The problem of how to discount values in the far future is reviewed, and shown to lead down a blind alley. An alternative is proposed that allows long term consequences to be addressed by decisions using a relatively short term time horizon. A simple model investigating the optimal containment of radioactive waste in a deterministic world is used to show that current generations can indeed cater for the interests of the far future while optimising over the short term; however, this is not always possible. The proposed method seems to address most of the critiques of long term discounting found in the literature.