose few restrictions on the type of the task to be performed. They just provide tools that users may or may not use.

Adjustability means often extensive predefined functionality. There exists a perception that the more functions are provided the better. More functions means more flexibility and more complexity in the system. Different kind of users (for example first-time, casual or expert users to categorise roughly) have different kind of needs for adjustability and these needs probably change while users move from one level of expertise to another. Providing extensive functionality is not enough, users must understand what the functions do and how to use them.

A study done by Walter and O'Neil (källa 11) shows that flexibility (usage of abbreviations, default values, common synonyms etc.) improves an expert users performance, but hinders a novice. On the other hand Shneiderman (källa 12) argues that even expert users are penalised by poor design. Goodwin (källa 13) finds an interaction between functionality and usability in a comparison of two message-handling systems. On system was functionally rich but difficult to use. The other system was functionally limited but easier to use. He points out that the richer but harder system provided less effective functionality. Good interface design can though enhance overall functionality by making the functions accessible to users and facilitating effective use of functional capabilities.

Since the far ends of the continuum from programmer provided adjustability (too expensive and time-consuming for enterprises and organisations of today) to entirely end-user tailored adjustability are though technically possible but not likely to be successful in most of the environments where traditionally designed asyncronous computer conferencing systems are used, this study focuses on the middle or on the right side of the continuum as shown in the Fig. 1.

Programmer provided       End-user adjustability    End-user tailored         
adjustability (C, C++)    predefined basic          adjustability             
                          primitives (templates,    (interface/rules/object   
                          tools etc.)               editors, high-level       
                                                    programming)              

Fig. 1 The Continuum for user adjustability

So called tool kits and UIMS (user interface management systems) have been used as tools for end-user tailored systems. Even the mechanisms that are used to support end-user adjustability acquire a considerable amount of skills. Users could be categorised very roughly into three different groups according to their skills and interests to adjust conferencing systems they use: 1) those who want to get the work done and do not have interest in the system per se, 2) those who like exploring the system, but may not fully understand it and 3) those who understand the system thoroughly (usually with formal training or experience in computing). Different adjusting mechanisms (a collection of possible options such as different tools, building blocks, parameters etc. and design information) should be easily accessible and adjusting per se could be a community effort thus encouraging members of user community to help each other by sharing insights and expertise.

4.2 Expressive systems approach

Expressive systems (källa xx) approach is based on different philosophy. These systems are designed to support both standard operating procedures and exceptions from standard operating procedures or rather there are not any predefined standard operating procedures. These systems are composed according to every user's individual needs and changes of the systems are made according to changes that occur in user's real world needs. These changes might include support for exploring new communication possibilities. Changes should be able to be implemented in the time frame appropriate to user's needs and by the users themselves with occasional aid from IT-experts when needed.

These systems could be based on distributed client/server architecture, use the concept of object-orientation and might borrow techniques from the world of artificial intelligence. As new, more user-friendly techniques emerge, they will substitute today's technical solutions.

The three different levels that could provide adjustability are user interface (how the data is gathered and presented), functionality (how the data is processed) and datastuctures (how the data is stored). According to expressive systems approach users should be able to choose the suitable interface from a set of possibilities reaching from a simple text-editor to sophisticated graphical representations. Datastructure should be very flexible with atomic level components e.g. nodes - either atomic or composite - and links - typed, bearing semantic meanings and directional for ease of forward and backward navigation. These systems do not present to a user a series of predefined options in attempt to cover existing and future user requirements. Functionality is created by the user from the atomic level components and rules along with the capability to manipulate and extend them in new ways. All basic constructs should be represented graphically and there might be built-in constraints in specific settings. These basic constructs vary according to the actual domain. These systems could be regarded more like tool kits which users explore and implement according to their specific needs.

One of the benefits of the expressive systems approach is that it facilitates the handling of unexpected changes and events in the future. New kinds of communications, such as composing multimedia (olfactory or tactile) messages, could easily be added when needed into asyncronous computer conferencing system designed according to this approach.

The basic components and capabilities of a conferencing systems, built according to accepted or future expressive systems standards, could provide a starting point. One suggestion for these basic components of an asyncronous computer conferencing system is described in ISO/IEC JTC 1/SC 18/WG 1 N 855: Preliminary User Requirements for Group Communication (källa x). The users would design their actual conferencing application according to their specific needs either from these basic elements or from more powerful high-level components, such as templates or like, while they are implementing their own application of an asyncronous computer conferencing system providing that the actual system has been designed and constructed according to expressive systems approach.

4.2.1 Work done with the notation of expressive systems approach

Thomas W. Malone et al. (källa 16) have defined four kinds of building blocks: objects, views, agents and links, that give users the flexibility to tailor the basic functionality found in some groupware systems (more specifically gIBIS, Co-ordinator, Lotus Notes and Object Lens). Their study proposes that these four primitives would be enough to provide a "tailoring language" for end-users.

According to Malone et al. objects represent things from the real world, such as people, tasks, conferences, messages, forms etc. These objects include attributes (collection of fields and field contents, buttons etc.) and set of actions that can be performed on objects. Views are customised collections of objects and/or content of objects. Users should have the possibility to modify these views. Agents perform tasks for users without their direct attention and links represent relationships between objects or fields within an object.

Malone et al. identified six primary user level types of modifications:

1) defining new object types (new content)

2) adding fields to existing object types (modifying content)

3) selecting views for objects and sets of objects (display)

4) specifying parameters for views (how to create displays)

5) creating new agents and rules (how to process)

6) inserting new link types (how to connect).

4.2.2 Using hypermedia paradigm according to expressive systems approach.

Hypermedia systems are used to create, annotate, link together and share information from a variety of media. Flexible access to information is provided by navigation, by annotations and by tailoring the presentation. Thus hypermedia systems try to exploit the networked nature of human cognition.

One way of adding adjustability to asyncronous conferencing systems designed according to a hypermedia paradigm is the usage of hypermedia templates. Hypermedia templates could be defined as sets of pre-linked documents that can be duplicated (källa 19) or as "a partially-created, properly formatted collection of document skeletons that can be filled in by the user" (källa 20). A template could be considered as a composite object, which is comprised of other objects, such as nodes and links. Many applications might have some common basis that can be transformed into hypermedia templates or hypertext templates when it suffice with text-based content.

Some requirements for a hypermedia system to provide templates are:

Generic operations to create, duplicate, edit or delete a template. Duplications should create empty documents with nodes and links. Facilities to add contents to empty documents, to list templates (inclusive documents and links), to display overviews of templates, to access a template by different criteria (author, date etc.). Some control operations to overview templates, to zoom into specific parts of web and to look at the content of documents. Choices to find out the master templates (the one that was copied) and facilities to edit master template. Changes to master templates should propagate to all templates created from it. Facilities to specify formats and screen layouts for a template and to add help should exist. Direct manipulation of the content of the documents should be possible.(Källa 21)

Hypermedia document should consist of three components: the content part, the organisational part and the presentation part. The content part contains content nodes (atomic or composite), that carry information and content links that connect content nodes. The organisational part structures the network under a reader-oriented perspective, providing thus a possibility to adjust to different reader needs. The presentation part handles the actual display of structure and content and provide the means of navigation. Textual, graphical or combined styles could be adapted as needed. Hypermedia systems that incorporate virtual structures, computation and filters would be more dynamic.

Information retrieval through queries are important for large hypermedia systems. Conventional mechanisms; keyword based, weighting of words, ranking of documents etc. have been used. Queries by navigation, structural queries, inference networks, cluster hierarchies, aggregates and AI techniques are some other successions to facilitate browsing and searching. (källa 23)

Rao & Turoff (källa 24) propose a general framework for hypertext functionality. They classify nodes into six different semantic types: detail, collection, proposition, summary, issue and observation. Links can be categorised into two major types: convergent and divergent links. Convergent links can be classified into specification, membership, association, path, alternative and interference links. These links help narrowing the pattern of relationships. Divergent links are classified into elaboration, opposition, tentative, branch, lateral and extrapolation links. These links expand or broaden relationships. This approach allows the users to freely assign their own semantics to objects (nodes) and relationships (links) and provides also information about the composition or evolution process, the way network is dynamically changing.

To create a good user interface for hypermedia based asyncronous computer conferencing system requires an understanding of the organisational setting, the targeted task domain, the user population and the desired outcome of hypermedia navigation.

5 Basic properties for a asyncronous conferencing system

An asyncronous conferencing system should have following basic properties. There should be separate activities (conferences) for broad subject areas. This would provide the basic level of organisation. Threaded discussions should be allowed within activities. Different threading-structures (tree, star, web etc.) might be needed. Informative topic lists or webs or other suitable structures to provide overviews are needed. The criteria, which these overviews are based on, should be user choosable. For example subjects, dates, authors, keywords and like could be used. The integrity of the topic i.e. the items from the beginning to the most recent one, should be stored. Personalised search and filtering tools ought to be provided. Appropriate access control mechanisms and support for different usage types could also be needed. (källa 14)

These properties are based on the experience of mainly text-based asyncronous computer conferencing systems. Users have different needs for communication: text based, animations, audio, graphics and different kind of combinations just to mention some of the existing needs. Thus multimedia messaging is needed since the human communication is usually multimodal. Increasing international communication requires multilingual and multicultural conferencing systems. Needs for integrating non-traditional input devises (such as olfactory devices) and output-devices (such as from text to synthetic voice etc.) are also emerging.

There is so far not enough knowledge (at least to my knowing) of the basic properties of the multimedial, multilingual or multicultural computer conferencing systems. These new conferencing systems will be more complex and they will be used by people with more diverse backgrounds, knowledge levels, tasks and needs.

5.1 An example of adjustability in existing asyncronous systems used for computer conferencing.

5.1.1 Lotus Notes

The information in Notes databases is organised and maintain using four basic building blocks: forms, fields, views and documents. These can be regarded as different levels of the basic components according to the expressive systems approach. Lotus Notes provides predefine sets of these basic building blocks. In addition there are formulas that define the content of documents and determine how data is displayed. Lotus Notes provides a broad range of possibilities for user adjustability. These possibilities require although some amount of computer skills. For example templates are empty skeletons of databases that provide a basic structure for new users. This is an example of the high-level components used according to expressive systems approach. These extensive sets of basic building blocks and the possibilities to combine these building blocks according to individual evolving user needs provide extensive adjustability possibilities. In addition there is the possibility of the usage of external tools to create the non-supported file types and allowing these files to be stored and transported as attachments to Notes messages.

The mechanisms used to provide adjustability are so extensive, that the users of the two first categories i.e. those without interest in the system per se and those who just like to explore the system, might be overwhelmed by all these possibilities to adjust. Providing IT-expert help while designing and creating the needed conferences would probably aid these users to achieve effective conferencing applications. For the third group of users, those with experience or formal education in computer sciences, these basic building blocs provide broad possibilities to adjust the conferences as they need.

It seems that while designing and creating Lotus Notes the expressive systems approach to adjustability has been used at least to some degree. The visual presentation and the functionality to some extent are composable from predefined building blocks.

6 Some possibilities to add adjustability into the in-house computer conferencing system.
6.1 Adjustability according to the traditional approach
6.1.1 Defining new object types

There could be a basic object type, for example called "message", in the system. This type would for example have following structured fields: author, receiver, subject, date, keyword(s). The content of the message could be unstructured. This basic object could be further defined, for example as a form. This form-object would have some additional structured fields, such as form type etc., defined. The content of the form could be structured or unstructured. The next level could for example be a form for " received phone call". Even the content of this object would be structured. These more or less specified forms could be composed of the predefined sets of fields, texts, buttons, icons, item lists with rules etc. that would be provided by the system. The users could choose the predefined possibilities from item lists, which they receive through dialogue boxes while defining for example that they want to create a new object type. The inheritance hierarchy for objects could be used to simplify the creation and modification of type definitions.

Basic set of messaging objects and fields should be provided by the system as a standard for conferencing application. Several messaging types and their reply types could be predefined. The new requirements should be built on the top of these basic types according to the individual needs.

6.1.2 Adding fields to object types

Adding fields to existing object types could be done by choosing "add field" action. This action could for example open a dialogue box which guides the user to choose of a list of predefined field types and the rules which are appropriate for the chosen field type. The content of the fields should naturally be added by the users. The content could be typed, copied or if it is a link "add link" action should be performed.

6.1.3 Views

Users should be able to view the content of the objects (regardless of type ) and the relationships (links) between the objects or fields within objects according to their own needs. The system should provide users with a possibility to define e.g.. in which order (s)he wants to see the messages (the oldest first, the newest first, sorted by the author, sorted according to keywords or subject etc.). This definition should be done by providing the user all the alternatives and letting her / him to choose (dialogue boxes, item lists, pop-up menus etc.). The users should also be able to define which fields they want to see from objects, users should be able to rename selected fields if desired. Various kinds of summary displays should be predefined so that the users could choose the view most suitable for their needs. An example of these predefine views could be "all instances of an object type" e.g. all "phone calls" for a specific user. These summary displays should be more permanent and could be regarded as folders or stores.

6.1.4 Specifying parameters for views

Users should be able to select the parameters (fields, links etc.) from a predefine item list that defines the actual object type, from which the user would like to define a new view. A dialogue box should guide the user to choose from item lists. For example if the relationships between the objects are desired then the user have to specify which relationships are to be shown e.g. links from which fields or from which object types. Users should also specify what are to be shown in every node: subject, author or some other field.

6.1.5 Agents and rules (creating new rules)

Users should be able to compose their own settings of commands (either as menus or as toolbars) of the total amount of existing commands. User commands could also consist of macros / scripts to simplify a stream of commands. The total amount of commands should not be too extensive, since there is a danger of less effective functionality then. (compare Goodwin).

6.1.6 Creating new link types

Users should be able to create new link types through defining the relationship between source and target objects (or fields). This could be done for example through "new link" function. The dialogue box would guide the user to choose the source and target objects (fields) and to name the link between objects (fields) and define the processing regarding the new link type if applicable. Default value for source object (field) could be the object or the field where the cursor is positioned.

Another way of adding user adjustability regarding links would be a set of predefined link types that users could choose from while connecting objects (fields).

6.1.7 Search possibilities

Users should be able to search the contents of objects according to own needs. Subject, file name, conference name, author, receiver, different fields or content of the message etc. are examples of some search factors. The software should assist the user by providing an extensive list of search factors while (s)he is searching. The results should be shown according to users needs either according to standard views or individually defined views.

6.2 Adjustability according to expressive systems approach
6.2.1 External tools

An alternative way of providing user adjustability would be using the attachments-mechanism to transport files from the creator to a reader. Using this adjustability approach means that the end-users would have basic set of conferencing functionality to provide a shell for attachments. The users would be able to use their favourite tools for creating the content of the attachments (for example word-processor, spreadsheet product, database product, presentation product and so forth). Both the creator and the reader(s) should use same or at least compatible tools to process the attachments.

According to this approach for example forms would be created outside the conference system (in word-processing- or drawing packages) and saved as attachments which users could copy and refill and provide as new items or new attachments.

6.2.2 Message based adjustability

The MIME (Multipurpose Internet Mail Extension) messaging provides adjustability on the very basic level of the message, while a MIME-message itself can be an atomic message or a composite message (with several body parts). The initial content-types are text, multipart, message, image, audio, video and application. These content types can have several subtypes. In addition both internal (inclusion) and external (attachments) file-handling is allowed. Since the attachments can also be pointers to local or remote files the allowed transfer mechanism has to be specified. For additional information of MIME see Borenstein & Freed.

To be able to handle the non-textual body parts or files external agents could be launched. These agents should be according to user preference and a should be changeable by the user. The visualisation of these non-textual parts or files could be done by icons, preferable by user changeable icons.

6.2.3 Designing user-interface, functionality and data structures according to the "just-in-time" needs.

Users should be able to choose their favourite tools to create/view the content body parts(s) of the messages. Users should be able to compose the appropriate functionality, usually the appropriate semantic naming according to the groups conventions and the order in which the parts are to be shown. If it is a question of a composite structure users either choose from pre-existing structures (templates) or create graphically appropriate structures according to the actual needs. The basic data structures - objects (atomic and composite) - and directional, typed links provide flexibility to handle evolving functionality and presentation needs.

7 Discussion

Solutions based on predefine standard operating processes are usually not adjustable enough to changing requirements.

A) From centralised to decentralised

B) Freedom of "cosmetic design" (user - interface level => presentation)

Regardless of the technology on which the asyncronous computer conferencing system is implemented some general adjustability features should be included.

The so called cosmetic changes i.e. changing colours, changing terminology or symbols according to the actual user-groups, changing a language, changing sound alerts and so on should be able to be done simply by any user. The guidance could be provided by dialogue-boxes with pre-filled lists of the alternatives and possibilities to create "own" alternatives and buttons to start the processes which are available in the specific settings.

C) Freedom of tools (evolving selection. according to evolving expertise)

The functional changes that should be able to be done by any user are task dependent, but some general properties that should be adjustable are among others following.

The modifications should be done interactively. The choices of favourite editors, external agents, default printers, optional access mechanisms when only pointers are provided, defining own shortcut buttons, creating aliases, defining own sort and search parameters, choosing and changing fonts and font sizes, customising windows, defining different printing-options (headers, all body parts or whatever is actual), handling of containers (body parts that contain body parts) are all example of parameters that should be user adjustable.

D) Freedom of media, multimodal according to "natural" human communication

Flexible basic building blocs (nodes and links ) combined and composed into multiple composite levels presented as multilevel (predefined?) structures and allowing the users to rebuilt and rename these structures according to their own needs and own terminology seems to provide adjustability that could cover the needs of many users. Basic building blocks and multilevel structures combined with the usage of environment - and task-related metaphor(s) and terminology could shortened up the initial learning period and provide examples of how to use the features of adjustable asyncronous computer conferencing systems.

E) Freedom of functionality (created from atomic functions)

7.1 Epilogue

As important as the tools (conferencing systems) are for mediating the communication, the main elements are those who/which communicate and the context of the communication. These systems as such substitute paper, telephone, fax or whatever media has previously been used and provide some additional features (e.g. rapidness, possibilities for non-linear reading)