The argument against LNT is simple and conclusive. LNT cannot replicate both the increased cancer incidence we clearly see when a large dose is received over a very short period and the lack of detectable increase when the same or much larger doses are acquired over far longer periods.1 Any model that cannot do that must be rejected.

This means we must have a replacement. That replacement harm model must be well defined. Unless you can program your harm model, you do not have a model. No qualitative handwaving about unspecified thresholds and hormesis need apply.2 The replacement must allow one to compute the harm (eg Lost Life Expectancy) associated with a given plume and population distribution. LNT can do that. The replacement must as well.

At the same time, the replacement must recognize our ability to repair radiation damage, and the fact that repair takes time. LNT cannot do this. Einstein once said ``as simple as possible, but no simpler". The Sigmoid No Threshold Model (SNT) is about as simple as you can get and still replicate both the clear harm inflicted by a dose rate spike and the lack of detectable harm from the same dose spread over a long period. //

1 In some cases, LNT is off by more than a factor of 100.
2 Hormesis holds that small, properly timed doses of radiation can be beneficial, in much the same way as vaccines.