Balint, 'COMPUTER-AIDED HUMAN-TO-HUMAN INTERACTION: THE HCHI-APPROACH', Arachnet Electronic Journal on Virtual Culture v2n02 (May 16, 1994) URL = http://hegel.lib.ncsu.edu/stacks/serials/aejvc/aejvc-v2n02-balint-computeraided The Arachnet Electronic Journal on Virtual Culture __________________________________________________________________ ISSN 1068-5723 May 16, 1994 Volume 2 Issue 2 BALINT V2N2 ======================================================================== COMPUTER-AIDED HUMAN-TO-HUMAN INTERACTION: THE HCHI-APPROACH L. Balint Hungarian Academy of Sciences h48bal@ella.hu Abstract General aspects and basic concepts of computer-aided human-to- human interaction are investigated. A new method of using specific computerized tools in the support of human communication is suggested. The key concept in the HCHI (human-computer-human interaction) approach is the introduction of specific and intentional computer-based processing of human messages, in which an intelligent machine aids human-to-human information transfer. The machine support covers not only acquisition, storage, sorting, merging and retrieval of transmitted information, but also translation, formalization, re-recording, analysis and re- synthesizing of human-to-human messages. Moreover, filtering, adjusting, correcting and extraction of selected key facts from messages is possible, as is the constrained transfer of the related mental models via suitable decoding/encoding of the messages. All of these properties result in enhanced precision and increased reliability in human-to-human information transfer -- especially in complex man-machine systems. 1.0 Introduction Theoretical and practical questions of human-computer interaction have been at the focus of research and development devoted to man-machine systems (MMS) for many years (Helander, 1988; Rasmussen, 1991). Related activities have sought to achieve effective and reliable communication between humans and machines. However, it is rarely emphasized that the widely accepted need of investigating methods of human-computer interaction should be complemented by similarly intensive investigations with respect to human-to-human interaction (HHI). This HHI topic encompasses a wide variety of unanswered questions, both in theory and practice (Bayley, 1982; Vickers, 1989). No satisfactory effort has been devoted to increasing the efficiency and reliability of human-to-human interaction in complex man-machine systems. This communication domain is sometimes more ambiguous and more error prone than that of the mere exchanging of information between human and machine (Reilly, 1987). Interaction-like relationships between machines can be handled with relative ease, because machines are normally constructed with attention to prospective communication requirements. However, the similarly straightforward and technically established handling of emerging problems is unlikely when humans are communicating with each other. Human-to-human interaction traditionally exploits standard communication modes which apply well-known verbal, written and gestural ways of transmitting and receiving explicit or implicit information. Here we investigate only those forms which utilize advanced computer technology -- especially the devices and tools of human-computer interaction. It will be shown here that the achievements of computer technology (as well as of human-computer interaction theory and techniques) may greatly increase the efficiency and reliability of human-to-human communication -- especially if precision of transferred information is required. Therefore, - Human-computer interaction (i.e. combined human-to-computer and computer-to-human interaction) is examined by aiming at selecting those features of highest potential benefit for the human-to-human interaction and - Computer-supported means of human-to-human interaction are introduced, with the aim of applying computers as tools for the enhancement of communication between humans, and for achieving increased efficiency and reliability in the exact information exchange among humans in a complex, cooperative man-machine system. Results of these two analyses may be summarized in the following corollaries. - Human-computer interaction tools (human-computer interfaces) may be separated into human-to-computer and computer-to-human tools, each well distinguished by its basic tasks and properties. We shall call them human-to-computer interfaces (HCI) and computer-to-human interfaces (CHI). - The separate human-to-computer and computer-to-human tools may be categorized with respect to human-centredness (i.e., with respect to their ability to handle human communication and to match human behavior appropriately). - Human-centred interaction tools are well exploitable in human- to-human communication/interaction. - The computer-based interaction tools allow intentional computer-based processing of human messages. - As a consequence, a computer-supported human-to-human interaction scheme can be introduced and widely applied, resulting in elevated efficiency and reliability for human-to- human information transfer. The suggested approach is called HCHI (i.e., human-computer-human interaction). The key concept in HCHI is the intentional computer-based processing of human messages: intelligent machine support is provided to the human-to-human information transfer. This machine support covers not only the acquisition, storage, sorting, merging and retrieval of the transmitted information, but also the translation, formalization, re-recording, analysis and re- synthesizing of the human-to-human messages. Moreover, filtering, adjusting, correcting and storage of selected key facts from the messages is also involved. In this way, intelligent machines perform a transfer of related mental models by the suitable decoding/encoding of messages. Of course, HCHI in this context assumes appropriate expertise and a knowledge base built into the applied computers/interfaces (i.e., a form of AI-configured tools is postulated). It will be shown that HCHI, by using human-centred interfaces as computer-based tools in information exchange between humans, constitutes a new man-machine relationship. This new relationship enhances reliability, formal correctness and exactness (sometimes, even the adequacy and efficiency of the human-to- human interaction). On the other hand, the price for these benefits is (in most cases) a somewhat rigid and mechanized way of interpersonal communication. The suggested approach implies significant involvement of computerized tools in human communication, rather than the mere increase of convenience and efficiency of cooperation between the involved parties. Although the machine doesn't take over full control and regulation of human-to-human interaction, it strongly influences the human communication process. The extent of this influence depends on the requirements and on the practical implementation. Several characteristic examples from among the most feasible potential applications will be discussed. Some of these are simple utilizations of the suggested method (e.g., the exchange of mail between humans, the preparation of minutes of meetings and agreement memoranda, the control and correction of technical drawings, and the translation of man-made texts from one language into another). However, other utilizations discussed are really sophisticated applications (e.g., information exchange among members of an engineering design team, communication between operators of a complex industrial system, and messaging between persons sharing tasks in traffic control). Still other potential applications of HCHI will only be mentioned in passing, without further detail (e.g., banking, medical practice, education, and the law). In Section 2.0, the background for application of computer tools for enhancing human-to-human communication is discussed briefly. Section 3.0 goes into some detail concerning human-centredness and HCHI concepts. In Sections 4.0 and 5.0, some further theoretical and practical aspects of HCHI are investigated. Section 6.0 deals with potential applications, while Section 7.0 provides a brief commentary on the benefits and drawbacks resulting from the introduction of HCHI (and the related human- centredness). Finally, Section 8.0 considers prospects for applied HCHI, including a set of open questions to be answered in the future. It should be noted that only the basic concepts of the HCHI method are suggested in this paper. No actual, practical applications (or even working prototypes) are constructed. Only some qualitative experiences (e.g., with CAD/CAM systems) attest to the usefulness of the HCHI approach. Therefore, this paper can't present quantitative analysis of experimental data. However, it is hoped that discussion of the means to enhance exactness and reliability in human-to-human interaction is sufficiently intriguing to motivate future work in the direction described. 2.0 Background: Human vs. Machine Characteristics The human-computer interaction discipline has developed because of the many differences between humans and machines typical of man-machine systems (Callatay, 1986). Some of these differences between humans and machines (specifically, computers) as information transmitters, processors and receivers are listed in Table 1. Table 1 PERTINENT DIFFERENCES BETWEEN HUMANS AND MACHINES ------------------------------------------------------------ Property Humans Machines ---------------------- ---------------- ----------------- Information exchange Mostly verbal Mostly formal Information processing Loose formalism Firm formalism Stress in information Random/casual Stable/regular processing Communication Coupled channels Separate channels Information processing Lacking Present determinism Adaptability Built-in Limited Capabilities Changing Fixed ------------------------------------------------------------ Note: "Stress" signifies a very complex union of properties including disposition, inclination, focusing, concentration, care, apprehension, concern, pressure, and strain in sophisticated task-related behavior and attitude. Moreover, machine behavior may be considered causal, whereas human behavior is essentially non-causal. This means that machine input and output (I/O) relationships can be well modelled, but no adequate models are available for the description of human I/O relationships. In other words, proper machine reactions to well- defined stimuli are exactly predictable (for all but the most complex cases), while prediction of human reactions -- even to exactly defined changes of situation, circumstance, and motivation -- is extremely difficult. These discrepancies between human and machine properties suggest two new ideas. The first concerns the classification of human-to- computer and computer-to-human interfaces (in common: human- computer interfaces). The other is the potential for exploitation of machine characteristics in human-to-human interaction and communication. These concepts are introduced and investigated briefly in the following section. 3.0 The Basic Ideas: HCHI and Human-Centred Interfaces 3.1 HCHI: Human-to-Human Interaction Aided By Computers and By Human-Computer Interfaces There are a large number of applications where machine-based systems require human intervention to achieve proper system operation. The use of man-machine systems is made necessary by the fact that in most cases no fully automated operation can be achieved; computer-controlled subsystems are not able to cooperate correctly without the involvement of human operators. Among the many examples of how "fully automated" systems fail to complete their tasks appropriately are the following. - Design automation without human participation could not be achieved even in the most straightforward technical design problems (e.g., in electronics), because of the limitations of even the most complex physical models applied in the process. - Despite the high sophistication of onboard computers, without human intervention space vehicles couldn't handle appropriately the entirely unexpected events during long space missions. - Even by applying the most advanced automata and computer technology, nuclear power plants can't operate without human supervision, again because of the complexity of the technology. The main reason for these machine failures is the inability to anticipate all possible events in the models upon which their operating systems are based, in any high complexity application. It should be noted also that the requirements for human-to-human as well as human-computer interaction depend on the situation and circumstances under consideration. In general, the demands typical of human and/or machine activity depend on qualitative and quantitative task complexity, reaction speed requirements, spatial properties of events, the well- or ill-conditioned character of required actions, and the complexity and coverage of models implemented in machine support. From this, one can conclude that - Human judgment is essential if qualitative complexity is not addressed adequately by computerized models, or if the machine is ill-conditioned for possible events and required actions. - Machine involvement is necessary if rapid decisions are needed and/or spatial distribution of events doesn't allow human perception of all related parameters. It also is necessary if quantitative complexity restricts or prohibits purely human evaluation. - In all other cases, either humans or machines alone are able to perform the required job acceptably, provided that no extra parameters and no additional event types become required in the decisions and/or actions. However, because of the theoretically unlimited number of (closely or loosely) related parameters and because of the finite probability of unforeseen events, virtually any practical applications for which pure machine operation might be assumed still should involve human supervision or participation. Despite the commonly held conviction that human-machine interaction is a must in virtually all machine-implemented systems -- and that this kind of interaction poses many theoretical and practical problems -- it is rarely recognized that similar problems arise in cooperation between humans. However, an important new understanding has emerged recently: Acknowledgement that exactness and reliability in the collaboration of humans as system components may be considerably enhanced by introducing computers into the human-to-human interaction processes (much as the introduction of humans into machine-to-machine interaction-based systems is helpful). Consequently, the basic precept of this paper is the introduction of the computer as an intermediary component into the human-to- human transfer of information, both in cases of pure human-to- human interaction and of man-machine systems comprising several cooperating humans. This involvement of computers in human-to-human communication procedures (i.e., the use of "HCHI" rather than traditional HHI and HCI/CHI schema) requires a special approach to human-to- computer and computer-to-human interfaces -- an approach which at first sight may seem to be a strange idea (Why complicate the otherwise most natural and traditional means of human communication?). However, the properties of human-to-human information exchange and the requirements posed in many cases for this exchange make necessary (where possible) a new way of aiding humans in their interactions and cooperation. Consider the following three arguments why the introduction of such computer-based communication aids is required. - First, the number of operational systems where inappropriate decisions and actions may result in serious failures or even disasters is rapidly increasing (e.g., in industry, traffic, banking, and medical systems). Adequate and precise information exchange between humans is a must in these systems. - Second, the complexity of most operational systems is rapidly increasing, requiring machine support in human-to-human information transfer (e.g., in data acquisition, storage, filtering, selection, and retrieval). - Third, human attitudes, behavior and cultural background are continuously changing. Individuals become increasingly unique, with increasingly disparate knowledge, experiences, motivation and habits. Thus, a common understanding of what human messages mean and imply requires the more formal support provided by computerized tools. Despite its newness, the concept of HCHI may seem at the first sight to be somewhat familiar to those who are involved in one area of man-machine theory and practice: computer-supported cooperative work (CSCW). This rapidly evolving discipline also seeks to aid humans by using computers in their cooperation (Bowers, 1991; Galegher, 1990). However, CSCW addresses only one of the many practical applications of HCHI. It concentrates on human collaboration with the aim of achieving a well defined goal related to a specific activity (or of solving a well defined task). HCHI may be useful in a much wider spectrum of human-to-human communication and interaction, especially in relation to complex man-machine system operation. It differs from CSCW in its objective (aiding human-to-human communication/interaction in general, not merely supporting cooperative work). Therefore, HCHI must entail the significant involvement of computerized tools (intelligent interfaces and an AI-configured computer, for many essential information processing tasks), whereas CSCW (in the widely accepted sense of the acronym), provides instrumentation merely for increasing the convenience and efficiency of communication and cooperation among the involved parties). 3.2 Human-Centred vs. Machine-Centred Human-Computer Interfaces As we have seen, human-computer interfaces play an important role in HCHI. Human nature and human cognition/emotional behavior are important aspects in the construction of such human-computer interfaces. The most important related properties were listed in Table 1. A classification of HCHI interfaces is suggested below. This classification is based on the relationship of the interface characteristics to the aspects in Table 1 (i.e., how the interfaces behave and what properties they have with respect to human characteristics). Interfaces may be either human-centred (capable of dealing with human nature and behavior) or machine-centred (constructed by applying formal and rigid interaction schema which make for easy realization). Human-centred interfaces attempt to bridge the gap between human and machine properties, while machine-centred interfaces presume machine-like human properties (therefore cannot bridge a wide human-machine gap. Of course, constructing machine-centred interfaces is much easier than constructing human-centred interfaces, but machine- centred interfaces require much more adaptation by humans (familiarity with computer usage is assumed). Machine-centred interfaces may be constructed either intentionally (i.e., based consciously on a cost-related selection of machine-centredness rather than human-centred solutions), or because of a lack of resources necessary to the more demanding task of constructing a human-centred interface (e.g., knowledge and devices). The HCHI-based human-to-human interaction/communication principle suggested in Section 3.1 presumes human-centred interfaces. Application of the principle and of human-centred interfaces is the key to best exploiting the potential of computers in making human-to-human information exchange efficient and reliable (without assuming too much human familiarity with computers). 4.0 Elaboration of the Basic Ideas: Properties and Utilization of the HCHI Approach The introduction of computers into human-to-human information transfer as intermediary components is affected by two sets of factors: First, the main distinctive properties of human and machine participants in traditional man-computer interaction, and second, the characteristic situations of information exchange arising if humans are interacting with humans. The intermediary computer is a crucial tool in the HCHI approach. The enhancement achievable by means of this computer support is based on the capability of the introduced machine to transmit not only the formulated messages but also the mental models associated with individuals involved in the human-to-human communication. This transfer of mental models is useful in decoding/encoding the formulated messages. Computer processing can filter, adjust, and correct the messages as well as store key facts deduced from them. Processing, of course, assumes that appropriate expertise and a related knowledge base are built into the applied interfaces. Therefore, HCHI presumes an AI configuration which supports human-to-human communication and interaction (Barr, 1986 & 1989; Partridge, 1990). The degree of complexity and sophistication of this knowledge base and expertise is driven by the required HCHI properties and characteristics of the application. Potential applications (discussed in detail in sections 6.0 and 8.0) include - Perhaps most important, the area of industrial systems, in which humans should perform their tasks with an exactness commensurate to that of the technological processes themselves. - Medical practice, where even the survival of patients sometimes depends on the precise and clear exchange of information among cooperating doctors and assistants. - Business (especially, financial) administration, where the accurate, perfect and trustworthy transfer of information between humans is of key importance. In fact, it is hard to find any field of human activity or functions of man-machine systems where there is no long-range feasibility of exploiting the benefits provided by the HCHI principle. However, in the short term, the picture is somewhat less optimistic. First, HCHI is not a cheap technique. Second, there are a lot of open theoretical and technical questions regarding true HCHI solutions. And third, the dissemination of such an approach requires a long time. (Practical realizations will slowly disperse, appearing only in crucial applications at the beginning). The HCHI principle assumes the involvement of HCI and CHI devices (human-to-computer and computer-to-human interfaces). This role of HCI and CHI devices suggests a new way of categorizing interfaces: How human and machine properties are taken into account in the realization of interfaces. By using this categorization schema, we can distinguish between human-centred and machine-centred interfaces in the area of HCI and CHI, as well. The main differences between human- and machine-centred interface realizations are in how information is forwarded from the source (human or computer) to the target (computer or human). In the case of a human-to-computer interface, the system may transform human output (e.g., written or spoken messages) into a form acceptable as machine input. Alternatively, it may use fixed form (i.e., machine-centred) information transfer, which requires adaptation by the human to rules dictated by the computer (behind the interface). In free form (i.e., human- centred) information transfer, the interface should take care of interpreting the messages or information received from the human agent. In the case of a computer-to-human interface, the system may transform machine output (e.g., data files) into a form acceptable as a human input (i.e., human-perceivable signals and messages). Output towards the human is of fixed form. The human- centredness of the interface depends on whether the interface is ready to adapt to human needs. Such adaptation may allow freedom for the human, by making his or her interpretation and understanding of the (human-centred) output from the interface as easy as possible. The key component in computer-aided human-to-human communication is the human-to-computer interface. The human-centredness of the interface allows more freedom in the form of the information transfer, while machine-centredness implies more rigidity in the same information exchange. In order to be truly human-centred, an interface must have - Nearly free-form text input (i.e., no fixed form of information is presumed -- for example, there is no assumption of menu- driven or direct manipulation input), - Nearly free-form speech input (e.g., no reliance upon commands based on a limited vocabulary), - Nearly free-form graphical input (i.e., no fixed form editing of graphical constructs and drawings is necessary), and/or - Nearly free-form gestural input (i.e., no limited gesture-set is introduced). The introduction of computers (and thus human-to-computer and computer-to-human interfaces) into the process of human-to-human communication, with the aim of achieving the required interpretation and filtering by the computer, requires the most flexible and adaptive human-centred interfaces. It is through such systems that the potential advantages of computer-aided human-to-human interaction can be best exploited. Although adaptivity (to human needs, behaviors, properties, abilities, and stress states) and flexibility (in terms of changing circumstances, situations, and requirements) are doubtless crucial elements of human-centred interface operation, they are not the only properties which a truly human-centred interface must have. Such an interface must be able to - Perceive most forms of human information sources (e.g., sound or movement), - Recognize an appropriate variety of human message types (e.g., voice or gesture), - Comprehend a wide variety of background cultures and education levels, - Help with (and, if necessary, manage) a proper spectrum of the communication/interaction-oriented human actions, - Allow and initiate checking/control and acknowledgement of many kinds of transformed/processed messages, - Take over some human actions in case of the human's inability to continue communication/interaction, and (finally) - Make communication/interaction as easy and convenient as possible. In the approach described here, the traditional human-to-human interaction scheme is replaced by either human-computer-human interaction (pure HCHI) or by a more complex scheme combining both the traditional and the novel forms. HCHI would provide the advantages of - recording, - translating (by using a built-in vocabulary), - formalizing (by using a built-in syntax grammar), - re-recording, - analyzing (by using a built-in semantic analyzer), - re-synthesizing (by using a built-in knowledge-base), and - transmitting to the human addressee all the messages received by the machines from the originating humans (allowing exact but naked information transfer only). The more complex combination of traditional and novel forms may support in addition a more flexible information/message channel based on individual contact between the human participants. In considering the role of computers in human-to-human communication, the question arises of under what circumstances we may call machine involvement "significant" (versus considering the presence of machines only a convenience of information transfer). Building upon the previous discussion, it can be stated that the involvement of computer-based tools should be considered significant if - Machine support to the human-to-human information transfer covers acquisition, storage, sorting, merging and retrieval of the transmitted information; - Translation, formalization, re-recording, analysis and re- synthesizing of the human-to-human messages is included in the machine functions; - Computer processing involves filtering, adjusting and correcting (as well as storage) of selected key facts from the messages; - The transfer of mental models is performed by suitable decoding/encoding of the messages, and - Appropriate expertise and a related knowledge-base (i.e., some form of artificial intelligence) is built into the applied computers/interfaces. The same criteria may be used to determine whether any given system of computer-supported human-to-human communication and interaction should be considered a true HCHI solution. 5.0 HCHI In Practice: Some Remarks On How to Apply the Suggested Approach On the basis of the HCHI principle, a new man-machine relationship may be established by which human-to-human interaction achieves a new horizon, allowing more reliable, more formal and more exact (although more rigid and more mechanized) ways of exchanging information. The realization of HCHI would feature the implementation of human-to-computer interfaces and computer-to-human interfaces as intermediary components built into the process of human-to-human communication. (The interfaces would support communication, act as catalyst and/or mediate or negotiate between humans.) Such an involvement is possible in any manned system where human-to-human interaction is a critical element in the operation of the system. In contrast to the traditional human-computer interaction practice in which discrepancies between human and machine are considered inconvenient (and problems stemming from these discrepancies are to be avoided or eliminated), the HCHI approach directly exploits such discrepancies. The properties which are lacking (causality, firm formalism, homogeneity, channel separation and determinism) are provided by the inserted machine negotiator. Thus, the versatile capabilities and flexible adaptivity of the humans may be coordinated, adjusted and integrated by the more rigid machine cooperator. The result is that the computerized tool involved in HCHI enhances human-to- human communication. This is done, in cooperation with the communicating humans, by improving the quality of message transfer and by increasing the reliability of exact and true message interpretation on the recipient side. A critical feature of the machine-based form of human-to-human interaction is its ability to introduce formalized protocols into the communication processes among humans. Such protocols may be adjusted to the needs and preferences of the humans. Despite being adjustable, the protocols will retain exactness and full handshaking in the information transfer. They should (in most cases) allow and even motivate or force originating humans to check, control, critique and acknowledge machine interpretations before the messages/information is stored-processed-forwarded to recipient humans. Such validation is especially important in terms of the key facts extracted from related messages. 6.0 Applications: From Simple Bilateral Conversation to Large, Complex Industrial Systems A wide spectrum of man-machine systems rely heavily on human-to- human interaction and on the quality and outcome of human-to- human communication. Instances include information exchange among members of an engineering design team, communication between operators of a complex industrial system, and message transactions between persons sharing tasks in traffic control. Future applications of HCHI may occur in fields as diverse as banking, medical practice, education and law. Although it may be difficult to implement HCHI in fields of this sort, all include activities for which the exactness and reliability of human-to- human information exchange are of key importance. There are some very simple everyday applications essentially ready for HCHI. These include - Interpersonal correspondence (not only would the correctness of grammar and style be evaluated and corrected by the computer, but facts (dates and times, places) also would be verified). - Minutes of meetings and agreement memoranda could be checked for correctness, exactness, and exact information content (depending on how much intelligence is provided by the AI knowledge base and inference engine). - Original man-made texts could be translated from one language into an other. In translation, the equivalence of the source and target texts sometimes is extremely important. However, really effective machine-supported translation would need to mirror the full intentional content and cultural context of those communicating, a capability that probably won't be achieved in the near future. There also are a number of crucial industrial applications that could be improved by HCHI, from simple cases of graphical information exchange to the most complex design and planning activities (Begg, 1984; Cushman, 1991). - Technical drawings (especially sketches) are perhaps the most important traditional form of information exchange between engineering and technical staff members in industrial organizations. Filtering and correcting such graphical information carriers would be important to achieving error-free technical information exchange. - Integrated functional, electrical and physical design of electronic systems and circuits is an excellent example of computer aided/controlled human-to-human information transfer. The traditional design process results step-by-step in a rapid increase of design information and yields, for example, drawings, master artworks, and numerical control files for production/testing. This process suffers often as a result of uncontrolled interactions among the functional, electrical and packaging designers. HCHI could add desperately needed formalism to the communication process. - Another potential HCHI application would involve the complex, integrated process of engineering design (product design) and process planning (technological planning) in a production or manufacturing organization. Contradictions between design concepts/targets and associated manufacturing processes sometimes results in a failure to meet original specifications. HCHI could help avoid misunderstandings and misinterpretations, in support of efficient and reliable engineering activities. Despite its advantages and wide potential for application, care should be taken in terms of where and when to use HCHI. Its most important limitations, weaknesses and shortcomings should be kept in mind. - HCHI probably will be an expensive technique, especially in the early stages of it's application. (However, human-centred interfaces are expensive, in general.) - Many open theoretical and technical questions still must be answered before true HCHI solutions can be achieved. - The wide dissemination of HCHI principles can be anticipated only in long range. Practical applications will emerge slowly and sporadically, especially during the starting period. - At the beginning, HCHI solutions probably will appear only in critical applications. The first proof of concept application areas probably will be the most demanding man-machine systems (those that require exact and precise human-to-human information exchange). - The first HCHI implementations predictably will suffer from users' feelings of being inconvenienced, an effect introduced by the approach itself. Until truly efficient solutions appear, users may feel that the new way of communicating is irritating and disturbing, compared to traditional (informal) human-to- human interaction. The effect is similar to that observed in even the simplest cases of introducing computer techniques into traditional organisations and operations (Card, 1986; Salvendy, 1987). Thus, at present HCHI might be considered just a collection of principles with considerable feasibility and an evident potential for future utilization. 7.0 On the Benefits and Drawbacks of the HCHI Approach The intent behind introduction of the HCHI approach (and of human-centred interfaces) is to establish a new man-machine relationship. Via this relationship, human-to-human interaction/communication becomes more reliable, more formal and more exact, although more rigid and more mechanized. The significance of applying the HCHI concept depends on how important exactness is in the information traffic between humans. Such exactness seems to be most important in technical information exchange (situations in which facts and data may play a critical role in adequate communication). The idea of introducing computers into human-to-human communication is not entirely new. It has been attempted in several practical applications, especially the use of human-to- human message/information exchange tools when humans are distant from each other in time and/or space. However, in most cases the unaltered recording or transparent transmission of the human messages is the only aim; no intentional filtering is involved. The key differences between such earlier attempts and HCHI are that HCHI is targeted at - allowing a more general approach to all cases of human interaction/communication requiring exactness, and - allowing intentional computer-based intelligent processing of the human messages. Table 2 contrasts HCHI to traditional HHI (human-to-human interaction), HCI/CHI (combined human-to-computer and computer- to-human interaction) and CSCW. Table 2 CHARACTERISTICS OF COMPETING INTERFACE PARADIGMS ------------------------------------------------------------ Characteristic HHI HCI/CHI CSCW HCHI ---------------- ---- ------- ------- ------ Efficiency high low medium medium Reliability low medium medium high Cost low high high medium Limitations many average average few Applications wide limited bounded wide Future horizon wide medium medium wide Open questions few many average many ------------------------------------------------------------ Analysis of Table 2 indicates that HCHI probably is inferior to HHI in terms of efficiency, cost and the number of open questions, and to CSCW in terms of the number of open questions. However, considering typical applications and normal circumstances, HCHI appears to be predictably superior to all other interface paradigms in terms of other characteristics. The many open questions related to HCHI should be taken, however, as a warning about the lack of immediate feasibility in applying HCHI in candidate applications. Clearly, overestimating the potential of the suggested HCHI approach should be avoided. 8.0 Outlook: Potential Applications and Open Questions Future applications may include not only the most obvious technical fields, but also a wide spectrum of society related and political affairs situations. In all potential application areas, misunderstandings, disregarded messages and a lack of information frequently cause virtually unavoidable problems -- problems which might be bypassed through use of HCHI. However (as noted in Table 2), there remain a number of open questions regarding computer-aided human-to-human interaction. - Is it feasible today to develop the amount of intelligence, expertise and knowledge necessary to realize true HCHI solutions? - Is HCHI truly totally different from HCI/CHI and CSCW, or is HCHI merely a tangential approach benefiting from the successes of other paradigms? - Are humans ready to communicate/interact with each other by using HCHI-type support, or will resistance with respect to such a novel "tool" prohibit HCHI's introduction in the near term? - What truly straightforward applications of the HCHI principle are possible in the near future (if any)? - Is it possible to model human characteristics, habits, behavior and cognition in such a way that they can be represented in and manipulated by likely AI configurations supporting the computer/interface? - Is the potential cost/benefit ratio sufficient to permit/validate the application of HCHI in practical applications -- at least in critical industrial man-machine systems characterized by high risk of potential disasters? Answering these questions by building a broader theoretical foundation and conducting practical experiments may help advance the chances for successful and wide-ranging future applications of the HCHI approach. The primary goal of this paper has been to motivate and inspire future research into the challenging area of how machines may aid in one of the most subjective human activities: interpersonal communication. 9.0 Acknowledgement This paper is based on a talk presented by the author at the People and Computers VIII conference (HCI'93) in Loughborough, UK, during September 1993. Thanks are due to all who provided a preliminary critique and who made comments regarding the paper, especially to Donald L. Day for his support and assistance. The author also would like to express his thanks to the Hungarian National Research Fund (OTKA) for supporting the related research. 10.0 References Barr, A., Cohen, P.R., and Feigenbaum, E.A. (1986 & 1989). Handbook of Artificial Intelligence, Vols. 1 & 4. Reading, Mass.: Addison-Wesley. Bayley, R.W. (1982). Human Performance Engineering: A Guide for System Designers. Englewood Cliffs, N.J.: Prentice Hall. Begg, V. (1984). Making Computer Aided Design Tools More Usable: A Study of A Complex Task Shared By People and Machines. London: Kogan Page. Bowers, J.M. & Benford, S.D. (eds.) (1991). Studies in Computer Supported Cooperative Work: Theory, Practice and Design. Amsterdam: North-Holland. Callatay, A.M. (1986). Natural and Artificial Intelligence. Amsterdam: Elsevier. Card, S.K., Moran, I.P., & Newell, A. (1986). The Psychology of Human-Computer Interaction. Hillsdale, N.J.: Lawrence Erlbaum Associates. Cushman, W.H. & Rosenberg, D.J. (1991). Human Factors in Product Design. Amsterdam: Elsevier. Galegher, J., Kraut, R., & Egido, C. (eds.) (1990). Intellectual Teamwork: Social and Technological Foundations of Cooperative Work. Hillsdale, N.J.: Lawrence Erlbaum Associates. Helander, M. (ed.) (1988). Handbook of Human-Computer Interaction. Amsterdam: North-Holland. Partridge, D. & Wilks, Y. (eds.) (1990). The Foundation of Artificial Intelligence. Cambridge, Mass.: Cambridge University Press. Rasmussen, J. & Andersen, H.B. (1991). Human-Computer Interaction. Hillsdale, N.J.: Lawrence Erlbaum Associates. Reilly, R.G. (ed.) (1987). Communication Failure in Dialogue and Discourse. Amsterdam: Elsevier. Salvendy, G. (ed.) (1987). Social, Ergonomic and Stress Aspects of Work with Computers. Amsterdam: Elsevier. Vickers, D. & Smith, P.L. (eds.) (1989). Human Information Processing: Measures, Mechanisms and Models. Amsterdam: Elsevier. ------------------------------------------------------------ **************************************** * * * Dr.Lajos Balint * * senior councillor * * Hungarian Academy of Sciences * * Department of Natural Sciences * * Budapest, Nador-u.7. * * H-1051 Hungary * * Tel: + 36 1 117 5700 * * Fax: + 36 1 117 8486 * * e-mail: h48bal@huella.bitnet * * * **************************************** _____ Articles and Sections of this issue of the _Electronic Journal on Virtual Culture_ may be retrieved via anonymous ftp to byrd.mu.wvnet.edu or via e-mail message addressed to LISTSERV@KENTVM or LISTSERV@KENTVM.KENT.EDU (instructions below) or GOPHER gopher.cic.net Papers may be submitted at anytime by email or send/file to: Ermel Stepp - Editor-in-Chief, _Electronic Journal on Virtual Culture_ M034050@MARSHALL.WVNET.EDU _________________________________ *Copyright Declaration* Copyright of articles published by Electronic Journal on Virtual Culture is held by the author of a given article. If an article is re-published elsewhere it must include a statement that it was originally published by Electronic Journal on Virtual Culture. The EJVC Editors reserve the right to maintain permanent archival copies of all submissions and to provide print copies to appropriate indexing services for for indexing and microforming. _________________________________ _________________________________ _THE ELECTRONIC JOURNAL ON VIRTUAL CULTURE_ ISSN 1068-5327 Ermel Stepp, Marshall University, Editor-in-Chief M034050@Marshall.wvnet.edu Diane (Di) Kovacs, Kent State University, Co-Editor DKOVACS@Kentvm.Kent.edu ____________________________ GOPHER Instructions ____________________________ GOPHER to gopher.cic.net 70 ____________________________ Anonymous FTP Instructions ____________________________ ftp byrd.mu.wvnet.edu login anonymous password: users' electronic address cd /pub/ejvc type EJVC.INDEX.FTP get filename (where filename = exact name of file in INDEX) quit LISTSERV Retrieval Instructions _______________________________ Send e-mail addressed to LISTSERV@KENTVM (Bitnet) or LISTSERV@KENTVM.KENT.EDU Leave the subject line empty. The message must read: GET EJVCV2N2 CONTENTS Use this file to identify particular articles or sections then send e-mail to LISTSERV@KENTVM or LISTSERV@KENTVM.KENT.EDU with the command: GET where is the name of the article or section (e.g., author name) and is the V#N# of that issue of EJVC