Sorry i cannot resist this chance to feed back to the Game Developer directly, so i decide to post again in this thread.
It is about AI.
My altitude is that new AI change is not 100% good or 100% bad. It is a experiment which will be followed by more experiments.
I will analyse it from a programmer's perspective.
Quite a few newly reported issues are actually caused by an AI change that is target to another old issue.
1. Pathfinding is changed.
Remedy for old issue: Previously if a rally is outside a wall block, units inside will attack the wall blocking them but not thinking of turning around and walking through the gap in the opposite wall.
Now AI is smarter that units will take a longer route to walk to rally instead of attacking wall. Also archers are smarter that they will walk a short distance to attack a building outside of wall, instead of stopping too early to start attacking wall.
Problem caused by this change: though the AI change enhances something, it fails something else. Now the HC units and infantry units are so "willing" to find a walkable route that they often unpredictably choose a long route and being attacked by many defending towers along the route.
It becomes even worse because of another change that you made, which is to teach units be more sensitive to nearby defender units. I will talk about that point later.
My suggestion: There need to be a balance when calculation route cost. If the current method is base on lowest "cost" optimisation mechanism. Then the cost of "wall block" and the cost of "unblocked space to walk" need to be tweaked. Now the units are too willing to "walk", previously they are too willing to "attack blocking wall", maybe compromise a bit?
I also suggest this judgement should be different for different unit types. Archer should be more willing to "walk" but HC and infantry should be more willing to "attack blocking wall". It is complained by many players that their HC are attacked during unexpected hanging around the walls, the reason is HC too easily decide finding a walking route instead of crushing the wall blocking him.
2. Attacking wall preference is changed so nearby units will try focusing on attacking the same wall block instead of inefficiently attacking the nearest wall
Remedy for old issue: The effect is very obvious. I can observe very often that group of archers or infantries are now working together on same wall block.
Problem caused by this change: Though it works the purpose but the AI change causes another unexpected and more serious problem. Now the units are so obsessed to attack wall block that they often ignore a open wall gap nearby them. The priority should be: "walk through near by wall gap" > "attack a wall my nearby fellow guys are attacking" > "attack my nearest building block wall", but now obviously their first choice is to "attacking a wall block" together. Gosh, sometimes i really want to roar because of their stupid behaviou: guys, your task is to attacking key buildings, not attaking a wall, even you guys sing song together and attack the wall very happily, but please dont be obssessed by a god damn wall......
My suggestion: Prioritise the correct sequence, also please increase the frequency that a unit checks his own surrounding environment so he can find out there is open wall gap nearby as early as possible.
3. The attacking preference is changed that units are much more sensitive to the nearby denfenders
Remedy for old issue: The effect is also very obvious (though i do not know if this really wantted by all people). Now the archer or HC will head toward and chase moving nearby defender. It is not like before that they were obssessd by the current task and ignored defender killing their comrades nearby.
Problem caused by this change: In my opinion this AI change is most debatable. Is it really wanted by majority of people? I really doubt it and it does not suit me obviously. My units are so easily distracted that they forget their main task and blindly chase defenders in all directions. It makes it worse that now the defender are buffed. There are so many of them and they are not easy to be killed as before. All these plus together causes a chain reaction that greatly impact raiding troops efficiency. Hi guys, i deploy you at that location and at that time for a reason, why chase a defender wandering around? who cares that defender??
My suggestion: For me, i will rather this change does not happen at all. Or at least, raiding units will not chase the defender so aggressively. They should be obtuse until the defenders move into a very close range.
Summary:
As a developer, i appreciate the boldness your developers took to solve problems in old AI. I wont complain even it brings equally many new problems if not more. Programming is like that, code->performance benchmark->change code and so on and on. At least the change reflects the good will.
It is not because we players are smarter so we see the problems but your developers did not. It is only because we check them out when we have the new AI but you needed to imagine the behaviour in mind when planned and developed the new AI.
If we were assigned the same task to improve old AI, I am sure very likely we did even worse or equal bad. It is natural and acceptable that improve through experiments.
So all in all, I hope you developers can move on to continue tweaking the AI. My suggestions above are not perfect. They will trigger some other unexpected problems which I just can not see now, as you previously did not predict the defects in your latest AI development activity. But let me wish you good and please keep on enhancing. Nothing is perfect, but we all wish it better as you do.