Yea but how is the AI?

[mutterfudder]

Most of the threads about test play is with another human player.
Has anyone playtested the AI?
I will be playing solo for some time and would like a decent AI.
Could someone start a thread with the AI and describe the play for play account?

Thanks

[Gyrene]

I prefer solo play myself too. Not sure how many testers played against the computer. I know Kid mentioned he played so many turns but didn't seem to go too deep into the entire campaign.

Just how intelligent & aggressive is the Computer player? Hopefully more so than in UV.

Does the computer repeatedly launch attacks against invading ground forces, or will they just continue to mindlessly go into bombardment mode?

How aggressive is the computer towards taking new bases? In UV, besides from base point value, there were pre-determined primary and secondary objectives. How is defined in WITP?

Any input from you guys would be much appreciated!

Thanks!

[tiredoftryingnames]

The AI will never replace a human but does it in any game?

It'll give you a good game from what I've seen but keep in mind from restarting due to code changes I have never seen it play out a long scenario. It seems to do well in the smaller ones with the limited objectives like in UV. Since the last time we've had to restart our games, I'm up to 8/42 against the japanese AI in the long campaign game. It's very aggressive out of the gate conquering territory and moving it's squadrons forward to the new bases.

I've finally checked the AI at Rangoon and I'm building up to push back. I was getting hammered from the air but hurricanes and spitfires have started to allow me to unload ships at Rangoon without getting the whole TF decimated from Tavoy.

The south pacific is starting to pick up as Rabaul airstrikes are hammering away at Port Moresby and in a surprise move his CVs after disappearing after PH showed up supporting his last few invasions in the NEI. The AI has finally taken all the NEI and I only hold out at Guadacanal and PM. I sent out some CVs to hit his transport forces moving around new guinea and from a sighting report saw that our CV TFs just missed each other. A few hexes closer and they probably would have reacted and launched strikes and it would have been bloody in the South Pacific. So now I'm trying to track him but his CVs have moved north away from my coast watchers and PBYs. Now I have to try and figure out where he's headed next.

The AI took Wake early on but I've retaken it and started building up the defenses there in case he comes back. I've finally started getting enough base forces with aviation support that I'm now building up the support at forward bases so I can start moving the air units from stateside to the action. Currently most bases can support around 30 aircraft. Once I get my forces in place i can drastically increase my air power locally and start my offensives. Which I can't wait to do to see how it handles the defensive. Of course the AI isn't done with offensive actions. The AI recently tried a push in china that failed. My pushes fail so it's developed into a stalemate there.

So it gives an interesting game. From restarting several times I know it doesn't always follow the same path and it's more fluid with it's CVs early on in which operations it supports so you never know where they might show up. On top of that you can do the varied starts to really get a different game. It will attack, not just do bombardment attacks. If it fails it will move forces up to try and get superiority. The battle for Rangoon started with 3 enemy units attacking me. By the time the 3 month running battle in around rangoon was done i counted 11 enemy units being reported in the action reports. he's now pushed back from rangoon and has probably been reinforced.

[tsimmonds]

The AI took Wake early on but I've retaken it

That's the only thing in your report that really sounds off. A decent human player, having taken Wake, would not let you re-take it by 8/42. Maybe the AI isn't so good at evaluating what size garrison is appropriate for a base not worth much in VPs but extremely valuable (and vulnerable) nevertheless.....

[tiredoftryingnames]

I don't know he was pretty strong there even though yes a human might have put more there. But it's a trade off. You either defend heavily pulling some forces from your invasion TFs or you let them take more territory. The AI was rolling up other areas when I struck. I had to use two army infantry divisions supported by two Artillery battalions after bombarding him (which damaged several DDs from his coastal guns he put there) and several turns of airstrikes from 3 CVs. Using 2 divisions meant I had to delay reinforcing the South Pacific so I had to gamble too with my force dispositions.

[Damien Thorn]

I would like to know if it continues to send transports to their death trying to supply islands that are totally under enemy air control zones. In UV you can sink hundreds of shipsby letting the AI take a base near where you have air superiority.

I hope the WitP AI will evaluate the cost of letting isolated troops die vs the cost of loosing ships trying to resupply them.

[tsimmonds]

Using 2 divisions meant I had to delay reinforcing the South Pacific so I had to gamble too with my force dispositions.

2 Divisions would be hard to stop. But this might not have been so much of a gamble vs the AI as it would be in PBEM, where a capable opponent would realize this sort of strength used in the central Pacific would mean relative weakness and opportunity elsewhere....or perhaps in PBEM you would find yourself needing to use those 2 divisions in a more reactive and defensive manner[;)]

[Gyrene]

Good to Go! This was just the type of info i was anticipating! I'm pleased (I'm sure others are too) the AI does seems to put up an interesting game.

If you wouldn't mind, could you please post an ARR or two vs. the computer as well?

Thanks Much!

[Mr.Frag]

That's the only thing in your report that really sounds off. A decent human player, having taken Wake, would not let you re-take it by 8/42. Maybe the AI isn't so good at evaluating what size garrison is appropriate for a base not worth much in VPs but extremely valuable (and vulnerable) nevertheless.....

Thats not true at all.

Wake can *always* be retaken by an Allied player at any time. The point for Japan to take it at the start is to prevent it being used to act as a ferry point for the initial Allied aircraft coming to help PI. Once PI is gone down in flames, Wake has little value to the Japanese except perhaps to act as a fueling spot for subs, but by then most shipping has moved far south and it is not even of value there.

[tsimmonds]

Wake has little value to the Japanese

Well, you oughta know. I was just going by my experience in the old paper and cardboard days. Wake wasn't so much worth anything to the Japanese as it was important to keep away from the US, who could use it as a springboard to bigger and better things. Perhaps the new game mechanics preclude this possibility....

[Mr.Frag]

Well, you oughta know. I was just going by my experience in the old paper and cardboard days. Wake wasn't so much worth anything to the Japanese as it was important to keep away from the US, who could use it as a springboard to bigger and better things. Perhaps the new game mechanics preclude this possibility....

Since Japanese troop vp are half USA, it is actually better to let the USA retake it then hit it again. remember the 2:1 rule on land troops.

[Nikademus]

The AI is phenominal at supply management. Just start an AI game, wait a few turns and then load it' side and look at all the TF's. Mogami would be proud [;)]


Wake is hardly useless to Japan. It constitutes a vital part of Japan's early warning net

[Damien Thorn]

ORIGINAL: Nikademus

The AI is phenominal at supply management. Just start an AI game, wait a few turns and then load it' side and look at all the TF's. Mogami would be proud [;)]

Yes, but will it send those transports into enemy controlled air zones and certain death? UV's AI was notorious for doing this.

[Nikademus]

new code was introduced to make the AI (and the tacAI for the human player) more aware of enemy air zones of control.

[tsimmonds]

Since Japanese troop vp are half USA, it is actually better to let the USA retake it then hit it again. remember the 2:1 rule on land troops.

VPs...kinda gamey, IMHO. Got tons of respect for you sir, but I like to take something once, and either fight to keep it or let it go; to lose it in order to retake it because of the VPs for killing the boys...[:-]

[Mr.Frag]

VPs...kinda gamey, IMHO. Got tons of respect for you sir, but I like to take something once, and either fight to keep it or let it go; to lose it in order to retake it because of the VPs for killing the boys...

Why not let the USA donate supplys for you and save yours for elsewhere? [;)]

Supplying Wake as Japan diverts shipping and supplys from places that really need it, such as the front line.

[tsimmonds]

Supplying Wake as Japan diverts shipping and supplys from places that really need it, such as the front line.

Wake always seemed to me like an important part of my front line. Again, this from the paper and cardboard days, maybe electrons have rendered Wake meaningless.....and maybe not[;)]

[The Gnome]

I would like to know if it continues to send transports to their death trying to supply islands that are totally under enemy air control zones. In UV you can sink hundreds of shipsby letting the AI take a base near where you have air superiority.

I hope the WitP AI will evaluate the cost of letting isolated troops die vs the cost of loosing ships trying to resupply them.

Amen to that! I had sunk so much shipping at Shortland, I was surprised anything new could fit into the harbor. In addition I hope they fixed the corrilary problem of sending even the most slightly damaged ships into ports under enemy air supremacy.

I could damage a BB trying to bombard Lunga, scratch him up as he was outboung and he would take refuge in Shortland. A couple of turns later he was sunk. This continued as naseum to the entire IJN BB roster.

[ZOOMIE1980]

ORIGINAL: tiredoftryingnames

The AI will never replace a human but does it in any game?

It'll give you a good game from what I've seen but keep in mind from restarting due to code changes I have never seen it play out a long scenario. It seems to do well in the smaller ones with the limited objectives like in UV. Since the last time we've had to restart our games, I'm up to 8/42 against the japanese AI in the long campaign game. It's very aggressive out of the gate conquering territory and moving it's squadrons forward to the new bases.

I've finally checked the AI at Rangoon and I'm building up to push back. I was getting hammered from the air but hurricanes and spitfires have started to allow me to unload ships at Rangoon without getting the whole TF decimated from Tavoy.

The south pacific is starting to pick up as Rabaul airstrikes are hammering away at Port Moresby and in a surprise move his CVs after disappearing after PH showed up supporting his last few invasions in the NEI. The AI has finally taken all the NEI and I only hold out at Guadacanal and PM. I sent out some CVs to hit his transport forces moving around new guinea and from a sighting report saw that our CV TFs just missed each other. A few hexes closer and they probably would have reacted and launched strikes and it would have been bloody in the South Pacific. So now I'm trying to track him but his CVs have moved north away from my coast watchers and PBYs. Now I have to try and figure out where he's headed next.

The AI took Wake early on but I've retaken it and started building up the defenses there in case he comes back. I've finally started getting enough base forces with aviation support that I'm now building up the support at forward bases so I can start moving the air units from stateside to the action. Currently most bases can support around 30 aircraft. Once I get my forces in place i can drastically increase my air power locally and start my offensives. Which I can't wait to do to see how it handles the defensive. Of course the AI isn't done with offensive actions. The AI recently tried a push in china that failed. My pushes fail so it's developed into a stalemate there.

So it gives an interesting game. From restarting several times I know it doesn't always follow the same path and it's more fluid with it's CVs early on in which operations it supports so you never know where they might show up. On top of that you can do the varied starts to really get a different game. It will attack, not just do bombardment attacks. If it fails it will move forces up to try and get superiority. The battle for Rangoon started with 3 enemy units attacking me. By the time the 3 month running battle in around rangoon was done i counted 11 enemy units being reported in the action reports. he's now pushed back from rangoon and has probably been reinforced.

The big issue in my mind is not with what the AI does to take and/or hold, what base. It (in this case the Japanese AI) is obviously tighly scripted, at least during the Japanese conquest phase. The issue I have, is that, lets say on MAr 15 the AI lands with the X, Y, and Z divisions, supporte by a Bombardment and Carrier TF, on Java to finish taking the Dutch East Indies. If I play the scenario five times will the AI ALWAYS land divisions X, Y, and Z supported by a BB and CV TF on Mar 15??? If so, and I believe it is, then that's lousy. No two ways about it.

From what I'm reading the actual combat AI seems to be very good. The idea, that Frag stated, that Coastal batteries will work in conjuction with minefields to take out the invaders MSW's is pretty good stuff! But it's the Big Picture stuff, the Strategic side of the AI that I'm afraid is going to be disappointing, at least after the first run or two through a scenario. It looks like I'm going to know exactly where, when and how the AI is going to attack me after a two runs through any scenario, right down to the exact same composition of ship types in the TF's..... I would have hoped they would have at least had a LITTLE randomness in the scripting to keep the human player guessing and least a little each time through a scenario....

[Grotius]

I know it doesn't always follow the same path and it's more fluid with it's CVs early on in which operations it supports so you never know where they might show up. On top of that you can do the varied starts to really get a different game.

Well, Zoomie, these statements would seem to suggest there's hope.