Race for Rostov

Let's keep only one thread active per scenario. What do you think about a particular scenario. Was it too easy, too hard? Did you win or lose at first? Post descriptions of your brilliant victories and unfortunate defeats here.
Posts: 46
Joined: Fri Apr 27, 2012 3:20 pm

Race for Rostov

Unread postby RambOrc » Mon May 14, 2012 12:36 pm

A case where the AI could probably be further improved:

While progressing very fast in the South, I overextended my troops and so a single German armor unit could drive straight through a hole and capture hexes up to nearly the river, blocking further railway access for the supply center and thus putting 10+ of my units out of supply. I had to race back 3 armor units that could only reach hexes next to the German unit by using extended move (thus no attack).

On its next turn, the German armor had a clear way to proceed to the river, over the bridge and to the hex with the supply center. If it had done that, my tanks would have gone from one turn out of supply to two turns out of supply and unable to attack the German unit, I would have had to bring in more forces from the Center or North. If the AI had done so, I would have lost several turns. Instead, the German unit didn't move during its turn, meaning my tanks on the neighboring hexes were close enough to the supply center to get resupplied even with the railway line blocked. They took turns shooting the German unit to rubble, then raced back to the frontline. Thanks to this missed opportunity by the AI after a smart move (I didn't really leave the railway open, it was a nearly full strength infantry unit he completely smashed for its tank to get all the way through), I still achieved a BV, with the Germans having I think 3 infantry units left, way Northwest of Rostov.

Sadly I didn't save the scenario so can't post the replay.

User avatar
Tomislav Uzelac
2x2 Games
Posts: 1877
Joined: Mon Apr 04, 2011 11:24 pm
Location: Zagreb, Croatia

Re: Race for Rostov

Unread postby Tomislav Uzelac » Mon May 14, 2012 12:45 pm

Hm, we really do need the savegame in cases like this.

Tuning AI agression is problematic, you might make a change that will work in this case, but then be over-aggressive somewhere else. If you provide a savegame, we can build a base of savegames against which to test.

Thanks for the report though, ante may still comment upon reading it, but in principle I think we need the savegame.