Problem with AI's build routines

WW2: Road to Victory is the first grand strategy release from IQ Software/Wastelands Interactive, which covers World War II in Europe and the Mediterranean. Hex-based and Turn-based, it allows you to choose any combination of Axis, Allied, Neutral, Major or Minor countries to play and gives you full control over production, diplomacy, land, air and naval strategy. Start your campaign in 1939, 1940 or 1941 and see if you can better the results of your historical counterparts. A series of historical events and choices add flavor and strategic options for great replayability.
User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

Problem with AI's build routines

Post by Jim D Burns »

I’ve run into a problem with the AI’s production choices in game. In my game as Germany its late 1941, and as you can see from the below screen captures, Russia is almost totally out of units (FOW is off).

The AI generally chooses to rebuild expensive damaged units every turn instead of building new units. There needs to be a default OOB structure for each country that the AI has to build before things like replacing losses in damaged units can be chosen. That or the PP’s need to be divided into two pools, one that the AI can only build new units from, and the other that the AI can replace losses with.

Otherwise Countries like Russia will soon run out of units and become pushovers. This kind of glaring problem should have been caught in beta testing.

Jim

Image


Hanal
Posts: 2295
Joined: Sat Nov 01, 2003 6:08 am

RE: Problem with AI's build routines

Post by Hanal »

I have not gotten far enough into the game to see this yet, but unless there is some massive re-enforcement schedule that kicks in once the invasion occurs, then what I'm seeing here makes no sense...have you invaded yet?

and as an aside...why is there an all sea/lake hex in the Crimea?
User avatar
Erik Rutins
Posts: 39759
Joined: Tue Mar 28, 2000 4:00 pm
Location: Vermont, USA
Contact:

RE: Problem with AI's build routines

Post by Erik Rutins »

Jim,

A few questions:

Was this a game started with the 1939, 1940 or 1941 scenario?

It looks like you did very well in your invasion, did you encircle and destroy most of the initial Soviet forces? I'm wondering if some of thiis is because you have just seriously outplayed the AI as well as it not spending its points wisely.

In any case, I'll make sure we test this and re-balance it in the first update. I'm also surprised if this was an issue in earlier pre-release versions but made it through to release. I don't recall it being reported though.

Regards,

- Erik
Erik Rutins
CEO, Matrix Games LLC


Image

For official support, please use our Help Desk: http://www.matrixgames.com/helpdesk/

Freedom is not Free.
User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

RE: Problem with AI's build routines

Post by Jim D Burns »

1939 start, all countries on easy including Germany. I chose to go the easy route because Germany usually gets about 180 production after all initial conquests and the US only gets 200 production total. That’s just a 20 point difference, so by doubling it, the US will get 40 more points than Germany. If you’re going to be rebalancing, you should double the US at minimum, they are far too weak in this game.

Of course I played well by using encirclements to destroy units, but this is late 41,and I haven’t yet reached as far in the South as was historically reached, so it’s not like I was doing extremely better than historical. I should note I left the south pretty much alone and allowed Italy and Slovakia to head there under AI control. Were I in control, Sevastopol would most likely be taken by now.

The AI did well when it had units to fight with, but gradually it simply ran out of units because it spent way too much PP on rebuilding losses and not enough on building new units.

I think Russia has a unique situation to other countries. It is going to lose its starting army, so it needs to be busy building a new one in the rear, not trying to rebuild the one that is going to die. The Ai did no such thing, it kept pouring points into units that were almost encircled and they just died the next turn. The AI needs some kind of recognition ability that recognizes a unit is a write-off even if it has a thin supply route open to it for one last turn.

Jim
comrade
Posts: 167
Joined: Wed Jun 07, 2006 6:35 pm
Contact:

RE: Problem with AI's build routines

Post by comrade »

The AI spends PP on replacing losses because buying new units is generally less effective. 1 strength point replacement is 50% cheaper than 1 strength point of brand new land unit. In case of air units it is 66% cheaper.

I'm curious how many PPs did Russia have? They might have been accumulating PPs for some tech investment or buying an expensive unit (wchich is of course an AI mistake in this circumstances) . It would be interesting to see the save game. You can also check it yourself, just press F1, this will open a 'cheat' window, and set USSR control to human, click end turn and see what's going on there :) If you find that Russia has loads of PP than it means that they were sparing it for some other purpose...

The other way to test this is to stop the offensive for a few turns, allow them to rebuild all existing units and then and see if they are buying new units and at what rate. The reason I ask about it is to find out if something just stucked, and Russian AI wasn't buying anything at all, or is your theory true and they did spend everything on replacement (which in turn may mean that this is a balance issue and Russia should have more PP)
User avatar
Vypuero
Posts: 232
Joined: Sat Apr 07, 2007 5:11 pm
Location: Philadelphia, PA USA
Contact:

RE: Problem with AI's build routines

Post by Vypuero »

At least you could adjust by adding to Russians PP (setting Russia to Easy)
User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

RE: Problem with AI's build routines

Post by Jim D Burns »

For clarification, this is the turn just prior to the Pearl Harbor event.

The USSR has 120 PP saved and has a production of 225 (they are set to easy along with all other countries). Their war economy is only at 80% though, this could be a big part of the problem. A country should jump to 100% when war starts and perhaps gradually increase 2% a month or so from there.

I’ll try running the game for a month or so without moving and see what they build.

Jim


User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

RE: Problem with AI's build routines

Post by Jim D Burns »

I ran it for about 6-8 turns. Russia built a few infantry corps, but not much else. It had about 100 PP saved after that, so it spent everything it could, it simply lacked the funds to rebuild quickly. I guess it would eventually rebuild itself if I left it alone for a year or so.

I’m now convinced that the slow gradual buildup of the war economy hurt them, probably almost as much as the AI pouring PPs into units about to be surrounded. War economies should grow due to political events when a country is at peace (France surrenders, Yugoslavia, Denmark and Norway attacked, etc.), and should immediately jump to 100% once war starts. Germany simply has too great an advantage when it collects 100% war economy and its opponents get 20%, or 40% or 60% etc. for such a long a period once they are attacked.

Jim
comrade
Posts: 167
Joined: Wed Jun 07, 2006 6:35 pm
Contact:

RE: Problem with AI's build routines

Post by comrade »

War Economy was supposed to model gradual switching from peace-time economy to war economy, this is why it grows only when a country is at war (it grows a bit faster for communist and national regimes). Russia starts with WE=45%, this may be too low. I'll increase this on Ribbentrop-Molotov pact and Vichy france events (will do the same for USA). I'll also consider adding a completely new event for Russia that will set WE to 100% as soon as the war breaks out. And adding Lend-lease event (which I also plan for next patch) with additional PP from USA should also help :)

BTW, there is a "Siberian reinforcements" event that fires in November or October 1941, it gives Russia +500 PP. Will also consider increasing this value a bit.


benpark
Posts: 3072
Joined: Mon Aug 12, 2002 1:48 pm

RE: Problem with AI's build routines

Post by benpark »

I haven't looked through the events, but it might be a good idea to give Germany a production boost in 1943 due to the "Total War" production "miracle". Maybe a "Bulge" increase for the winter of 44, too. It would help make Allied vs AI play a bit tougher.
"Fear is a darkroom where the devil develops his negatives" Gary Busey
User avatar
geozero
Posts: 1816
Joined: Wed May 22, 2002 4:00 pm
Location: Southern California, U.S.A.
Contact:

RE: Problem with AI's build routines

Post by geozero »

Just a question...for Comrade... assuming you are the DEV.

While Mr. Burns has done a great job playing the game and providing suggestions, I'm wondering how many people actually beta tested this game prior to release and how many more are still testing... I would think that reliance solely on one gamer's experience (no offense Mr. Burns) to tweak events and balance would be folly. I do not have the game (yet) so I can't chip in... so how many people have played the game through?

If you need a tester... sign me up as well.

JUST SAY NO... To Hideous Graphics.
User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

RE: Problem with AI's build routines

Post by Jim D Burns »

ORIGINAL: comrade
War Economy was supposed to model gradual switching from peace-time economy to war economy, this is why it grows only when a country is at war (it grows a bit faster for communist and national regimes). Russia starts with WE=45%, this may be too low. I'll increase this on Ribbentrop-Molotov pact and Vichy france events (will do the same for USA). I'll also consider adding a completely new event for Russia that will set WE to 100% as soon as the war breaks out. And adding Lend-lease event (which I also plan for next patch) with additional PP from USA should also help :)

BTW, there is a "Siberian reinforcements" event that fires in November or October 1941, it gives Russia +500 PP. Will also consider increasing this value a bit.

Many of the countries should be at or near 100% when war starts, and if they are not, they should immediately jump to 100%. Otherwise the simple math of Germany collecting far more PP then they are, will guarantee a quick end. Events like Vichy, declaring war on minors etc. should all increase war economies of all countries (not just a select few) when those events happen. That way the final jump to 100% isn’t too dramatic of a jump.

The gradual increase you currently have should occur after a country hits 100%. After all, most countries didn’t max out their production until near the end of 1943. So a small 2% increase every month would be a good way to model this slow maximization of available industries. My guess is most countries will eventually hit about 150% before the end of the game, which isn’t too bad considering the changes their economies went through after 1941.

As to the US production, you have to remember, the US out-produced the rest of the world combined during the war. Granted a large chunk went to the Pacific, but it should at a minimum be out-producing the combined efforts of Germany and Italy all by itself.

Which reminds me, how exactly can the US get its air groups to Europe? It’s too far for a transfer move and planes can’t board ships. Am I missing something obvious here?

As far as Russia is concerned, there is a problem unique to Russia that demands the AI choose new units over repairs. That is simply the vast area of the country. Russia needs a standing army large enough to span the length of the country, or Germany can easily surround and eliminate its army at will.

That means the AI code needs to have a default OOB structure it will always choose to build before it spends money on upgrades, tech or repairs. Right now it doesn’t do that, so Russia runs out of units pretty quickly. So if it takes 100 divisions (just a guess, may take far less) to create a solid front, Russia should spend 100% of its income on new units until it has 100 divisions on map. After that then it can default to the regular production code.

Jim
Mraah
Posts: 1085
Joined: Wed Feb 20, 2008 6:11 am

RE: Problem with AI's build routines

Post by Mraah »


Considering the problem with the build routine ... Has anyone played the Allies yet?

I don't have the game (yet) ... But, if Russia had a human player would it be able to stop the Germans at Poland whilst the German AI spent PP on rebuilding the forward line units instead of creating new units?

I'm kinda flipping the coin around.

Rob
User avatar
Vypuero
Posts: 232
Joined: Sat Apr 07, 2007 5:11 pm
Location: Philadelphia, PA USA
Contact:

RE: Problem with AI's build routines

Post by Vypuero »

Has anyone else activated Vichy?  I did and it did not seem to own any cities so got no money- nor did it have any units - can this be fixed?
 
I am playing a campaign on normal - France, Low Countries, Poland all easy - Russia not so much.  They have big swarms around cities and you can't just ignore them.
User avatar
Vypuero
Posts: 232
Joined: Sat Apr 07, 2007 5:11 pm
Location: Philadelphia, PA USA
Contact:

RE: Problem with AI's build routines

Post by Vypuero »

Oh another problem - I sent an Italian Air Division to Crete - it can't move now forever because it is OOS!
 
Fix maybe?
comrade
Posts: 167
Joined: Wed Jun 07, 2006 6:35 pm
Contact:

RE: Problem with AI's build routines

Post by comrade »

I can confirm the Vichy problem - they don't use the PP from their cities because these cities still have France as an original owner. Will be fixed.

Don't send anything to Crete - there's no port there, which makes this island not important in the game.

As for the AI build issue, this will be addressed in 1.20 by series of fixes. I also suggest using difficulty settings, e.g. setting USSR/UK/France to very easy and Germany to Very Hard.

User avatar
Jim D Burns
Posts: 4001
Joined: Mon Feb 25, 2002 6:00 pm
Location: Salida, CA.

RE: Problem with AI's build routines

Post by Jim D Burns »

ORIGINAL: comrade
and Germany to Very Hard.

The problem with this solution is Germany will then simply be unable to afford to buy an army large enough to protect all its borders, let alone replace losses and spend money on tech. I realize this would be very challenging to try and play, but it has no basis in the reality that is WWII.

Germany (and all other countries) needs enough PP to at least accomplish what they did historically. Then balance everyone else against Germany based on what they accomplished.

The US should out-produce both Germany and Italy combined by about 50% after the initial conquests of France, Yugoslavia, Norway, etc. Some of this PP would of course then go into lend lease convoys to both Britain and Russia.

Russia should out-produce Germany by about 20%-30% after the initial conquests. Much of Russia’s production will end up being captured, so it needs a large enough production that it can sustain the hit and still be able rebuild its army several times by the end of 1942.

Great Britain should produce about 60%-70% of what Germany produces after the initial conquests.

These are rough estimations, but you get the picture, all production needs to be balanced against Germany’s capabilities, since Germany is the main antagonist of the game.

With the current income levels, Germany can barely afford an army large enough to take on Russia by June 41. It definitely cannot afford to build its historical Barbarossa force.

I’d say Germany should be building about 250 PP every turn by the time the 1940 conquests are done instead of the 180 or so it gets now. This should give Germany just enough PP to get a decent army built for Barbarossa, but not so much that they can massively overbuild.

You can easily up the costs of tech to offset the extra income, but I’d leave everything else the same.

Man, I’d really love a scenario editor for this game.

Jim
comrade
Posts: 167
Joined: Wed Jun 07, 2006 6:35 pm
Contact:

RE: Problem with AI's build routines

Post by comrade »

Well this is what difficulty levels are for, you have to take into account that some players will do better than others using the same settings.

I agree that Russia and USA are a bit under-powered (next patch will address that) but I also think that Germany-Normal is a bit overpowered. In 1.20 I set War Economy for Germany to 50% in 1939 scenario, it will reach 100% around April, then it will grow 1% per month (this rule will apply to all countries that reach 100%), according to your suggestion. USA and Russia will have more PP and there'll be Lend-Lease.

At the moment game has only a simple map+units editor, to edit all other things I use excel and edit csv files manually. In 1.20 I will translate countries.csv column headers to english - this will make life easier for any modders ;)

Thanks for all suggestions and opinions.
User avatar
Redmarkus5
Posts: 4454
Joined: Sat Dec 01, 2007 1:59 pm
Location: 0.00

RE: Problem with AI's build routines

Post by Redmarkus5 »

I wasn't happy from the outset with the presentation of the map and units etc.  I've only played up until the invasion of the Low Countries, but after reading what Jim has to say, I'm going to put this on my virtual shelf until the patch comes out and the modders have had a chance to get busy.  The invasion of Russia is the main event of WW2 in Europe - how could that not be the core focus of development and testing?  I still think this game has potential, but it's not quite ready yet.  Should not have been released at this stage IMO...
WitE2 tester, WitW, WitP, CMMO, CM2, GTOS, GTMF, WP & WPP, TOAW4, BA2
User avatar
Widell
Posts: 890
Joined: Wed Apr 27, 2005 2:25 pm
Location: Trollhättan, Sweden

RE: Problem with AI's build routines

Post by Widell »

ORIGINAL: Jim D Burns
Germany (and all other countries) needs enough PP to at least accomplish what they did historically. Then balance everyone else against Germany based on what they accomplished.

The US should out-produce both Germany and Italy combined by about 50% after the initial conquests of France, Yugoslavia, Norway, etc. Some of this PP would of course then go into lend lease convoys to both Britain and Russia.

I’d say Germany should be building about 250 PP every turn by the time the 1940 conquests are done instead of the 180 or so it gets now. This should give Germany just enough PP to get a decent army built for Barbarossa, but not so much that they can massively overbuild.

You can easily up the costs of tech to offset the extra income, but I’d leave everything else the same.

Well, here's a classic problem when dealing with this type of game (and HOI has the same "issue". Also AT in some scenarios) and that is the eternal question what it is the game should accomplish, and what the player should accomplish. I may be completely off here as I haven't gotten this title yet (trying to constraint myself until after summer/vacation, but it is very hard [;)]), so please correct me if I'm wrong. This, and HOI are what I would refer to as "open ended" games, while for example many of the historical scenarios in TOAW as well as ACW, WITP etc, are much more decided in their outcome. Thereby not saying, the scenarios in these games only have one ending, only the % of games that end with the historical outcome is high compared to, for example, HOI. The open endedness comes from the fact that the player(s) control production, tech levels, diplomacy etc etc. In the case of WW2:RTR, there is also the fact that the player can control as many nations as he wants. This of course open up for more of what I'd call "role-play". Have a look at the AAR,s over at the HOI forums and you will see what I mean.

I do agree that the PP levels need to reflect the situation at the start of each scenario, and that subsequent events have a reasonable impact on the PP for each nation, but to put a system in place that would guide the German player to do what Germany did historically, does in my mind forfeit some of this games potential dynamics. If historical accuracy beyond the starting points (incl. PP, Tech levels and OOB's etc) plus a reasonable set of events to cover for at least historical choices by the players is what we are looking for, there are other very well developed and proven titles out there, most of them with mods to make them even more historical an many levels.

I don't want this to be the "can of worms" discussion comparing historical to a-historical, but more of a discussion about what game to choose depending on what you as a player expect the game to do for you, and what options to choose from you have in terms of products.
ORIGINAL: Jim D Burns
Man, I’d really love a scenario editor for this game.

Could not agree more, although the files seem moddable as "stand alone files" from what I read on the forum. Maybe someone with coding skills could create a simple tool to get started?
Post Reply

Return to “WW2: Road to Victory”