Pitcher abuse points are an attempt to measure overuse. They're based on the idea that once a pitcher reaches a point of fatigue, each additional pitch significantly increases the risk of injury due to fatigued muscles not handling the stresses of pitching combined with the deterioration of mechanics.
The metric has evolved over time. Initially, it assigned one or more abuse points to each pitch over 100 per game with higher pitch count ranges getting more points per pitch. Then the metric evolved such that each pitch over 100 was counted as 3 pitches (i.e. was tripled). Finally, it was decided that the higher the pitch count goes over 100, the more of an impact each pitch has so the metric evolved such that the number of pitches above 100 was cubed instead of tripled. This is where the "P-cubed" name comes from. By the way, 100 was adopted as the magic number that most pro pitchers could get to before showing signs of fatigue. Obviously, the actual number differs from pitcher to pitcher but this was used as the average.
Now, one important point to note is that there seems to have been only limited studies as to whether the P-cubed number translates to injuries or anything else important. If you search, you'll find articles that are quick to point this out. But you'll also find that there seems to be quite a few folks paying attention to this metric. Intuitively, the concept makes sense. But there are too many vairables in the injury equation and it's hard to establish any direct correlations. So, take this with a grain of salt. It is still an interesting idea.
The following article provides good background for the original pitcher abuse points metric:
And this is a follow-up article to the one above:
This is a recent assessment of pitchers in the big leage. Guess who 2 of the top 3 pitchers in abuse points are... Wood and Prior. Prior also had a high P-cubed number in college.
An interesting article on new Red Sox pitcher Matsuzaka and his abuse points: