How confident should courts be in the opinions of expert witnesses testifying at the bleeding edge of science when the life, liberty or property of a citizen hangs in the balance? Not very, if a new study is any indication. The authors of Conflicting Biomedical Assumptions for Mathematical Modeling: The Case of Cancer Metastasis wanted to build an exemplar mathematical model of cancer metastasis and open it up for testing and systematic evaluation. Instead they found that none of the twenty-eight leading academic experts in metastasis could agree on even the basics of the process. There were, in fact, as many opinions about the course of metastasis as there were researchers.

The authors found that a wide range of incompatible assumptions are held by scientists studying the same subject and that no two experts advanced identical scenarios for cancer metastasis. Most tellingly, the differences were largely invisible to the experts themselves.

As the authors wrote: "In their description of metastasis, experts grouped the same symbols/events differently, they varied their ordering of events, and often suggested recurrent events absent in the outline that we showed them (the 'textbook' version of cancer metastasis). While some disagreements were minor, such as proposing that 'some unknown extra steps occur between these events', others were substantial." They went on to write: "It was clear after 28 interviews that despite similarities, experts think differently about metastasis."

So what to make of it?  If nothing else the paper supports the view that (1) finding an expert whose views on an uncertain area of science align with your client's pleadings is probably no more challenging than going through the buffet line at your local cafeteria and picking out what you want; and, (2) a given expert's opinion, made under such conditions of uncertain science, is almost certainly wrong.