If R&D difficulty increases as it accumulates, as recent evidence suggests, then there is a value in the diversification of the aggregate R&D efforts over the whole range of industries in the economy. However, this paper proves that the quality ladders models allow for self-fulfilling prophecies to divert R&D from an arbitrary number of sectors, with potentially dramatic effects on the dynamics of industry structure and on the long-run aggregate growth performance. Moreover, in developing countries, perfect markets might allow stagnant monopolies to resist despite the removal of fundamental entry barriers to dynamic competition. This kind of coordination failure implies potentially strong effects of socio-cultural factors on growth and suggests a positive role for public R&D policy even in a framework of semi-exogenous growth.