We need to know whether useful methods must be able to tell the difference between barely noticeable earthquakes and ones that cause great damage, not whether useful methods must be reliable. We already know that the new method has reliably predicted several earthquakes.

The new method’s predictions can’t always differentiate between earthquakes that are barely noticeable and ones that cause great damage. If a method that can’t always differentiate between these things is unlikely to be useful, then the new method is unlikely to be useful.

We know that the new method has reliably predicted at least several earthquakes, but (C) fails to establish that a method that can’t always tell the difference between non-severe and severe earthquakes is unlikely to be useful.

The geologist’s argument only addresses the new method; the predictions of any other methods are irrelevant. Instead, we need to prove that the new method is indeed unlikely to be useful.

The new method’s prediction range can be the difference between a barely noticeable earthquake and one that causes considerable damage. We don’t know that it never distinguishes between the two. (E) also seems to point to a much greater range, from “imperceptible” to “catastrophic.”