What broke
The only user reported something on screen; the first response reached for reasons he might be mistaken — display scale, a driver, an update — none evidenced. He filmed it, and one frame in sixty showed exactly what he had described.
The rule it paid for
When the only user reports what he sees, ask what the instrument missed. Where an observation cannot be reproduced, the finding is not reproduced by this method, named with the method — never probably not there.