We investigated active deformation within the Akato Tagh restraining double bend to determine the age of the active Altyn Tagh fault relative to the Altyn Tagh system and thereby evaluate the extent to which this system evolved by net strain-hardening or softening. Active structures were mapped based on their geomorphology and disruption of Quaternary(?) deposits/surfaces. The style of active faulting is strongly correlated with fault strike: 065°–070° striking segments show pure left-slip whereas faults with more northward or eastward strikes are transtensional or transpressional, respectively. Our mapping further suggests that the 065° to 070°-striking western and eastern segments of the Akato Tagh bend are characterized by pure strike-slip motion, with partitioned transpression along the ∼090°-striking central segment of the double bend. It remains unclear how this active, bend-perpendicular shortening is absorbed. In conjunction with previous work, the present study fails to support the commonly held idea that Tarim-Tibet motion is strongly oblique to the Altyn Tagh system. Estimates for the age of the Akato Tagh bend derived in a companion study suggest the bend is only a few million years old. The current principal trace is probably similarly young because formation of the bend by recent deformation of an old trace should result in transpression along the western and eastern segments, contrary to the pure left-slip shown here. The Altyn Tagh system comprises multiple fault strands in a zone ∼100 km wide across strike. Because the main trace appears to be much younger than the system in which it is embedded, we speculate that this system evolved by the sequential formation and death of short-lived fault strands. In particular, we suggest that geometrically complex strike-slip fault systems such as the Altyn Tagh may form via system strain hardening, where this net response reflects a dynamic competition between hardening and softening processes that are active simultaneously within the fault zone. Hardening mechanisms may include growth of restraining bend topography or material hardening of phyllosilicate-rich gouge, whereas softening processes might include R-P shear linkage or reduction in bend angle by vertical axis rotation. Our analysis suggests that net strain hardening of a fault system can produce continental deformation that is spatially localized over the 1–5 m.y. during which an individual strand is active, but distributed over the 10–100 m.y. corresponding to the life-span of the whole fault system and the collision zone in which it is contained. Thus, time scale is critically important in determining whether or not continental deformation is spatially distributed or localized.