And people point to how PG had more hexes with which to work and that somehow this made things EASIER to program its AI?
It is not the pure number of tiles or hexes, it is about freedom to manouvre your units around.
That is one of the biggest problems in Shafer_5.
And not to mention the fact that PG combat was a good bit more complex than CiV in terms of how various types of units performed against one another. PGs 1994 AI seemed to be able to take THOSE issues into account a hell of a lot better than what we see in CiV.
People really seem to not see what Panzer General was: mainly a scenario-based game with a given number of units for the defender, low number of replacements and - most important! - typically a one-way-street-like flow of movements.
And, very important, you did have support fire. Why this has not been implemented in Shafer_5 is completely beyond me. I can only guess that the completely misdesigned archer-type units would have become even more obviously stronger than their successors, the fire-arm infantry.
You did start in clearly defined positions, you had clearly defined targets and a very restricted timespan to use.
So, the "grand strategy" for the AI was quite easy (in comparison) to program. And yet, it made many severe faults.
Tactically, it was mostly better, but nevertheless not much of an obstacle for the human.
The biggest problem was the limited time. Typically, for an overwhelming victory you only had ~12 to 15 turns time. Maybe 16 to 20 for a standard victory. A bit more for a minor victory. Beyond that, you would have lost.
So, creating an AI for such a game is not as complex as it is for a Civilization game.
Defenders of CiV say PG had well-defined objectives. And CiV does not? It would seem to me that if the AI is at war with me it would not be incredibly hard to teach it to regard my cities or resources in the way that PG regarded those victory point hexes.
First of all, defending typically is easier than attacking.
In the PG scenarios where the AI was to attack, it typically had overwhelmingly more units, very often even better units.
Second, PG took place on pre-made maps. Number, type and starting position of units were balanced against these maps. This is obviously not true for a Civilization game.
Third, due to more space, in general much higher movement rates and no eras going by, the AI could try to outflank you. In Shafer_5, in most cases the map holds just one obstacle after the other, minimizing the speed of units.
Cities in Shafer_5 are much closer located to each other than victory points were in Panzer General. Both, human as AI could really manouvre their units.
I will be the first to admit that I am not a programmer and have never tried to craft a tactical AI. But I do not see ANY reason (other than lack of skill and/or experience on the part of the programmers) why CiV should be so utterly and completely inferior to PG from a tactical standpoint.
In my last game I had innumerable opportunities to wipe out unscreened siege units as they wandered aimlessly about while purporting to "invade" me. Including one occasion where the AI decided to mount an invasion consisting of three artillery units and exactly nothing else. What exactly was that supposed to accomplish? Lots of hexes, fewer hexes, defined objectives or not - that is just plain stupid, ridiculous, and immersion-breaking.
Defending the CiV tactical AI is kind of like defending the US college football BCS system. I guess you can do it, and you will see some people try. But the arguments they use are never particularly logical and never completely make sense.
To this one I agree.
The deficencies of programming are on top of the design-inducted problems.
Which, measured against the advertising is such a shame that one misses the words to express it.
Yes, Shafer_5's AI is in many cases even weaker than PG's AI. After having tested 1.141 now in some games, I would even say, in most cases.
Nevertheless, even PG's AI was quite weak.
And I say it quite frankly: whoever was testing this AI should give a s**t about the NDA and tell the truth (in case the developers really didn't listen, as some try to indicate), or should remain silent and turn away in shame (as I think most testers were just too enthusiastic about being "awarded" by their tester status to really have a close look to what was going on) .