in reply to [OT] Perl / Computer Science Science Fair Projects
I won the Maine State Science Fair in the "Talks And Demonstrations" category. That category is the longer and more thorough category where there are subject matter experts judging you. I am not telling you this to boast, but rather to lead credence to what advice I have.
If your sister is interested in winning and doing well she needs to realize it may be in direct conflict of doing an exciting project that she has fun with. At its core, a science fair is about testing a hypothesis using the scientific method and ensuring the results are statistically valid. Display and presentation will go a long way at the local level, but if she advances - everyone she will be competing against will have a whiz-bang show to put on. At that level, it will boil down to fundamentals.
She should decide early on if she wants the software portion of the project to be the experiment itself, or if it is going to be a tool in facilitating and validating the experiment. Let me explain. Conway's Game Of Life is an example where the program itself could be the experiment. It is possible to start with a few basic set of rules that evolves into a dynamic system. On the other hand, you could do like I did and use the program to perform statistical validation on the results of a psychological hypothesis (it was 1993).
The hypothesis could be that it is possible to write a tic-tac-toe computer player that is impossible to beat (ties are possible) using only N rules. It could be that she uses a program to test some advanced game theory (I have Scientific American magazines with lots of great ideas). No matter what, at the local level it will be more about show than substance and at the advanced levels it will be all about substance and how scientific the results and process are.