This paper describes a dynamic analysis of technological advances among hard disk drive (HDD) manufacturers in the areal density of their products across the history of the industry. The study provides (additional) evidence supporting a view of technological racing with leap-frogging rather than with persistent leadership domination in the HDD contextlike others, we find that technology leaders in one year are less likely to innovate in the next year than those firms right behind them in technology. We also uncover new evidence that technological laggards do not disappear as quickly as expected by technology racing metaphors where a winner-take-all outcome is expected. Our efforts to explain this pattern of persistent heterogeneity with the usual kind of strategic positioning stories and specifications were not notably successful. But we did find evidence that technological innovation in HDD follows a trajectory consistent with a proportionate random process (akin to a Gibrat process) that favors technology leaders but only stochastically. We demonstrate through simulation that evolution in a population with selection favoring a characteristic evolving as a proportionate random process generates increased variation. This contrasts with a common social science framework for viewing evolution that assumes a fixed characteristic and implies decreasing variation.