Is there a relation between the way people are in real life and the way they are in games? People differ in their personality, culture, and position, and the decisions they make are for a part determined by these differences. We have researched the relation between personality and player motivation and how virtual environments can be adapted to different types of people. Furthermore, persons' characteristics influence the way one collaborates with others. We are developing models (for gaming characters) that incorporate different decision making styles. These models should result in more human-like characters and in improved collaboration between players and game characters.

When processing large data sets that do not fit in the main memory of a computer, data has to be swapped between memory and disks during the computation. Since accessing the disk easily takes a million times longer than an elementary operation on the CPU, disk access ("I/O") will often dominate the running time of the computation. The hardware can help by transferring data to and from disk in large blocks, transferring lots of bytes in almost the same time as accessing a single byte. But which algorithms can really take advantage of these extra bytes, and which algorithms still need a different block to be loaded into memory at every step of the computation?

It turns out that algorithms designed for data that fits in main memory are often horribly inefficient when data is stored on disk. This inspired a large area of research to design so-called "I/O-efficient" algorithms. Many useful standard techniques were found to make algorithms I/O-efficient. Still there are some seemingly trivial problems (particularly graph traversal problems) for which it is not known how to solve them optimally with respect to disk accesses. In this talk I will discuss the standard model to analyse the I/O-efficiency of algorithms and discuss some basic results and open problems in the field.