Face the reality. For a program to put a single pixel on the screen, the call will have to go through at least five layers of code, each coded by a different person. There is something like 30 miliion lines of code in XP, not including drivers and applications. It would be naive to think that much code could be made without some errors or that every contingency could be tested. The advantage compared to the DOS days is that the programmer does not have to care about what graphics card (or other hardware) is used. If there is a bug in the drivers for the graphics card, it is the responsibility of the manufacturer to fix it, not the application programmer. In DOS, you had to know exactly how to control the card/printer/whatever, in effect you had to more or less make your own driver. There is a huge difference between having to provide support for hardware variations and just identifying bugs in existing drivers. Sorry Ian, but your comparison is just plain stupid. /Troberg