Although informatics as science developed successfully, in some topics its progress is slow, and these topics are frequently of great fundamental or technological importance. The status of our knowledge in informatics is often insufficient to allow exploitation of existing or emerging technological possibilities. Moreover, informaticians know of many problems that have good solutions where nevertheless their entering practical work is inhibited. Finally, users point out that often solutions that are applicable in principle do not meet practical needs. Successful advancement of informaticians requires tight cooperation. The same is true for users of informatic knowledge; technologists and informaticians will need each other urgently.

In the following, some of the challenges for informatics are considered. Many of them have long been the subject of intense work in the TUM faculty of informatics and elsewhere.

The basic challenge for informatics (as for any other engineering science) is to construct from given specifications a technical system. Frequently, the exact formulation of the specification is a big hurdle. The range of systems that the informatician can have control over has been expanded to include more and more complex systems. It is foreseeable that increasing demands will require further improvements in design methodology. Because of the importance of information processing in all technical systems, progress in this area has consequences in all engineering sciences. The improvement of design methodology is being tackled from many sides, e.g., by the development of languages and tools. However, it also requires a better insight into the behaviour of the human being as a designer as well as that of groups cooperating in the design of a system.

In many applications, it is essential that the system and the processes performed by it are safeguarded against unauthorized manipulation. This may mean that any intervention into the system must be recorded. In other cases, it is required that persons can use the system without revealinging this fact to third parties. While techniques fulfilling such demands have been known for some time, their cooperation for the protection of large systems is an essential challenge.

A fundamental requirement is that a system does not commit errors - at least not undetected ones. No technical system will fulfill this requirement completely. To protect the environment from the aftermath of errors is a fully obvious requirement for those technical systems where the computer exerts control that could harm the environment. Much less understood but nevertheless more important is the prevention of damage to our juridical, economic, social, and political order by the mass use of computers. Not well understood either is the effect of computer use on the thinking and communication behaviour of single persons and groups.

A further challenge for informatics is the participation of laymen in designing places to work and methods of work based on computers. For reasons based on technological constraints, we have become used over decades to seeing the human being as the servant of the computer. But we have not learned how laymen can shape their intercourse with computers according to their needs.

The last two decades have greatly changed the technique of working with computers. The rigid exchange of texts has been replaced by interactive, graphical work. But still the outside picture of computers is determined by keyboards and monitor screens, which are inadequate for input and output purposes. Computers that adapt themselves to the needs of the user must be usable like active writing and drawing tables and must allow input and output of pictures and texts as well as language. Though techniques for doing so have been known for some time, they are far from perfect, and appropriate interaction techniques are still missing. Nevertheless, "multimedia" information representations can be found now on cheap handheld calculators. At present, single output techniques that have proven to be usable somewhere, like text, audio, and video output, are merely combined. The development of truly appropriate techniques for information representation on communication channels is expected to be rather detached from traditional forms, and the development of new forms of interaction between human beings and technical systems is a particular challenge.

An additional challenge is the growing bandwidth of technical communication. In the decades to come, increased use of glass-fiber cables and satellite lines will overcome existing limitations.

For no more than twenty years have we been freed from the idea that a computer follows a prefabricated program describing for every variant of the possible progress a concrete course of solution. The introduction of knowledge-based methods has resulted to some extent in the representation of the knowledge necessary for the solution in a suitable form within the computer and in the introduction of a general problem solving method - frequently based on a long chain of deductions - which generates solving programs. Present-day techniques of knowledge representation and deduction are basically quite powerful, but for many application fields practically insufficient. This is so in particular with fuzzy or time-dependent knowledge. Most likely, informatics will have to live with a whole spectrum of different techniques for knowledge representation and deduction. Transitions between these techniques are yet to be developed. First steps in the automatic acquisition of the necessary knowledge are today made. These techniques will be extended substantially, and we shall learn how from the concrete knowledge efficient abstractions and models may be derived.

For decades, informaticians have endeavoured to better understand concurrency and distribution, two fundamental degrees of freedom in the design of information processing systems, and to make them technically usable. The technological provisions are extremely favorable, thanks to the existence of microelectronics, which gives reliable processor components at low cost and broadband communication. Nevertheless, progress is slow, indicating that mastery of these degrees of freedom will be a challenge for some decades. What is lacking in particular is the availability of global concepts of division of labor and cooperation in systems consisting of many, largely autonomous actors. While today the decomposition of system functions with respect to concurrency and distribution is an essential step in systems design, there are conceivable systems, built from largely autonomous actors, which organize themselves competitively so as to establish collectivelyly the system function.

Around 1980 the idea of open systems was born. They originate from components whose outside behaviour follows consistent concepts and which thus can be combined freely. It is of great economic importance that open systems promise emancipation from dependency on single manufacturers. The practical use of this concept is still difficult and in many cases requires too much of informatics to be of help for those for whom it is intended. At the same time it can be expected that conventions for open systems will have to change, or progress in informatics will be inhibited. Thus, the interoperability of components of information processing systems - whether devices or programs - is a very important challenge for the coming decades.

Another task of informatics is the management of storage and retrieval of information. The technology of magnetic and optical storage already allows storage of all the information a person may use privately or professionally in a very compact and secure way. Exposing this information to machine processing offers new methods of retrieval, which are needed to make good use of an overwhelming flood of information.

A last challenge, maybe at first sight less conspicuous, is presented by the invasion of informatics as a working technique in other sciences. Surpassing the present ubiquitous drafting of texts with the help of a computer, we think of elementary techniques of the human brain which have been known for a long time, but for unavoidable practical reasons have been computerized only rarely: Gedanken experiments, simulation allowing one to investigate experimentally on the computer the behaviour of arbitrary fictitious or real systems. The behaviour can be visualized like a movie, can stimulate phantasy, or can suggest consequences or variations of the experiment. In this context, techniques for automatic proving of theorems can help to verify or falsify hypotheses. Inquiries in a knowledge base will possibly be used to clarify the circumstances of an investigation.

Such techniques will change science, as has happened already in some cases. It will be necessary to support intuitive thinking, or the use of computers will lead to impoverishment instead of enrichment.

Friedrich L. Bauer
Wilfried Brauer
Eike Jessen
Manfred Broy