AI too favored in large random scenarios?

[henri51]

I am starting to think the the AI is given too many advantages in large single-city random scenarios. Maybe I am just a poor player, but...[:-]

In my present AAR game (see AAR forum) I seemed to be doing OK, but in the last couple of moves, the wind has shifted and it looks like I will be overwhelmed.[:(]

Here are SOME of the advantages that the AI gets from the wiki:

* AI gets a little base resource and fuel income and a little bonus on top of each location [scripted in generic.at2]

* AI gets option to produce supply in enemy towns at 200% of the normal cost [scripted in generic.at2]

* AI gets 10% combat and 25% movement bonus [scripted in generic.at2]

* AI gets free transfers [rulevar for this enabled in generic.at2]

* AI gets free construction of units and HQs [rulevar for this enabled in generic.at2]

* AI will get a very limited number of free factories / resource upgrades over time, but the number increases for AI+ and AI++ [scripted in generic.at2]

In every random game I have started, within a couple of moves, at least one of the AI nations is producing over 10 times my oil and raw, and ALL of the AI nations are producing much more than I am. This is because to upgrade an oil well, one must spend a lot of PPs, which take a LOT of time to obtain, whereas the AI getsupgrades for free, as well as factories (it is not clear how many)that allow the production of units without having to use their city production, which can concentrate on supplies and infantry.

In small scenarios where the human player using superior tactics and organization can overwhelm the AI nations before they can get organized, but in large scenarios (say using xlarge maps) even the most badly placed AI with just a couple of cities and resources can eventually surpass the human production of 10 or more captured enemy cities, with the result that the human will eventually find himself fighting on 3 or 4 fronts against superior armies.[:-]

For example in my AAR game ("Random take 2"), in my last message I described how I was going to fight holding battles against the French and Russians while dealing with the British, who although holding only 4 of 11 cities of their own nationality, are managing to outproduce me and to hit me with stacks of 15 aircraft, which would instantly bankrupt my oil if I tried that and to hold the line. In the meantime, using a stack of some 15 aircraft, the Russians have obliterated my units holding the neck of land blocking their passage and will now have free rein in the Far East. In the meantime, despite being atttacked in the rear by the French, the Japanese are mounting a huge army that my holding action will probably not be able to hold for long.

In sum, for large maps and scenarios that last a long time, the AI cheats give the AI too much advantage and need to be toned down.Another possibility is to go back to the original AT idea and to let captured cities produce units and not only resources. At the present time, holding 5 times as many cities as an AI player is not enough to keep up![:-]

Henri

[Tac2i]

Note that after the first patch every 1 TOWN START game will begin with everyone at peace. Currently each game starts with everyone at war.

If you have captured your share of enemy cities and resource hexes, you should be able to field more tanks and planes and have the oil/ore to support them. That said, since everyone is at war with you, it may be an impossible situation.

EDIT: beta patch is now available.

[henri51]

ORIGINAL: Webizen

Note that after the first patch every 1 TOWN START game will begin with everyone at peace. Currently each game starts with everyone at war.

If you have captured your share of enemy cities and resource hexes, you should be able to field more tanks and planes and have the oil/ore to support them. That said, since everyone is at war with you, it may be an impossible situation.

EDIT: beta patch is now available.

Maybe and mavbe not. Here is a graph from the above AAR game. The graph has been similar throughout the game, and the graphs for force, raw and production are similar, and the graphs have been similar to this in every game I have played so far (all of the on xlarge maps).

I suspect that the problem is that the gain of capturing an enemy city (8 pp per turn once it ramps up production) is marginal compared to the PPs required for example to upgrade a resource or build a factory. For example, let us say that one has captured 5 enemy cities, each of which yields 8 PPs per turn; it requires 80 PPs to upgrade a mine to level 2 (not to mention engineering points), which means that it will take 5 turns to gather the PPs to upgrade a single mine, and 25 turns to upgrade 5 of them to level 2. In practice this means that it will take some 30 moves before the human is seriously geared up to have all his resources upgraded to level 2 and have a a few factories and airports. In the meantime, and the graph shows, the AI strength has been growing exponentially.

Anyway, I will try a new game with the patch and see. If it doesn't work out, I will try using TWO HUMAN slots to play against the AI...[X(]

Henri

[DakaSha]

My issue is its too easy on AI and too hard on AI+
:-/

[henri51]

ORIGINAL: DakaSha

My issue is its too easy on AI and too hard on AI+
:-/

I have yet to see a report of a human winning against the AI in a large random single-city-start scenario.In the AAR "Random take 2", I held my own and was even winning until about move 30, when the huge AI advantage in production caught up with me.[X(]

If anyone has ever beat the AI in a game like this, Iwould like to hear about it...[&o]

Henri

[Tac2i]

Give me your setup parameters and I may give it a shot.

[theonlystd]

Me and a friend smashed the Ai on a 1 city start rather easily.. Tho it was with no roads on so that played a big part.

Tho do agree with dak. Ai on default seems to easy and on + is really frackin hard/impossible on most start setups .

[henri51]

ORIGINAL: Webizen

Give me your setup parameters and I may give it a shot.

Five player, single city start, dry land. If you want to try the same situation I have about 10 saved games going from the beginning to the last move that I can email.

Henri

[Tac2i]

I've started one earlier this evening with 4 AI. I'm about 12-13 turns into it and I'm at war with three of the AI. Two declared on me and I declared on a third. I have offensives rolling up two of the AI at the moment, one offensive moving faster than the other. Across the ocean I have only one non-native city holding out against another AI. They are no direct threat to me as yet.

It does appear to me that 1 TOWN START maps basically build an "at start" nation but only give you control of the capital. You have to go out and take control of your native "neutral" cities before the AI captures them and forces you to fight to capture them. I had to wrestle control of one so far from the AI. Your regime's people group could be separated into separate land areas too. I have two areas in this game that I've noticed so far.

Definitely plays different than the old AT 1 TOWN START. At least in ATG you have two ways to start a random game. In the old AT it was 1 TOWN START only for random games. I'm thinking that the standard ATG random game with an at start nation might actually be easier for noobs to learn the game than these new 1 TOWN START games.

[DakaSha]

ORIGINAL: theonlystd

Me and a friend smashed the Ai on a 1 city start rather easily.. Tho it was with no roads on so that played a big part.

Tho do agree with dak. Ai on default seems to easy and on + is really frackin hard/impossible on most start setups .

Ive been suggesting intermediate levels but i dont get jack for a response. Please post about it
Its super important to me because i cant get a fun game started

[theonlystd]

ORIGINAL: DakaSha

ORIGINAL: theonlystd

Me and a friend smashed the Ai on a 1 city start rather easily.. Tho it was with no roads on so that played a big part.

Tho do agree with dak. Ai on default seems to easy and on + is really frackin hard/impossible on most start setups .

Ive been suggesting intermediate levels but i dont get jack for a response. Please post about it
Its super important to me because i cant get a fun game started

Well takes just a smidge of work but play on Ai

And give the Ai regimes a couple extra cities around there capital ...

[DakaSha]

shouldn't be necessary.

[Jeffrey H.]

ORIGINAL: DakaSha

ORIGINAL: theonlystd

Me and a friend smashed the Ai on a 1 city start rather easily.. Tho it was with no roads on so that played a big part.

Tho do agree with dak. Ai on default seems to easy and on + is really frackin hard/impossible on most start setups .

Ive been suggesting intermediate levels but i dont get jack for a response. Please post about it
Its super important to me because i cant get a fun game started

Try modding one up yourself. Many of the improvements to the PvAI game in ATG came from Zaratoughdas ARHS mod in AT. He found most of these improvements via trial and error.

[Jeffrey H.]

ORIGINAL: theonlystd

Me and a friend smashed the Ai on a 1 city start rather easily.. Tho it was with no roads on so that played a big part.

Tho do agree with dak. Ai on default seems to easy and on + is really frackin hard/impossible on most start setups .

I think this was the case with AT as well. It was modded to give a somewhat better challenge but there was still a large gap between AI and AI+ and in essence the game played the same except for the quantity of units that you had to fend off was increased with AI+.

The base character of the AI in ATG was essentially unchanged and I'm finding the same underlying behaviors surfacing up again in ATG. Not sure if modding will get us anywhere now, we'll have to wait and see about that.

ATG is a different game now for certain but against the AI it's about where we left off with modded AT.

[Stardog]

Hello<

Well my hats off to you all!.. (Guys kicking the AI's butt)

I haven't even started a random game just been playing the French vrs Arabs Scenario like 4 times now and can't win...?

Yall must be pretty good...

SD

[Tac2i]

At least in my games I'm not finding that the AI is too favored. I want a challenging battle and the AI usually gives me one. I started a 4 AI 1 TOWN START game and 3 of the 4 AI are at war with me (two declared on me and I declared on another). Thankfully one of the AI, after running me off his land after I invaded doesn't have a border area to get at me and is at war with another AI. I've been battling against the the Republic and Nippon. Two years into the war (24 turns) I've crushed the Republic and now mopping them up. The war with Nippon was somewhat stalemated but now I have, at least at the moment, the upper hand against them.

Click on the attachment link below to view the map.

[Krafty]

Next attempt at a huge game me and Theonlystd will post an AAR. I think we may lose our current game. One ai has almost 4x our production. Sad thing is that I made the map were playing from scratch!




Perhaps I gave the AI too much production? Heh. Theonlystd is the Soviets at the top right corner. Im the French, the purple at the lower left corner. Map loop is on, I think Im going to try a sea invasion on him and end it quickly before the AI gets us. So far we have pushed the AI back and captured a few  things (AI are all on +, sorry no ingame pic, its on his turn and I dont know his password to get some screenies) We have capital city areas about the same as the shown AI (the chinese) but with the bonuses and the +, the AI has some 400k a turn production, while we have around 120k or so. Their oil and raw income of course is through the roof. They have 100s, literally 100s of armored cars, inf gun IIs, mortar IIs, mg IIs, on our borders that were engaged with, while STD said he had 15 light tanks fending them off, I have maybe 30 armored cars. Of course attacking with those 30 armored cars insantly uses up my 5,000 oil. (This is with the beta patch, so oil movement costs should have been fixed, ive been moving 1 hex at a time anyways since I noticed the bug).

The key appears to be THROW INFANTRY AT THEM...throw them in hordes, have more hordes waiting behind the lines to take advantage of the lowered enemy readiness. The idea being that even with the increased production, the AI must still obey the the Stack Limit rules. 115 Rifle Is or Smg Is (if you can get rangers or guerrillas woohoo) is an economical 100 stack points plus some change, in our game im producing 800 rifles a turn.  So I get 7 divisions full of infantry to throw at the AI (they must be making 3 or 4 thousand a turn)  I use some artillery behind the lines to lower the enemies readiness, then throw a few hundred rifles at that hex...they die of course, but they drastically lower the readiness of the enemy unit. My next line behind the front of hundreds of rifles then steps up, and throws themselves at the AI unit too. With the enemies lowered readiness, breakthroughs are likely when outnumbered 4 or 5 to 1. Most of the enemy units surrenders, I pick up alot of XP, and roll the survivors into 100 Stack Point divisions, and prepare for the next turn. On the AIs turn, they attack with their rear lines, which are of course WELL over the stack limit, giving me a nice advantage. I have full, or near full readiness troops with good XP to hold the line. I end up with 1:1 kills/losses during the AIs turn, but the line holds. Over the course of the last 10 turns doing this, ive captured 3 cities (one french!! more infantry!) a few oil a level 3 raw, and concentrated on getting factories producing artillery (40 a turn now, which will be set with 42 horses, and a flak gun, enough to entirely decimate one enemy hex, over the next 20 turns, ill employ enough arty to totally destroy the enemies line, then will do the rifle horde trick for a net gain of TWO lines of enemy hexes...) My hope is that by rolling high xp units into other medium xp units, and the constant deaths keeping my division count the same, that I should have 99 XP staff in no time...my spare PPs will eventually upgrade my staff, and grant me more bonus.

Then I should have no problem slaughtering 10:1 odds.

But who knows...10 turns from now things may look different. Next game, ill put up an AAR. Theonlystd and I dont lose to the AI in any game very often (we've been playing together for almost 5 years now, in every game you can imagine) and when we do, we consider it a personal challenge to defeat it.


PS.

If the generic master file had Counter Battery, as in Bombur's mod, and then the AI actually BUILT any artillery, youd never beat them.

The AI seems to be terribly bad at dealing with hordes of artillery. 40 artillery will kill 100 stack points of just about anything in a hex, and RARELY will the AI attempt to destroy your artillery either with arty of its own, or with airpower, leaving you a unit that costs very little supply for its damage output, and needs no fuel, whose numbers grow every single turn...in this game I shall surpass 1000 artillery before the 2nd year is out. Even with 400k production, the AI I know and love from AT (even with multi attacks and bonuses) wont be able to compensate.

[mgaffn1]

We should call this the "henri51 challenge"

Has anyone beat the AI in 5 AI player (1 human, 4 AI) single city start?

(must confess I have not.... I've already made some "tweaks" with the editor to my random games & must consider that cheating)
if you have, then I salute you, and would like to hear your advice.[&o]

cheers,

[sIg3b]

ORIGINAL: henri51


I have yet to see a report of a human winning against the AI in a large random single-city-start scenario.

How about large random 1v1 with each player getting half of the map at start? Have you tried this? Would wonder if this is easier or even more difficult?

[henri51]

ORIGINAL: Tesuji

ORIGINAL: henri51


I have yet to see a report of a human winning against the AI in a large random single-city-start scenario.

How about large random 1v1 with each player getting half of the map at start? Have you tried this? Would wonder if this is easier or even more difficult?

Good question, this could be more or less equivalent to the AI nations being allies, but not quite. I am not sure, but I think that the human would win thanks to better tactics and strategy, despite the huge AI advantages.

Also it is not clear how the AI advantages scale up: for instance if the AI gets a certain fixed number of extra units at the start, this becomes negligible in the long run. Also in the long run, the AI advantages in raws and fuel become irrelevant (in my AAR, at this late point I have plenty of oil and raws - the shortages were in the early stages).

Henri