Slitherine

Hi, just finished my second win, this time as Allies. Took Germany down in Oct.'43!!! I think I gave AI minor advantage. I won as Axis, having 2
capitols by end of 1945. AND the last 14-15 mos of that one, was most boring, because of the AI's ineptness.

I know there's was a lot time and thought on makers of this upgrade obviously. The new manual is mind-blowing. I understand what you all set out to do. YOU accomplish your mission via the enhancements and changes. And for the most part the game functions well; WHICH is why I keep coming back to Slitherine and it's associations.

I don't want to sound to harsh, but because you did nothing with the AI's adjustments, ect. I find playing this boring. No challenge at all, because AI
is given so much unearned advantages, and just prone to unexpalinable actions. 2 examples. As Allies AI, as soon as USA enters war, AI sends transports to Normandy/Calle area. They arrive in winter months and sit there forever. Because I had time to transfer enough units to area and sit on coastal hexes, wait and here&there when one could land, just finished them off. There were at times, 10-12 transports sitting in Channel, from winter '42 'til game ended. Secondly, in one just finished, as Axis, AI was a year behind in taking countries,ie France May-ish '41, then went after Switzerland(I saw another post referring to this also) & Yugoslavia and in June-ish '42 USSR Dow on Germany, and the AI had hardly anything on that border!! Also the AXIS AI made no attempt at Atlantic Sub warfare, probally because nmy comvoys were guarded and knew it couldn't win. As ALLIES AI it made a very feable attempt at guarding or attacking subs. I just conscentrated on Soviet aide and left UK alone for most part.

I could go on, but I think point has been made. I will just make some observations about combat results now. The AI hasd a great habit of attacking out of dug-in positions and not suffing any negativity, EXCEPT if it receives a combat loss. AND here's my gripe, it suffers no entrenchment loss, which seems really unfair. I say, if it attacks and suffers losses, then entrenchment should be lost also, because the units ability to properly man that same entrenchment would be reduced. Also, the effectiveness of the AI's airforce is totally bias for it. No matter whether one plays Allies or Axis, the AI's air war can not be won, except to overrun it on the ground!!! And don't anyone suggest anything about research. I know how that helps, but one-on-one the AI's air of 30-40% more effective.

I know you all spent an super amount of time and effort, but by ignoring the AI's adjustment, for someone like me, it really doesn't thrill. I can only hope that sometime down line, you'll work on the AI's side. It still surrouds exposed subs, like before, expecting to able to isolate, I guess. I can only guess, because so much of gaming is MP, that a lot of this new content is aimed at MP.

In closing keep up the good work and I look forward to things. down the line.

Marshall Ney

Does not France, Belgium, etc. come back into play, when Allies retake??? I thought first time, was because I only had Paris. But this last time, I had just aout the whole country retaken, but nothing. Why not?? Also, in my previous post, I forgat to mention the AI as AXIS, made nor attempt to help in Africa. And as Allies, made no attempt to go for the under-belly of Europe, as Churchill argued and won.


Anyways, Cheers,

Marshall Ney

We're not going to alter the AI because the game is too complicated to get a working AI. Why aren't you playing against a real opponent? Then you get a true challenge.

This is true with most games, especially strategic ones. The AI can't give you a real threat. Usually it will follow a script and once you learn how the AI script is you can select the most effective strategy against that

program it like a chess computer, playing billions of games and put them all in a data base

I think there's few members on the forum who have been trying to improve the AI performance. You could ask if they want to share their findings.

H2H game is still much more fun though.

jkile wrote: I don't want to sound to harsh, but because you did nothing with the AI's adjustments, ect. I find playing this boring. No challenge at all, because AI
is given so much unearned advantages, and just prone to unexpalinable actions. 2 examples. As Allies AI, as soon as USA enters war, AI sends transports to Normandy/Calle area. They arrive in winter months and sit there forever. Because I had time to transfer enough units to area and sit on coastal hexes, wait and here&there when one could land, just finished them off. There were at times, 10-12 transports sitting in Channel, from winter '42 'til game ended. Secondly, in one just finished, as Axis, AI was a year behind in taking countries,ie France May-ish '41, then went after Switzerland(I saw another post referring to this also) & Yugoslavia and in June-ish '42 USSR Dow on Germany, and the AI had hardly anything on that border!! Also the AXIS AI made no attempt at Atlantic Sub warfare, probally because nmy comvoys were guarded and knew it couldn't win. As ALLIES AI it made a very feable attempt at guarding or attacking subs. I just conscentrated on Soviet aide and left UK alone for most part.

Marshall Ney

No game, so far, has provided an AI to match human play, they often provide a reasonable game as defender, but are rarely challenging in attack, One way to get round this, is to play the attacker, as the Axis, in the early war period and the Allies in the later war.

The transports sitting in the Channel, turn after turn, looks odd and very unrealistic. Maybe there should be some penalty that the ground unit being carried should suffer step losses, starting after the first one or two turns (which are loss free), until they are either eliminated, or returned to a friendly port.

If that can't be done then I can live with it by the way you view the situation, looking at it as Grand Strategy and a high command situation map, the transports represent the threat of invasion. If you don't react the threat, then a landing may take place, but by placing units on the coast you can counter the threat. The threat of invasion starts in 1942 and continues thereafter, to remove the threat you would have to commit enough air and naval units to destroy the transports, in strategic terms you would have won control of the Channel. Meanwhile your enemy will be trying to keep the transports on station to maintain the threat and destroy your coastal ground units so that a landing can take place.

The same strategic view can apply to naval units, when I place a unit at sea, I am designating an area of patrol. It doesn't mean the ships are sailing there for months without end, but that the fleet is maintaining patrols in that area.

Stauffenberg wrote:the game is too complicated to get a working AI... The AI can't give you a real threat. Usually it will follow a script and once you learn how the AI script is you can select the most effective strategy against that

Comments like this are unnecessary. It is possible to program/script a challenging AI for WWII grand strategy games similar to this one, it just takes a lot of time and effort. It would be more helpful to just explain why it is difficult to improve/enhance the AI with the current game engine and share any development thoughts for a future version? But don't say it can't be done and dismiss computer wargame customers' reasonable expectations for a decent computer opponent.

Rasputitsa wrote: No game, so far, has provided an AI to match human play, they often provide a reasonable game as defender, but are rarely challenging in attack

It is unreasonable to expect a programmed/scripted AI to "match" human play, but enough variability can be introduced to provide a reasonable game, for either Axis or Allied play. Again, it just takes a lot of time and effort and perhaps several game versions to get there, but it is not impossible.

pzgndr wrote:

Stauffenberg wrote:the game is too complicated to get a working AI... The AI can't give you a real threat. Usually it will follow a script and once you learn how the AI script is you can select the most effective strategy against that

Comments like this are unnecessary. It is possible to program/script a challenging AI for WWII grand strategy games similar to this one, it just takes a lot of time and effort. It would be more helpful to just explain why it is difficult to improve/enhance the AI with the current game engine and share any development thoughts for a future version? But don't say it can't be done and dismiss computer wargame customers' reasonable expectations for a decent computer opponent.

First of all. We're NOT computer wargames developers who sell any games. We're just modding a game following our ideas about how to improve CeaW. We in the development team prefer to play against other players and not the AI. Therefore PBEM was our focus. We have no obligation to do anything with the AI code at all. The AI code in GS is the same as it was in vanilla CeaW.

The AI code in CeAW is not very extensive and if you want to make it somewhat challenging you need to code many thousands lines and even that won't be enough. We're using our leisure time to mod what we can and I don't think any of us have the time to invest into creating an AI that can maybe only give novice GS players a challenge.

I've yet to see any computer wargame giving players a decent challenge. In some games the AI cheat, but good players can learn how the AI works and get the upper hand anyway. Do you have examples of strategic wargames with a challenging AI?

GS v2.0 is particularly challenging because there are so many things the AI would have to consider, like rules we've added to the game. I've seen the vanilla CeaW AI struggling to properly attack e. g. in Belgium and France to get a breakthrough. The AI doesn't know how to finish off enemy units to use as base for the next attack. Instead the AI continues to attack the same hexes and end up with 2-4 steps left. Then you can swap the full strength rear units to the front and the AI must try again. If you use the rivers for the defense the AI can be stuck for a long time in Belgium / France.

So the AI doesn't even master the simple basics of the game, i. e. how to attack to get results. It would be even harder to make the AI understand the strategic aspect of the game. You need to analyze how to maximize the firepower on critical hexes so you can open up a line. E. g. I described how a human can break a double defense line if there are hexes in front on the defense line that are controlled by the defending side. It will be a nightmare trying to code a working code to do the same.

It's a bit easier to program a pretty decent AI for tactical games because you only need to analyze the map and move your pieces to capture enemy hexes. In strategic games this won't be enough. You need to have a plan to build up your strength so you can maybe attack 6-12 months into the future. You need to know when you can be on the offense and when you must be on the defense. You need to code so the Allies can attack something in 1941-1942 before they can confront the Axis on the continent. So you need to code a way for the Allies to probe the Axis controlled areas for weaknesses and attack there.

It's possible that talented AI coders can make a better job than the AI in the vanilla CeaW, but not even the best programmer can
give a good player a challenge without extensive cheating.

None of us in the GS development team have a lot of experience with AI coding and we can use our time better to improve the game for hotseat, PBEM. If someone wants to give it a shot then we're happy to invite the person to the development team, but I don't think you should hope for miracles.

Stauffenberg wrote: First of all. We're NOT computer wargames developers who sell any games. We're just modding a game following our ideas about how to improve CeaW... I've yet to see any computer wargame giving players a decent challenge. In some games the AI cheat, but good players can learn how the AI works and get the upper hand anyway. Do you have examples of strategic wargames with a challenging AI?

I understand the pbem objectives of the GS 2.0 mod team, and that the AI was not the team's interest. Fine. But regarding AI improvements and enhancements, Slitherine should reconsider what it could do, either as modest upgrades in a patch or as more extensive game engine redesigns for future versions.

I disagree with the assertion that a decent challenge from a computer opponent is out of reach. It's not; it's been done. It just takes a lot of time and effort and patience on the part of developers and modders. See http://www.wargamer.com/article/2526/Ar ... hird-Reich

Btw, it's been suggested that some human players cheat too (hard to believe!) and that it's possible to learn how they play and possibly get the upper hand.

pzgndr wrote:

Rasputitsa wrote: No game, so far, has provided an AI to match human play, they often provide a reasonable game as defender, but are rarely challenging in attack

It is unreasonable to expect a programmed/scripted AI to "match" human play, but enough variability can be introduced to provide a reasonable game, for either Axis or Allied play. Again, it just takes a lot of time and effort and perhaps several game versions to get there, but it is not impossible.

I never said I expected AI to perform as well as a human, I was just stating the obvious, with the comment that the AI can often give a reasonable game as a defender. I have played countless games against AI over the years and enjoyed most of them, including CEaW-GS.

I terms of realism, at many times during the war, all nations suffered bad strategic decisions, or had such decisions forced on them by their unpreparedness. Compared with some of the events which actually happened, the AI can reproduce a good historic feel.

I now see what the makers and programers purpose is. And it's been right below the surface for me for a while. MP is the only thing on their minds these days. It's understandable. It's what has evolved thru the years and it's a generational thing. I started in gaming with die-cutter board ones and thought early computer adoptations were the coolest, with simple AIs. And I must now come to realise:

Now because the fab-fad is MP, the AI is totally ignored. So be it. I just thought some adjustments, would apply to the AI.

I'll be more understandable in future, but nonetheless, still hopeful, this part won't be totally flushed. Then again, dreaming.

It just sounds like you guys should maybe be more forth coming in this aspect, and/or maybe I was expecting to much. It's just real disappointing to see The AI do ill-logical and unhistorical things.

Anyways, I guess us old timers will just have to suck it up. Have a good day.

Marshall Ney