I didn't check here for a few days, sorry.

As I think Chris was saying, a slip predictable model argues there is a critical stress at which an earthquake occurs, so a larger slip means a longer time until the next earthquake on a fault segment.

Time predictable means there is a constant stress left behind after an earthquake, so the longer it has been since the last earthquake, the bigger will be the earthquake when it finally occurs.

The models have been tested in a variety of recurrences of characteristic earthquakes - I think Justin Rubenstein has such an abstract for AGU this year for Parkfield events - but neither generally works very well.