I'll try.

Take a sheet of semi-log graph paper. It has a logarithmic scale vertically and a linear scale horizontally.

Let the horizontal scale be magnitude in tenths. Let the vertical scale be number of quakes.

Take the catalog of quakes and count the number of quakes which have a given magnitude. Plot a point representing the number on the graph.

Do that for each magnitude; 4.1, 4.2, 4.3, etc.

You end up with a graph which will be a downward-sloping line because there are more small quakes than large ones. a LOT more, which is why we use a log scale for numbers.

As long as that line is straight you can be pretty sure the catalog is complete but when the slope changes you've reached the limits of completeness.

In the databases I've been using, that happens around mag 4.5 or so. So I can only get good estimates of probability for quakes that are 4.5 or larger.

Roger