ACR:
What is "ping ponging"?
Ping-pong is an exploit of the fact that the AI only sees the present game board and makes all its evaluations against that, i.e. it neither remembers nor learns anything.
For example, if an AI unit is going to somewhere (and there's no war) you can block its movement by 3-4 units depending on the terrain. The idea is that you leave the AI a clear path around your units but so that the AI needs at least 2 moves to get there. The AI moves towards the opening. You close it and open another path on the other side. Now the AI evaluates the situation again and moves back towards the new opening. You close it and open the original. You can keep this up ad infinitum. Hence the name ping-pong. With it you can effectively block AI settlers wandering through your territory to the city spots you have reserved for later.
A very cruel form of ping-ponging is to block and isthmus with a maginot line except for one open square. If there are good spots in "your territory" the AI will send settlers towards that opening. At the last moment you close the line and all the AI settlers start wandering back. Once they have goon all the way back home you open the path again and watch the settlers turn around and start their long walk again.
Ping-ponging is an exploit in all its forms because the AI doesn't learn anything. A human would quickly notice that he isn't going to get through and would then try something else. The AI will go on forever which effectively means that he has wasted a unit that could have been useful somewhere else.
The ping-pong principle can be used in wars, too. The AI is programmed to prefer lone weak targets. So, just keep switching a defending stack between two weak units and the AI will aimlessly wander between them. Well, until it has gathered enough strength to attack your stack but then it's usually too late...
Now, diversion is fine. It's a good acceptable tactic. But it ought to be used for the original ping only and not the pong

, or you are just turning the rigid AI programming logic against itself. IOW it's not fun because you know what the outcome will be.
A good defining question between tactics and exploits is whether a human could fall for it.