More background images.

There was some very interesting work done by @real_tabasco_sauce in this area here. I believe there were some cross-platform compatibility issues (IIRC related to mathematical calculations being handled differently on Linux, Windows, and macOS), which, in my view, would be worth revisiting.

Some ideas / thoughts: Ranged soldiers don't always attack the closest enemy unit, but instead semi-randomly any enemy unit in range, where the chance depends on the distance E.g. if two enemy units are almost equally far away the chance the the slightly closer enemy unit is attacked is 51%, the other one is attacked with a 49% chance. If one enemy units is far closer the chances would be more like 80% 20%. Overall that means that ranged soldiers spread their attacks more instead of many attacking the same units, preventing massive overkill and greatly reducing the effectiveness of ranged units During small scale skirmishes, e.g. in the early game, you can manually focus fire to intentionally kill a specific enemy unit Maybe the probability which enemy unit is attack depends on where you attack moved. Instead of prioritizing the closest unit to itself it prioritizes the closest unit to the target location. So you could manually move your archers into range of the opponent's Javileers and then attack move next to them so your archers attack them instead of your opponent's melee units     Melee soldiers behave similar. They don't always charge towards the closest enemy soldier but spread out based on probability   Random attack delays. SC2 does this. Each unit has a small random attack delay so that two units rarely attack at the same time. That's e.g. why Siege Tanks virtually never shot at an already dead unit. This would also cause units to slightly spread out their attacks as they attack at slightly different times      

I agree that automate repetitive tasks is an optimisation in production chains.... for industries, however I don't think this apply to a game, if the game is over automated will lose fun and will reduce player interaction with it. Snipe already exist in the game, is more important the game is fun than take care of mice. In the game this is a kind of "gadget" that could be very good technically, it shows very good skills in programming, however it don't actually improve UX. Instead of that, making the game performance better will. In the bicycle example I mentioned, improving "how you 'pedal' is more important than upgrading a fan to cool your face, while you keep doing a effort in a not-correct way." Think about it.

This selection box targeting was nicknamed autosnipe