Arfa Daily wrote:

When a diode power bridge - discrete devices or four-legged package - fails,
it's almost invariably one diode in the 'negative' arm which goes short
circuit.

No it isn't.

Any thoughts on why this should be, rather than any one of the four
failing at random ?

Your observation is obviously flawed. All the diodes pass the same current. By
chance you've simply experienced a statistically anomalous percentage of such
failures.

Graahm