This is the smallest example we show people first. A serial EEPROM on one of our test boards uses address pins so several devices can share a bus. Each address pin has two zero-ohm links: one up to the 3.3 V rail, one down to ground. You fit one of them — not both.
What was actually on the board
On two of the address pins, both links were fitted. That creates a series path from the 3.3 V rail, through the address node, to ground — a short. The BOM gave all nine zero-ohm links the same value and package, with an empty do-not-mount column, and the only parts listed as deliberately unpopulated were four capacitors.
The designer’s note puts the bus address at 0x50, which is both address bits low — so the ground links were the intended ones, and the links to the rail should never have been fitted.
Two things go wrong at once
The rail is shorted to ground in two places, so the board will not power on. And even without the short, the address bits are being driven high and low at the same time, so the EEPROM has no defined address. One mistake, two failures.
Why the rule stayed silent
A deterministic short check looks for a single net that touches both a power pin and a ground pin. No such net exists here — the short runs through an intermediate address node, so the check returns nothing. That is a blind spot in the rule, not a bug in it. An empty result and a blind spot look identical from the outside.
Probe traced the copper from the 3.3 V rail through both fitted links to ground, read the empty do-not-mount column, and logged two findings: a blocker on the board, and a gap in our own rule set.