This year marks the 101st anniversary of the sinking of the “unsinkable” RMS Titanic. On April 15, 1912, the Titanic struck an iceberg in the North Atlantic Ocean on its maiden voyage from Southampton, UK, to New York City. There was no single cause for the loss of the Titanic, rather the improbable combination of errors in human design and decision combined with unforeseeable circumstance lead to the loss of over 1,500 lives. The failure appears to have occurred over a range of spatial and temporal scales – from the atomic-scale process of embrittlement of iron rivets to global-scale fluctuations in climate and ocean currents. Regardless of the specific combination of causes, this failure in design and practice led to impressive improvements in both. Disaster and tragedy are harsh teachers, but critical to improvement and progress.

The important question for the nuclear waste management community is how do we learn and improve our waste management strategies in the absence of being able to fail. A geologic repository “operates” over a very distant time frame, and today’s scientists and engineers will never have the benefit of studying a failed system. In place of a failure that is followed by improvement and progress, we can only offer a general consensus on disposal strategies supported by a wide array of evidence and risk assessments. However, it may well be that consensus leads to complacency and compromise, both of which are harbingers of disaster. With this concern in mind, this is the time to review our fundamental approach, particularly the methodologies used in risk assessments that have us calculate risk out to one million years. The structure of standards and implementing regulations, as well as the standard-of-proof for compliance, should be reexamined in order to determine whether their requirements are scientifically possible or reasonable. The demonstration of compliance must not only be compelling, but must also be able to sustain scientific scrutiny and public inquiry. We should benefit from the sobering reality of how difficult it is to anticipate future failures even over a few decades. We should be humbled by the realization that for a geologic repository we are analyzing the performance, success vs. failure, over spatial and temporal scales that stretch over tens of kilometers and out to a hundreds of thousands of years.