Beta3, Risk Battle Mod

[unmerged(42223)]

Welcome to the Risk Battle Mod Beta3 test thread !

Beta3 is proving to be very exciting though admittedly I am the only one who has seen it.

I am running AI game test series 59 as I type away.

Risk Battle Mod has matured greatly via beta and has taken great strides forward. Of the issues 'uncovered' I have addressed the following:

-Unit balance; build, use and maintain.

-AI panic builds and corrections

-And most importantly, one thing I was very determined to eliminate, over large armies and IC build up. What will not be seen anymore are 500 plus division armies, with the possible exception of a consolidated China by one of the three Chinas. This menace remains for game play purposes. If you can build and maintain a 500 division army it should be only because you have ALREADY conquered the world !!

I have countered many a human gameplay move that I observed during beta. I will not elaborate too much in order to better see how this pans out yet again. I been feeling like my AI's are my munskins and have been taking too much of a beating...

Anyway, to conquer and be conquered...

There are added two new models of militia.
There are added a line of four new models of brigades

Some more highlights;
The mod runs faster, the AI's can drop off like flies as others grow to large estate. Or they might just duke it out.

Manpower pool is 'reworked'. You start with a large pool, that once is gone will most likely be gone forever, unless you get big very fast or you consolidate China as a China. Yet manpower is not stingy, I like building what I want and not starve out, unless I screw up badly. Same will generally go with the AI's. Stay at war expanding and you will not get into MP trouble. But it is far more tightly curbed, it needed to be, I was giving it out like candy before.

The mod remainds three-phased and I believe now the AI's are better positioned to challenge all along the way.

There is more I am sure but I wanted to get this posted.

Happy gaming !!


Oh I forgot to mention that I have been beefing up the starting OOB. All land divisions now start with brigades.

 

[GeneralHannibal]

Sounds good, I eagerly await its release.

 

[unmerged(42223)]

Additional information as I continue to fine-tune.

You will want to study unit modifiers, costs and attributes, same with mixing them with brigades. Loads of changes in there. Emphasis has been greatly amplied for each unit being unique; like what it can take with it, or what it is vulnerable against. As well some combat modifiers are changed so they all work for the AI. There is a special brigade redone and available via OOB and to build from game start. This is in addition to formerly mentioned brigades in post #1 of this thread. So far I cannot confirm an AI build as per my design, yet that is alright it is meant as much or more for the human who will better weather building it in game.

This mod has the 'difficulty' settings tweaked both as game governors or constants and for change of play with those who are brilliant top of line players. The rest of us should ONLY play on VERYEASY. This is no joke.

You can however keep the AI on 'aggressive'.

Back to specialized unit use. I have the AI's geared to build what they need, if in the desert they will favor cavalry, if confronted with jungle then marines or forts and mountains, 'Heavy Assault' as I call them. Infantry to attack cities and militia can be wiped out in very bad weather conditions, on and on. Cost is now very similiar so choosing will be more for use than IC/MP bargain hunting. (no offense Kanitatian ).

A strategy tip; if you lose your core provinces you will die. Simple and if you can do the same to an AI they will cripple sooner than later. Empires are more penalized for drifting off their core base.

Another tip is if you panic build militia; do not upgrade them they are extremely barbed for control purposes. Rather disband and rebuild if necessary, yet there is a price to pay here or do not research the next qualifying tech which again may or may not hurt for awhile depending on your overall needs.

 

[Kanitatlan]

Back to specialized unit use. I have the AI's geared to build what they need, if in the desert they will favor cavalry, if confronted with jungle then marines or forts and mountains, 'Heavy Assault' as I call them. Infantry to attack cities and militia can be wiped out in very bad weather conditions, on and on. Cost is now very similiar so choosing will be more for use than IC/MP bargain hunting. (no offense Kanitatian ).

No offense taken. I don't like being pushed down a one size fits all route for units any more than anyone else does. I just find that HOI2 has a very definite "vanilla infantry is the most efficient way of applying IC to a particular task" and this tends to flow into all the Mods including this one. With a risk Mod we are very much in a position of doing anything we like.

Are you still planning to let me look through a pre-release of the unit definitions for appraisal. From the sound of your changes I will need other files to factor in the situational modifiers.

My ideal unit set would incorporate the following issues ...

Different profiles of optimal unit selection for manpower optimisation as against IC optimisation.

All brigades having some effective purpose (in standard HOI2 AT, TD, AA are pointless and rocket brigades are too similar to artillery to be relevant).

Research always produced units that were more cost effective at winning battles than earlier tech levels.

Different unit types were more cost effective at their assigned tasks than other unit types.

Options for additional investment in unit production to optimise the operational costs of units - ie the trade off between supply cost and build cost real.

 

[unmerged(42223)]

Are you still planning to let me look through a pre-release of the unit definitions for appraisal. From the sound of your changes I will need other files to factor in the situational modifiers.

Yes, and I trust I have accomplished much of what we hope can be accomplished, however a detailed flow chart or look at unit stats and attributes all at once from several directions is very helpful. I have been going mostly via intuition and reaction to ever long and frequent game tests. AI only observing not human playing which still leaves a blind side.

Just last in running another long test out five years I have marked what I now can definitely say is a hard-coded AI build issue. (from my perspective). I am testing the MP flow to insure that armies are not built in sizes exceeding 500 divisions in less than five years. With two exceptions a consolidated China by a China or some very mega empire that controls a whole continent.

In this game neither came about and all armies numbered under 500 divisions but not by all that much, yet MP was drained and the mass building had ceased. However, here is the 'issue' there must be some hidden panic build the cheapest most cost effective unit (not militia) that the build queing allows to build once the MP lowers down to a we are now going to run out mode. It switched to mass quanitities of infantry. A couple hundred each for the most of the larger countries. But proportionally this is counter to my settings !!!

Relative build scheme is SET to NO.
That means build by the numbers no matter what ????
Crap.

I guess I need to edge up time/cost of infantry, yet I hope not imbalance choice or I may simply trigger the AI to resort to marines or something when MP drains down ?

This is where I am at now with this, raise the flow, lower the flow? Give more starting MP or more random MP later or less random MP later. What the computer does is now one thing and the humans another.

 

[Kanitatlan]

My thought about panic building was to instead create some events to simulate instant call up and deployment of militia. My experience playing the game was that the AI never achieved any panic building against me as I conquered them too quickly for any units to actually be deployed. The only solution to this that i can think of is either free units from events or set the build time for militia to be very very short with substantial IC cost per day to maintain the total cost - e.g. 10 IC for 10 days.

In fact there is no good reason for not making militia very cheap to build as long as their combat effectiveness remains low. The important element is that militia should not be capable of increasing the cost effectiveness combat power of an army cosisting of other units. They need to be designed to be purely defensive. Once I have the full set of unit definitions I can give you a "minimum safe cost" for your militia units that prevents them from becoming a normal element of an effecient army.

 

[unmerged(42223)]

I am looking forward to wrapping up here soon, I was sharing the lastest quandary.

Largely due to your feedback concerning militia I had already made these changes to them in Beta3 a while back:

Militia model 0
Cost 15
buildtime 12
defensiveness 20
softness 150
hardattack 1
softattack 4
etc.

Militia model 1
Cost 8
buildtime 24
defensiveness 21
softness 130
hardattack 1
softattack 4
etc.

Militia model 2
Cost 14
buildtime 17
defensiveness 22
softness 110
hardattack 2
softattack 10
etc.

Militia model 3
Cost 15
buildtime 17
defensiveness 20
softness 90
hardattack 3
softattack 12
etc.

There are other settings of course most significate is that militia model 0 doubles as partisans and also sports a 50 setting airdefense and airattack at 15. With these high setting the AI's do not AIR attack partisans anymore, with this they must be taken out with ground troops. Also because in the beginning quickly weakening AI's that need help fast panic build militia model 0 and there is no supply support, they are free.

To control how many militia are built upgrade is barbed.

Militia model 1 and 2 are added and billed as Professional Mercenaries. Stats reflect same. (I trust).

Whether this affords AI's to react in time against a human still will depend on how savvy the human is. For just knowing that a 'wounded' AI will mass produce and field dozens of militia inside a certain number of days gives the human great advantage.

Maybe explore a more 'instant' approach and presuppose the Country has in place the emergency call up of militia. So only cost of MP and IC will be the issue and these militia appear in like 2 days?

Nice to 'think' during a posting.

Have to try this it sounds juicy.

Now all the above is fine, however in addition to normal panic build as per the AI instruction available to us modders I believe there is another hard-coded command lurking about.

And this lurking about one triggers with MP starvation commencing. The AI forgoes building any ground troops excepting cheap infantry. The question is, can one sour the soap by barbing the cost in MP for infantry so far as it relates to other ground troops ? If so does it pick another to act as the cheap build if infantry are not found to be cheap. Or again does build infantry only no matter what?

At this point despite some testing I have more questions than answers.

 

[unmerged(42223)]

Maybe explore a more 'instant' approach and presuppose the Country has in place the emergency call up of militia. So only cost of MP and IC will be the issue and these militia appear in like 2 days?

Nice to 'think' during a posting.

Have to try this it sounds juicy.

Well a good idea or not I now remember doing this once and the AI will que in like 99 militia and stay put building this run of militia units until exhausted of MP, IC or resources. And not just one run either, several runs...

So what IC cost is reasonable for a two day run ? 50 IC ? And if 50 IC are not available it will still que it in but just take more time. So if it only has 25 IC available they will come in 4 days (double time). I remember toying with this too. That would solve the problem of them queing in too many for the most part. However short of an event, a dozen militia showing up at once is not probably feasable.

I always favor doing all I can without events. So I am not done experimenting I guess.

What do you think ?

 

[unmerged(42223)]

Holy wonder of wonders (for me anyways)

I just run AI test #60 something beta3 whatever to June 3rd 1930 and checked Fed Germany who is doing very well this time. I was confronted with '10 Convoy Escorts' just got built. And in the que more convoy escorts and transports are que'd, like a run of 11 convoy transports and a run of 3 escorts.

Gee, now if I could just figure what I did to encourage that behavior ??? Besides the obvious settings of course.

In 15 months up to 350 IC and a well mixed army of 119 which is fifth in size worldwide, Austria-Hungary is 161. MP is at 500 and declining with a daily MP flow of 1.51.

For testing purposes I just reset militia model 0 at 50IC in 2 day run and militia model 1 at 25 IC in 1 day run.

For the next test...

 

[unmerged(42223)]

For testing purposes I just reset militia model 0 at 50IC in 2 day run and militia model 1 at 25 IC in 1 day run.

For the next test...

This does not work, it just created a load of militia based empires. However I had failed to post the MP cost ? of those militia, model 0 and model 1. That being MP = 5.

But, when I reset MP cost to 15 with IC cost at 25 and 2 day build run it works. Once the AI recovered it ceased or rather suspended militia runs and went back to the regular que build. I raised Infantry so it would not default so easily to cheap runs of infantry. (where the line is exactly on this I do know ) But also its MP reserves are drained. So it remains important to keep specialized and expensive so the AI (or humans for that matter) not to resort to mass building of whatever seems to be the 'cheap build'.

 

[unmerged(41995)]

Any clue when we might give beta 3 a shake down Yukala?

I know I know, in the words of a certain software guru its ready when its ready....

But.....

Maybe a hint?

 

[unmerged(42223)]

Any clue when we might give beta 3 a shake down Yukala?

I know I know, in the words of a certain software guru its ready when its ready....

But.....

Maybe a hint?

Ya, it is ready when I quit screwing with it.

Actually I am fairly close to packing this puppy up. A couple of days. (I think :rofl: I think... :wacko: ,been a long day... )

 

[unmerged(42223)]

Well, I think I have found the culprit to build que spoiling or spaming of infantry. Max batch commands as there are three of these which means three triggers to 'load' the build que. I have rigged yet another test.

Well the first experiment was set up with 'holes' in it and the AI's drove a long line of Mac Trucks chuck full of Infantry right on through it... :rofl:

 

[unmerged(42223)]

I don't like being pushed down a one size fits all route for units any more than anyone else does. I just find that HOI2 has a very definite "vanilla infantry is the most efficient way of applying IC to a particular task" and this tends to flow into all the Mods including this one. With a risk Mod we are very much in a position of doing anything we like.

Well finally I believe I have broke this infantry everywhere crap !!

And why it is happening. The short of it is that are actually three AI build que not one. They appear as one to the player but the AI feeds in by differing criteria three que which each will favor infantry, thus compounding infantry builds no matter what. More especially as the game wears on and these various build que are triggered one on top the other.

 

[unmerged(42223)]

Beta3 update

Risk Battle Mod Beta3 Update:

I am getting happier.
The AI's are largely building exactly what I want. Out three years so far.

 

[Kanitatlan]

Speaking from the optimising player point of view I have a few comments.

1. Having a high air attack on militia is just asking for trouble. Players will be able to treat them as a specialist air defence unit. I think the only safe thing that can be done with militia versus air is to give them a very high air defence. This will help defend them whilst making sure that there is no impact on surrounding units.

2. My analysis of militia for HOI2 suggests a number of good rules to follow when designing militia units. (in the following CP {combat power} is firepower times org for a group of units. Note that CP should take into account the effects of low toughness/defence).
a) Militia mixed with infantry should achieve higher CP per IC on defence than just infantry.
b) Militia mixed with infantry should achieve higher CP per manpower on defence than just infantry
c) Militia and militia mixed with infantry should NOT achieve higher CP per IC/manpower on offense than pure infantry
d) In HOI2 militia I on its own does not provide higher CP per IC/manpower than other units but Militia II does. I'm not sure what is the best setting but if you have a unit similar to HOI2 Militia II then you should be aware players may use it extensively as a defensive unit.

Point (c) is achieved by having high defence but low toughness. I think this is essential as it is the only way to get the combination of capabilities. Also there is a need to keep the price of militia framed between strict limits so that (a) & (b) can be true without breaching (c). Your stated parameters for Militia II make it truly supreme as cost effective per IC although this is probably no great issue as long as (c) is not breached for manpower.

3. You should be wary of using upgrade costs to downgrade the value of militia. Any player with a combat style like mine will significantly downgrade the value of experience since I always use maximum force to win with minimum casualties (and hence minimum experience).

If you email me a zipped up copy of the current DB folder I can send you back a unit analysis. Do you want to consider a cycle of analysis before finalising Beta 3? I am concerned that the focus on AI unit management may leave holes that players can exploit.

Actually my real worry is that there will be some simple choice of optimal unit reducing the complexity of the game.

 

[unmerged(42223)]

Beta3 Play Testers

Yes, however there are now with you only four active beta tester I am aiming to give this to:

Kanitatian
therev
GeneralHannibal
EvilSanta
VILenin

(if anyone else has a burning desire let me know )

Last hands off AI test took near 24 hours of computer clock time running it out to summer 1938. So you guys must invest a lot of time into playing these beta games and I am very appreciative.

When I post up I will let you know first Kanitation just to save the rest the beta testers their test time. That way their time will be used to even better advantage, once any new needed tweaks are in place.

I have a business trip so even getting it to you Kanitatian is two or three days off.

When I was experimenting with the militia model 0 to inhibit AI air attacks upon them; that air attack setting was an unintended leftover. I am fairly sure that 50 air defense is sufficient deterent to AI air. And I only mean for it to be with militia model 0 anyway.

When you get it I will probably know for certain and will have amended that militia stat. You continue to share very valuable insight and I look forward to your full analsys of all units.

 

[GeneralHannibal]

So we should get it in about a week (by we I mean me, Therev and Evilsanta).

 

[unmerged(42223)]

So we should get it in about a week (by we I mean me, Therev and Evilsanta).

Yes and I am back from business trip but a little tired... it was a two day, one night but lots to do business trip.

 

[unmerged(42223)]

Hi Beta Testers !
And Kanitation waiting for his advance copy.

Again, again so far no matter what I do with the settings the AI que goes Infantry run upon run. This does not reflect any setting I can control.

I am tempted to get more radical.
What do you think ?

I can double the OOB Infantry at game start and nix the AI build que to zero Infantry?

OR and,

Swap the AI build instructions to build Infantry in the mid to late game say five years out.

The infantry stats I believe are fine enough, Kanitatian will know, but this AI doing what I cannot control is very fustrating when the command directly implies that it should behave one way and yet manages to default back to mass Infantry builds so far no matter build que instructions are set up. It goes worse faster with relative set to 'yes' and does it slower set to 'no'.

So I am delayed here and will test a time to two more.