Why does technology diffusion work to enhance productivity in some countries but not in others? We have examined panel data from 84 countries over fifty years to analyze which factors contribute to how productivity improvements diffuse from the technological frontier to those within it. We find that there is convergence of countries onto the frontier only amongst the very richest countries in our sample (OECD), and that there is little if any sign of convergence by others. This results in larger productivity gaps between the richer countries (at or near the frontier) and the poorer ones, and increasingly so for the very poorest. What can be done to aid diffusion to the problematic countries? We find that, for the very poorest countries, the most basic sorts of investments are required before absorption of technological change can occur. Educational attainment is an especially significant factor in poor countries, but of little significance in middle income states. Then, as basic needs are met, public policies become more important factors for diffusion in middle income countries. The openness of the economy becomes significant, as diffusion increases with the easy entry of new products and new technologies. Secondly, substantial public investments in R&D are also important - to help in the transfer of technologies from abroad. Finally effective IPR policies are important - to encourage private sector investments in diffusion. Ironically, these same policies have little or no benefit for the very poorest countries.