Slitherine

Kudos on the new AI opponents. Pretty slow, but for the first time I'm feeling challenged by the AI.

Thank you! We will let Miguel know.

Share some of your games here. I would love to see the results. I just lost on Typhoon as the Axis to the AI. Stopped me at 8 points.

Thanks! Our whole development team worked on that to make it as fast as it is, and let you interact with the map while the AI ponders its move.

Did making the ai smarter help them on their luck rolling dice? Before I could make a breakthrough in the prepared offensive stages of the game, but now I am lucky to not get cut off the first turn. Having a stack of armor in the clear is no advantage at all, as attacking infantry will score four or five hits while I may not get a single hit. Very discouraging.

I have to agree with the previous poster. Something seems odd with the odds, pun intended. Even though the odds are heavily in my favour I find myself losing more often than not. Will try to document this.

Having said that, you make great games and your support is top notch!

Really truly, no change to combat mechanics or bonuses: the AIs don't cheat in any way. It would have been easy to do (and avoid some of the justifiable flak we received that the early AI was too easy) - but we don't like it in other games and decided not to in our games. If that decision ever changes it will be disclosed.

Good play creates a situation where more random results will be beneficial, while adverse random results on the negative have less impact. The brain translates that into disproportionate luck; whereas it's actually disproportionate **consequence** of luck -- the sign of good play in games with random elements.

Many excellent backgammon AIs face similar accusations. Some have gone to extreme lengths to defend themselves (e.g. allowing players to roll and enter their own dice). Not something we have the resources to do (nor is it a feature that most players would use or value).

today I launched 5 games opeation tyhpoon, just to check the combat odds, and the battle with 3 stack of armor 13 points which 4 are elite against 7-8 pips of infantry and always (in these 5 tries) my tanks suffer 3-4 hits

I know it is possible but let's count the odds: infantry against armor in open terrain without defensive works score a hit with 20%. so I expect 1-2 hits at the most even 3 but not all 5 games 3-4 hits per game... and why first i didn't notice this?
it's not because the difficulty in fact I set a previous AI opponent.

so please check if the new relase bug somehow the soviets infantry roll or the bonus malus related to it.

Thanks in advance!

I haven't seen any evidence of 'bad dice' in almost a half dozen games vs. the AI:

Typhoon: 3, all wins, some easy.
At the Gates: 1, easy win. I thnk ObliqueStrats is correct about this one. It may benefit from a VP adjustment.
Moscow Campaign: 1 win, Moscow captured.

The Rokossoovsky AI is a vast improvement over the original attempts. Well done Shenandoah! Much, much better. It pushed and surprised me several times, although it is still prone to some errors in my opinion. In an attempt to rescue units out of supply, it will make high risk attacks that add to its supply problems. It will also commit suicidal attacks into overwhelming numbers, usually not to any purpose.

The sample is small, but I feel it also has difficulty weighting the importance of Moscow and the need to defend or recapture reinforcement spaces. In my one campaign vs the new AI, I got off to a slow start. By frost I'd managed to pressure Moscow. I held Tula. Throughout the late game, the AI continued to emphasize Voronezh. Rather than move to relieve the pressure on Moscow, the AI counter attacked west through Kursk and Orel. Moscow fell, but I would have won on points anyway due to VP accumulation around Moscow.

This is such a vast improvement, I'm very hopeful you will add this AI to Bulge, and launch your other games with it too.

Hey guys, just checking in to remind you that if you have problems or run into questionable situations, please use the "Feedback" button in game. We track and investigate all issues we receive. Everyone plays a bit differently, so seeing the game play out definitely helps. ( Plus I really enjoy watching you guys play, I learn a ton ) Thanks.

I can't say for certain if the outcomes are different of course. But it does seem like I'm getting a lot of extremely unlikely combat outcomes. I find myself saying, wow, I've never seen that combat result before.

Initially I was against suspecting that the hit roles against defending units were rigged or glitched somehow but after my last game, I'm on jumping on board.

17 infantry attacks, 3 full stacks, against a single, 2 pips remaining, utterly out of supply, non-elite tank, the only defender unit of Moscow and there all misses. Do the math, I just won the lottery with those odds; so I'm guessing it's a glitch and while this might be the most apparent example I've felt that something has been amiss since my first game. It doesn't help that the very same game contained a definite glitch just a few turns later in which the AI abandoned Moscow with my units currently inside of it, loosing itself the game instantly.

I am; however, appreciative of the AI's sense of fair play.

I hate to be repetitive since I have posted my feelings and opened a few cases through the feedback option in game, however I agree that 2.0 seem to favor the defender and I have decided to not play any scenario as an attacker...it is disappointing and will lower my score....

I've now played four campaigns vs. the new German AI:

1 draw, that could have been a win had I not bungled supply.

2 losses, Moscow captures. In both cases, the AI successfully formed a pocket on the first turn, trapping virtually every forward defender. One might have been a draw, the other a German win on points.

1 win at 7 VP (the most recent play-through). In this one I managed to extract several Soviet formations, enough to build a solid defense along river and wood lines. Despite the score, the AI had a chance. It bulled its way next to Moscow during frost, nearly breaking through.

At times, it seemed as if the German AI benefitted from unnaturally hot combat dice, particularly in the vicinity of Moscow. The battles were brutal and one-sided, conflagrations of massive damage to Soviet units. I'm no mathematician, but I've seen results that feel just as unlikely in multiplayer competition. I think it is the nature of the CIC combat system. Luck has the potential to decide the game.

The advanced German AI is now so well tuned, I imagine even the most expert of players will sometimes fall victim. Great job!

Hey guys. Thanks for all your comments. We're listening, and we're acting, so keep them coming.

In the mean time, if you've not liked our Facebook page you may not know that Miguel has posted a couple of great articles about the development of CiC's AI. A lot of you have questions and suggestions about our AI so hopefully this will give you a little more background info.

This first one is an overview of the changes in Drive on Moscow 2.0

http://www.shenandoah-studio.com/moscow ... rovements/

This one goes into detail of how it works.

http://www.shenandoah-studio.com/artifi ... alphabeta/

Cheers,

Pat

I just put three tanks full pips in open up against one infantry full pips in open and had a draw! Something's not right.

Played Battle of the Bulge yesterday and had the same feeling about the damage swings, especially compared to the predicted outcomes. It might be that the percieved problems with Drive on Moscow is rather a play balance issue. The improved AI makes a bad dice roll harder to recover from in Drive on Moscow, at least for me. The margins are so small that the outcome of the first round battles in Operation Typhoon decides the outcome of the whole scenario. Maybe the variance of the dice rolls is a bit too much.

Regards,
JT62

I'm afraid I also see hard-to-believe combat results with the new AIs. Let me say first that the new AIs are very good and I've been challenged by them, particularly Halder. But I'm frankly mystified by the fact that the Halder's planned offensives on turn 1 always produce the optimum result - KIA or retreat - while mine nearly always produce the lesser result in each case. Something is decidedly Not Right.

Last night I tried Typhoon against the new Russian AI. My turn 1 planned offensives produced a whopping six pips of damage and (obviously) no pocket. I still won easily, largely due to ill conceived Russian maneuvering on the flanks.

I felt my earlier post here might be unfair - memory is a funny thing & we tend to remember the poor results better than the good ones. So I decided to run some tests on opening planned offensives.

Halder's preferred opening offensives are 1) elite armor stack in Vladiminskoye, 2) infantry split into Andreapol & Olenino, & 3) armor / mech in Shostka. I started 10 games with Halder & he made these attacks every time. I tracked the results. In Vladiminskoye, Halder achieved 10 KIAs and got the armor exploit advantage. In Shostka, Halder achieved 8 KIAs with armor exploit, and 2 retreats by Sov defenders.

Then I switched sides and started 10 games vs Rokossovsky, using the same planned offensives. In Vladiminskoye, I got 6 KIAs and 4 retreats. In Shostka, I got 1 KIA, 7 retreats, and 2 failures to even dislodge the Soviet defenders.

I realize 10 tests doesn't prove anything, but this seems crazy to me. So, I stand by my earlier post. Something is wrong.