This paper addresses the design of user interfaces for aging adults. Older people differ vastly in how aging affects their perceptual, motor, and cognitive abilities. When it comes to interface design for aging users, the "one design for all" approach fails. We present first results from attempts to extend ability-based design to the aging population. We describe a novel approach using age-related differences as the principle of optimizing interactive tasks. We argue that, to be successful, predictive models must take into account how users adapt their behavioral strategies as a function of their abilities. When combined with design optimization, such models allow us to investigate optimal designs more broadly, examining trade-offs among several design factors. We present first results on optimizing text entry methods for user groups with different age-related declines.