In this paper I argue that forecasting presents us with a paradox. Our ability to make reasonable and accurate medium to long term social and technological forecasts is demonstrably dismal. The gap between what actually happens and what was predicted is usually huge. Promoters and technological forecasters continue to provide us with wide varieties of technological futures. The paper introduces a tentative typology of technology futures as a framework for empirical work on failed technology futures. Sociological, psychological and historical explanations (...) are considered for the forecasting paradox. But primarily I argue this is an ethical question. Forecasting is akin to promising and it is in this light it ought to be assessed. Forecasters are accountable for their forecasts in the same way I may be held accountable for my promises. (shrink)