One of the problems in comparing airlines with the military, of course, is that in the former we are generally referring to two-pilot cockpits (or multi-crew cockpits), whereas a high proportion of military cockpits are single-pilot: a very different game.
Then there's the question of quality control and incentives for self-improvement/development. In an airline I once worked for (and with which Bergerie1 is familiar), check assessment was at that time based on the average (perceived) standard for the airline, which was defined as "Good". There was a chart of the different grades of performance, which was symmetrical top and bottom and based on a so-called percentile of the pilot force.
IIRC, the middle 80% of pilots would be graded "Good" (percentiles 11 to 90 inclusive). Above "Good", percentiles 91 to 99 inclusive graded "Very Good", and percentile 100 was "Outstanding". Below "Good", percentiles 2 to 10 inclusive graded "Satisfactory", and percentile 1 was graded "Unsatisfactory".
The only fail grade was "Unsatisfactory". In practice, however, "Satisfactory" was a euphemism for unsatisfactory, and in need of further training and/or check. The 90% of pilots above that grade were unlikely to be offered any valuable simulator time to brush up their skills. A perennial problem with the grading in practice was that trainers and checkers were inclined to award VG when Good would have been more appropriate. On some fleets more VGs were awarded than Goods, and many pilots came to expect the latter merely for a trouble-free performance. Pilots tended to justify this on the perceived grounds that their fleet was better than average in the company. Grade inflation therefore spread from one fleet to another as new types were introduced.
Of course all that was many years ago...