Starfire Historical Mod - PreRelease

[Mithel]

We have released a first "half baked" version of our "Starfire Historical" Mod. You can download a copy from:
http://www.mnstarfire.com/ww2/hoi/recommendations.html

Don't expect it to work too well. It still needs major work. But I'd be curious if any of you play test it and I'd love to hear comments.

Note: this is intended for the 1936 scenario only! I suggest you also read the "player's guide".

Tonight I'll be tweaking up a few more values and releasing another update.

- Mithel

 

[unmerged(12544)]

It looks great, specially the IC level for minors. I will test it and report my impressions.
1)VII C subs have a sea defence value of 8, CORE uses 2, the Stony Road team even 0.
2)"Fleet Supply Convoy Org."(SC -0.1) and "Naval Underway Replenishment"(SC -0.2) tech benefits must be adjusted to your lower supply consumption levels for ships.
3)There is much more steel and oil available on the world market, compared to 1.05c or other MODs.
4)In my latest game the USA war production did not kick in at 75% warentry, ENG and FRA did. The war prod. concept is an excellent idea. The French military setup was great until BEL entered the Allies, then they left Lille, Sedan and Chaumont undefended.
5)The current brigade benefits for divisions make brigades unattractive, because they are so small (+10-20% soft attack). And when you consider the price, it is not worth it.

 

[Mithel]

Thank you, I appreciate all comments and will endeavor to improve it. I expect to revamp the "steel" resource levels worldwide sometime this week.

- Mithel

 

[Mithel]

Version 0.56 has been uploaded. This contains mostly fixes to the oil resources.

- Mithel

 

[Mithel]

Ögedei Khan - virtually all of your issues have been addressed in version 0.59 which is now available for download.

Two points:
1) It was historic for the French to leave the Belgium border undefended (actually the Dyle plan called for the strength of the French army to surge forward into Belgium which is what made the gap in the Ardennes open for the Germans). I did give those provinces a slight garrison value and tweaked some other garrison figures to better defend there.
2) My intent with brigades was to lower their relative impact (since the AI is stupid and builds randomly). You should notice the cost of my brigades are SIGNIFICANTLY cheaper than Paradox's figures. I feel brigades should be available for the player to add "flavor" but not so powerful as to create a potential exploit or dramatically alter the gameplay. (with Paradox's values I *never* built brigades)

- Mithel

 

[unmerged(12544)]

1)The French setup was good, quite balanced along the Belgium, Luxembourg, and German border, like historically. I was just shocked, that after Belgium joined the Allies the AI withdrew all its divs from Lille, Sedan and Chaumont to send them mostly to Metz, what is not historical. The HOI map is misleading, when you take the rivers as a guide line, the city of Sedan should be where Metz, Chaumont and the Belgium border meet. So in Chaumont there should be a part of the 5.Army and the most part of the 6.French Army and the 1.Army should move from Lille to Antwerpen on its way to Eindhoven, the 2. and 3.Army from Sedan province to Bruxelles. Maybe there is a special French.ai necessary, when BEL enters the Allies, when you want to get those historic moves. So Sedan and Lille were undefended historically, but not Chaumont(Sedan city), where the Germans had to fight through.
2)agree

2 more thoughts:
-An idea for your timing events:
When MEX is gone, they won't work anymore. So why not create an artificial country for them. My knowledge about that is limited, but I know that there are tags available (U00, U01, U02 and so on). It would own no provinces, would not be selectable.

-I have made some experiments with the "france.ai"-file, what you can try, is:
max_garrison_prop = 0.9 #war-time desired proportion of troops allocated to garrison duty
front = { reserve_prop = 0.1 #10% of front troops behave as reserve }
garrison = {
country_priorities = { ENG = 0 }
province_priorities = {
556 = ? #Chaumont
554 = ? #Metz
930 = ? #Toulon
931 = ? #Grenoble
}
Unfortunately, I have found no way to get the AI to move troops into Belgium, not even with military access. And the AI seems to count fighters in a province as garrison .

 

[Mithel]

Let me know what you think of my new French AI. I think with a bit of tweaking we should be able to get reasonable results (French troops rushing into Belgium rather than Metz).

That's an interesting idea to create an artificial country to run the timing events. But wouldn't they be "gone" if they had no provinces? My thinking was that Mexico should never be eliminated, thus it's a reasonably safe choice. I suppose we could create a fake country out of some obscure province in Russia or somewhere out of the way.

- Mithel

 

[unmerged(12544)]

Timer country

I see two possibilities to realize that:

1)Use e.g. tag = U12 and create it with
ownedprovinces = { 1 }
controlledprovinces = { 1 }
nationalprovinces = { }
When a country is not at war and controls no province, it will vanish. 1 is a sea zone and creates an error message box with every reload/start, after that it is working. I have tried to insert province 2202 in the province.csv, but that did not work. I have simply used MEX, first gave all provinces to USA in the save game file, load the game, save the game some days later and then I looked, if MEX still existet in the new saved file, with 1 it worked, then I did remove the leaders and ministers and it still worked.

2)tag = REB does not work, but tag = PIR does (probably from EU2), it is already present and does not require to control one province. Simply replace MEX with PIR in StarfireTimers.txt. The disadvantage is, that you get one more message box "January 14, 1936.....". There is also MER and NAT from EU2, those two are simply perfect as your timer, they are working and they do not produce this additional message box.

An economic point:
When you want your IC map to stay like it is, under construction in the AI-files there are changes necessary to limit IC build up. E.g. Bulgaria, it builds IC in 3 provs, result after 1 year +27% IC. But I have learned from a CORE discussion, that the world GDP in 1947 is about 50% higher than in 1938, so it should grow moderately. In some countries stronger, this will be hard to achieve.

 

[Mithel]

Version 0.60 has been released.

Enjoy!

- Mithel

 

[unmerged(12544)]

The MOD is getting better and better. I will give my latest thoughts.

1)Did you read my recommendation to use MER or NAT instead of MEX?

2)In the germany_barbarossa.ai the base_attack_odds are to high. The combat values for the Soviet divisions are lower and with 2.0 for a human SOV player it is easy to create a stalemate by simply placing enough divisions along the front and the GER AI will sit there and do nothing. I am thinking about 1.2 to 1.6. And you have to consider the offensive German leaders. What is your experience?

 

[Mithel]

Ögedei Khan, yes I intend to looking into an alternative for using Mexico to trigger the timer events. I just haven't gotten to that yet (it's easy to get lost in fiddling all kinds of things).

Interesting point about the attack odds. I wish we knew exactly how the code estimates/calculates the attack odds. Technically if an opponent could put enough strength along the entire front then you'd be foolish to attack. I'm assuming the AI will be clever enough to shift it's attacking forces to create a "spearhead". I'll certainly keep this in mind. Thank you!

I'm working on more improvements for today.

- Mithel

 

[unmerged(12544)]

IC development

1)My idea to limit IC build up, was to use contruction = { max_factor = [0.0-1.0] } in the ai-file. E.g. Bulgaria should be limited to build up in only one province (11 IC ->12 IC; 9% growth). They have 4 provinces, max_factor = 0.25 should in theory achieve that. But testing showed, that the AI does not care about that, 0.25, 0.1 or 0, it will always build IC in Ruse and a coastal fort in Varna. What I have found out, is that 0 does not mean no build up, 2 provinces always seem to be allowed to build something, 1 IC and one other improvements. I would recommend 0.0 for ALB, CZE, BUL, ROM, HUN, YUG, GRE, AUS, FIN, NOR, SWE, DEN. As a general line, maybe ic_end_year = 1946 and ic_at_war = yes. If you really want no IC build up, just set IC_end_year = 1935. You then would have to give IC by event at certain dates to the desired countries. When you lower the available resources, there might be a shortage later in the game. Which path will you choose?

2)Your agriculture resource, is it there to keep the minors alive, or for fundamental reasons?

 

[Mithel]

I am lowering the "build" (max_factor) for many countries. I'm a bit hesitant to lower this too much, such as to create an artificial constraint, because then it becomes an AI handicap (thus making the AI even weaker vs humans).

I also tend to limit the construction to pre-war years. My general thinking on that is that once the war has broken out countries will be more concerned with their military than with long term development of their economy.

Countries like Bulgaria that have limited resources will have limited growth due to lack of resources.

Yes, my agriculture is first and foremost to more accurately reflect the economies of minor (agrarian economies) nations and to avoid the bad "meltdown" problems minors have in stock Paradox HoI.

I've just finished a major revamp of coal and steel resources based on historical data (will probably need more tuning). I'm testing it now and hope to release a new version this morning.

- Mithel

 

[unmerged(12544)]

I was thinking about your nice timer events, you do not need agriculture to keep economies like Bulgaria running. You can give them the lacking resources by event every 3 months and so use real historic IC and resource values. Concerning IC, there I thought about preventing the AI from building IC and recommending the human player not to build. And adding IC by event every year or 6 months.

 

[Mithel]

Version 0.66 has been released.

Ögedei Khan - since you seem to be working actively with my mod would you care to e-mail me directly and we can work together perhaps a bit faster?

Bulgaria had a largely agrarian economy, it's historically accurate for them to have the agriculture.

Virtually every country now has as accurate of a historical IC value as I can determine (based on actual GDP figures). With my latest adjustments with Coal and Steel the world should be fairly accurate for resources too (there are a few distortions left).

It would become a nightmare to extend those timer events to EVERY country. But I will probably extend them more to cover more historical situations.

Sure we could code the AI to not build IC for many countries (with extra resources) but that opens up more potential exploits for humans vs AI. Certainly I'm not overly concerned with exploits but I do want to minimize the possibility.

- Mithel

 

[Mithel]

Version 0.69 has been released.

Opps!

There is a major error in v0.69.

I left in the test date for the Soviet infantry build change.

For event 2202603 the January 5, 1936 should be commented and the other date lines uncommented. Also the trigger for war between SOV and GER should be uncommented.

- Mithel

 

[Mithel]

Version 0.70 is released.

 

[Mithel]

Version 0.72 is released.

Saturday Nov 8:
And now Version 0.75 is released.

Major improvements to ground combat.

Sunday Nov 9:
Version 0.76 is released

Increased costs of units to slightly reduce overall production. Tech tree tweaked.

- Mithel

 

[unmerged(19545)]

Two thumbs up from the skinny guy

Nice start.

I'm a little concerned at the cost of doctrines though: Not that they are out of line with Vanilla HOI, just that the values they use are far too high as well.

I'm playing as Canada (naturally) and it's costing me 18% of my GNP to research one doctrine. That's like 1000 000 skilled labourers employed in determining that tanks should be grouped in armored formations, not scattered amongst the infantry battalions. Doesn't really make sense does it? Particularly when you consider that some old Brit guy figured out the same thing, without the other 999 999 people coming along for the ride.

Doctrines need some really serious thought.

I'd suggest making them very cheap, with an enormous variable in the time to research, from a month, to more than a year.

 

[Mithel]

Thanks for the feedback DC. In one of the latest updates I did reduce the costs for each "level" in all technologies. Are you using those values (give me a specific example) and are you suggesting the costs be reduced further?

It's tricky to get these costs correct. If we make them too low, then big countries will be able to research EVERYTHING at the same time, the only limit being the time it takes to research a pre-requisite. If we make the values too high then the AI has trouble researching important doctrines and small countries can forget researching many things completely.

I will be doing further work on this issue this week!

- Mithel