I cannot forecast better than Dave Forney. It is indeed very hard to tell what will be done even within five years. I think it is safe to predict that information theory is go- ing to be applicable in an even broader field than now. I think that in computer science information theory will be more important. I look forward to more substantial applications of information theory in statistics, in combinatorics. Now in information theory itself what will be the new basic results and new directions, I do not dare to forecast. I suggest that systems with memory will be studied more intensely -- the large part of knowledge today is mainly for the memoryless case. Of course the systems with memory are much more difficult, so we can hardly hope for a comprehensive theory of those. But I think it safe to predict that one will do more work on systems with memory. One can forsee development in the field of networks. As Prakash mentioned it would be very important to achieve such results, and it would be desirable if more talented young people worked on these problems; then one could predict many impor- tant new results. Well, everything depends on the quality of people who will work on the problems.