Isn't the most reliable the one that goes the longest without maintenance?
Not necessarily. It can be contextual, too, such as the frequency of malfunctions when fired clean and lubed for a hundred rounds. This is a more practical test of functional reliability in the context of civilian self defense than would be a thousand round torture test. To increase the statistical reliability of the results, multiple repetitions could be performed, always starting with a clean and lubed gun. If, during a thousand round torture test, however, the 1911 only started bobbling after 700 rounds, and the Glock kept going to the end, the difference would be meaningless apart from a situation like being dropped behind enemy lines in a jungle environment. It would have nearly no relevance for anyone looking for a concealed carry gun.