Technological choices largely determine the long-term characteristics of industrial society, including impacts on the natural environment. However, the treatment of technology in existing models that are used to project economic and environmental futures remains highly stylized. Based on work over two decades at IIASA, we present a useful typology for technology analysis and discuss methods that can be used to analyze the impact of technological changes on the global environment, especially global warming. Our focus is energy technologies, the main source of many atmospheric environmental problems. We show that much improved treatment of technology is possible with a combination of historical analysis and new modeling techniques. In the historical record, we identify characteristic "learning rates" that allow simple quantified characterization of the improvement in cost and performance due to cumulative experience and investments. We also identify patterns, processes and timescales that typify the diffusion of new technologies in competitive markets. Technologies that are long-lived and are components of interlocking networks typically require the longest time to diffuse and co-evolve with other technologies in the network; such network effects yield high barriers to entry even for superior competitors. These simple observations allow three improvements to modeling of technological change and its consequences for global environmental change. One is that the replacement of long-lived infrastructures over time has also replaced the fuels that power the economy to yield progressively more energy per unit of carbon pollution - from coal to oil to gas. Such replacement has ＇＇decarbonized" the global primary energy supply 0.3% per year. In contrast, most baseline projections for emissions of carbon, the chief cause of global warming, ignore this robust historical trend and show little or no decarbonization. A second improvement is that by incorporating learning curves and uncertainty into micro scale models it is possible to endogenously generate patterns of technological choice that mirror the real world. Those include S-shaped diffusion patterns and timescales of technological dynamics that are consistent with historical experience; they also include endogenous generation of "surprises" such as the appearance of radically new technologies. Third, it is possible to include learning phenomena stylistically in macro-scale models; we show that doing so can yield projections with lessened environmental impacts without necessarily incurring negative effect on the economy. Arriving on that path by the year 2100 depends on intervening actions, such as incentives to promote greater diversity in technology and lower barriers to entry for near infrastructures that could accelerate historical trends of decarbonization.