According to the surveys I’ve read, the average distance for police sniper shots in the US is just under 60 yards, and median is closer to 35 yards. At 60 yards, even a ratty AK47 with surplus ball will produce a group not much greater than two inches. A two inch group means dispersion of only one inch from the point of aim at fifty yards. So why is everyone obsessing about sub-MOA accuracy in sniper rifles?
The reasons are several. First, average and median numbers are no comfort for those snipers who get the less common but not altogether rare longer-range cases. Second, police snipers are working to much tighter standards than so military snipers. It’s not enough to score a hit, it’s usually necessary to hit the brainstem and completely shut down any voluntary or spasmatic movement, such as pulling the trigger on a hostage. The instant turn-off zones on a human are tiny, about five inches wide by two high from the front and roughly a two inch square from the side. Third, all deviations from the desired point of impact may add up — the errors in range estimation, the movement of the target, the imprecision of the sniper’s aim and the dispersion inherent in the firearm can all contribute to the point of impact being off by far more than the inch or so induced by the 4MOA limitation alone. This is also why higher-velocity rounds are preferred for most police sniping, to reduce the impact of any errors in range estimation as well as to limit over-penetration. A one-MOA rifle adds at most a quarter inch of error to the shot at fifty yards. Since the desired result is not an abstract group but a very precise match to the moving, bobbing target, that quarter inch minimizes the chance of the cumulative error growing beyond acceptable value. Other ways to reduce that error include getting closer, using a steady support, waiting for the perpetrator to become stationary and having multiple snipers fire on cue.