What happened?

[GBS]

We seemed to be on the verge of a patch then.......nothing. Any news??

[MarkShot]

There is no great mystery here. Dave (and occassionally other PG staff) have been hard at work resolving problems and tweaking the AI to make it play a more aggressive game on the attack. In some cases the problems being addressed had been previously reported by the customers and in other cases they arose due to further testing by the developers, betas, and auto-testing procedures (the AI plays itself).

Additionally, a few of the scenarios that have been our primary test cases have had some changes made to improve play balance.

By now, everyone here probably knows that Dave is a perfectionist. As such things always seem to take a tad longer than planned. But in the end, the result is always beyond description.

So, far I have played quite a few games of one particular scenario as each new iteration of AI tweaks is evaluated. I have been playing a fighting withdrawal role.

When I first started testing, the OPFOR AI tended to linger on objectives it had already acquired rather than advancing hard and fast.

After some work from Dave, the OPFOR AI began to rapidly fill the force vacuum of terrain which I had created as I pulled back.

Yesterday, in testing the latest round of tweaks, I played my first iteration where I found the AI actively hitting hard up against my delaying forces before I could pull them back. In fact, the AI came within just a few hours last night of bagging my arty firebase. I just managed to extract them in the nick of time (playing MAX order delays). Now, although, I usually portray myself as quite a humble player. I think I am very good at fighting retreats (terrain analysis, road blocks, night time withdrawals, dead space defense, bridge demolition) and, of course, I have played this one scenario numerous times. So, that was some very impressive improvements on the AI's part.

I am sure Dave to could explain it better in terms of C++ routine calls, replans, and force allocations. However, the bottom line is that the AI is without a doubt becomming more vicious and one can clearly feel it.

Psst ... so, keep the chatter to a whisper, because somewhere on the other side of that ridge is a panzer KG checking the action on their weapons and topping off their fuel. I think the dust be kicked up in the distance is more than just the action of a hot sun and a windy day. So, stop talking and get digging while there is still time!

[MarkShot]

Just to illustrate my point, this screen was from last night's test run.

You will notice to the NE German armor assaulting a very lightly defended position. And, of course, they might only be the tip of the spear. (In the past, the AI would have taken much longer to arrive here and would have arrived doing the road column stumble and bumble; but not this time ... they got here fast and they are itching for a stand up fight.)

My arty is just beginning to vacate the fire base. Of course, I had them set behind a ridge to prevent observation. But that is only concealment and as you can see, there was no real security detail protecting the guns. In the meantime, except for a handful of grunts there is only about 4km of highway between the German tanks and my precious guns. And that is nothing when they crank up those diesels and roll ... in the meantime, I have the full complement of order delays to get my guns to limber and pull back. In fact, my order delays are quite high, since I have over loaded my senior HQ with excessive micro-managing. In past, run throughs despite the load, I have managed to stay inside the German's decision cycle, but as you can see here, the Germans (OPFOR AI) have almost flipped the hold which I had on them and are attempting to pin my shoulders to the mat.

It was all very close. (If I lost my arty, the rest of my defensive arrangements had nothing with any bang to back them up.)

[Arjuna]

GBS,
 
Yes sorry for the delay. I was not really happy with the strategic AI's lack of aggression . I am now though. We are just finishing off a few other bugs discovered in our aggression testing.

[Panther Paul]

Is that "Agression" testing, or "Regression" testing [:D]

[MarkShot]

Paul,

Well considering that I keep playing the same scenario doing a fighting withdrawal, I guess it is both.

[MarkShot]

One thing I must say about playing the same battle over and over again (and using the same initial plan each time) ... you would think that it would get pretty dull ... but actually, it has been a lot more interesting than I expected. Why? One simple reason - the bridges. There are so many bridges and you can never be 100% sure about whether you'll blow a given bridge or not. So, the presence or absence of a bridge compared to a prior playing effectively makes it a whole new ball game each time out. I had never really thought about how bridge dynamics effectively becomes the "great randomizer" of scenario play.