AI for MWiF - France

[Froonp]

Ok. But I do not see how France gets 5 BP in Sep/Oct. Is the CW convoying the French Indo-China resource to Marseilles?

There are 6 resources in Metropolitan France and the 1 Algerian resource can be brought in by rail (through Spain). 3 convoys can bring the 1 Senegal resource up to Brest. With 8 resources delivered, that only leaves 4 French convoys remaining to bring in the 2 remaining Pacific resources.

I've looked at it closer. France starts with 10 CP.
- Senegal RP : 2 CP bring it back to Gibraltar, then by rail into France.
- Algeria RP : Rail into France, through Gribraltar & Spain.
- Indo-China RP : 6 CP bring it back to Marseilles.
- Iraq Oil : 2 CP bring it back to Marseilles.

Those, added to the 6 French RP, and the French produce at 5 BP in 1939 and 7 BP in 1940, and get to save 1 Oil per turn.
Looks like French paradise for me.

Using French CP to bring back RP for production looks like a primary target for me.

Using them as losses takers is not my prefered way of using them. Except if the CW uses CW CP to bring back the French RP, in safe waters : 4 CP French in the Indian Ocean could be swapped with 4 CW CP, then those 4 French CP can be used as losses takers in Bay of Biscay or preferably the Med to replace CW CP.

[Froonp]

ORIGINAL: CBoehm
Early build for France:

"incl. blitzbonus"
1939 (assuming no early Ge attack - otherwise Mil and Inf)
S/O Mech ...then Mil or a ATgun if one avail.
N/D Mil, Mech-d ...then more Mil or pilot and/or fighter depending on inf-gearing

I agree. With 5 BP per turn, in S/O only the MECH gets done, and in N/D 1 MECH DIV + MIL or GARR.
I would not produce more than 1 total gun (if any). They are long to come, expensive, and France already have a lot of them.

1940
J/F Gar, Inf ...then fighter if more bp avail.

With 7 BP available, I'll do same, except I'll do a pilot now, and a FTR next turn, so that they arrive at the same moment.

M/A Mil ...rest LL to CW
M/J Mil or inf, gar (Mil has the added benefit that these are removed in case of Vichy installation, so they, and RES, dont contribute to the Vichy garrison) ...Normally I would try to LL 5 bp or so to CW and build with the rest ...
J/A (assuming France is close to collapse) LL 5-8 bp to CW ...build Mil for rest.

I disagree with any lend lease to CW. First, there will not be any CP for the job, and as France I have something else to do with my time that move all my CP to make the link. Next, I prefer to defend more strongly, and built until the end. You never know how many blunders the Germans will experience, and if you've got no reinforcements, you're making his errors and bad rolls forgaved.

M/A : MIL + INF + FTR2, or MIL + PIL + FTR2
M/J : Same.
GARR can be swapped with MIL if MIL are not available.

"Excl. Blitz bonus"
S/O 39: HQ ...then MIL and pilots
N/D+: Inf, Mil, Fighters ...guns ...ART mostly usefull if NOT playing with fractional odds.

I agree, except that I would not produce more than 1 total gun. They are long to come, expensive, and France already have a lot of them. I'd prefer to build MIL / GARR / INF, and FTRs (with pilots).

Free France
try to build 1-3 good INF...then Fighters & pilots. If playing with Amphrule then try to either build or breakdown to inf-d and mtn-d for invasion purposes, since this is a real addition to CWs invasion capability in 41-42.

Excellent point about French DIV.
In addition of what you said, I love to build the 5-strength Tactical bombers of 1942 & 1943 (LeO 455 & Amiot 355) for FF. I'd only build 1-2 FTR before going for those LND, and then I'd build more FTRs if I get more BP.
Also an HQ is necessary, to reorg the French force, and give supply. This HQ being DeGaulle is not necessary, but if the BP can be gathered, this HQ represent a good deal because it is a decent ARM unit, and a good HQ, better than most of the others.

Later on (42+) trying to lay down a FF TRN or Amph can be a good idea too. If 3 factor HQI is available this can be a good investment too since it allows the FF to do a airimpulse and reorg AC. De Gaulle is IMO only really worth the cost if NOT playing with allied combat friction.

I disagree with FF TRS / AMPH. The Wallies will have plenty of them, and it is unecessary from my point of view to have a FF one. Having 1 is 1 less than necessary to conduct an Invasion, and buildng 2 is far too expensive. I'd prefer to build either land units, 2-3 as you said plus 1 HQ, that will all have an impulse after impulse use, and the most air units I can. Those 5-strength LND plus any number of FF FTRS will come really handy and very usefull for the Wallies.

Quite often it pays to LL a few bps from CW to France in N/D 39 and J/F 40 ...then LL back to CW the following turns - this has often nedded me a transfer of 10-15 bps to CW before France is conquered/Vichy-installed ...

I don't like lending the French BP to CW, and neither do I like the reverse. Both need to build at full possible production, all the time. The CW needs to lay down its AMPH from turn one (unless a France first happens), to have offensive capability soon in 1941 against Italy, or anywhere else, so lending CW BP is not an option for me.

[CBoehm]

Patrice I think we agree on most and I do believe its worth to try to fight for France ...IMO its very important to make sure France doesnt fall early M/J ...and streching it to J/A can be worth not only 10-15 bp, but also some very critical "time" ...should Ge try to go for Spain or Suez ...AND try to make a 41-barb. I do however think that very often if Ge does the slow and safe bet of attacking Bel in M/A or so ...France will have excess bp in M/A untill conquest which can be LL to CW without any major detrimant effect ...LL of 5 bp per turn will still leave enough french bp to build/rebuild Mil and Inf enough to give france a fighting chance and ensure reinforcements are on the track in case Ge should threaten a partial conquest....
 
with FF I too prefer building only airunits and yes the 5 factor TACs are wonderful, ...sometimes IMO it can also pay to build STR with 11-extended move since this can be build quite early and have enough range to threated the most of Ge occupied europe...unlike the CW STR ...(the FF TRN or Amph is more a luxury ...its just nice to have a bit of extra transport capacity for the odd move when CW-US does land or land & combined)

[Shannon V. OKeets]

I am surprised that no one has mentioned building the 3rd French HQ-I for the first turn's builds.

Currently, my plan is to not create a specific build schedule for the French. Instead I will place the emphasis on defending against land attacks. That has the effect of making the value of good defensive units higher. Besides the obvious strength factors for land units, there are bonuses for anti-tank guns, artillery, armor, and HQs (because of their ability to reorganize units). Additionally, tactical air and fighters get bonuses. Offensive chits are considered too (again because they can be used to reorganize HQs and then units). To a lesser degree, AA gets an value increase.

The net effect is that each available unit in the force pool has an effective Combat Value which is then divided by its Build Point cost to get its CV/BP worth.

The second step is to see how long it takes to build each unit. This is compared against when the crisis is expected to occur. For France that is normally M/J 1940, though it might be earlier if Germany uses a France first strategy, or later if Germany goes off conquering other minor countries (e.g., Denmark, Norway, Yugoslavia, Greece). The build schedule attempts to maximize the total CVs of units arriving before or on the crisis turn.

The third step is to allow for gearing. Since there are unlikely to be very many alternatives possible (at least for France early in the game), a straightforward examination and evaluation of all legal build schedules involving units over a certain CV/BP can be performed. The schedule that delivers the highest CV on or before the crises turn 'wins'.

This process will be performed fresh each turn to allow for changing circumstances. Besides the obvious annual additions to the force pool, there are also casualties to be considered, both past and expected future ones. If the crisis has arrived, then the strategic position is one of "Desperate Defense", which gives greater importance to how quickly builds arrive.

To this mix, I expect to provide some randomness. For example, say there are 3 build schedules that deliver total CVs of 20, 18, and 16 on or before the crisis turn. They might be given 60%, 30%, and 10% chance of being selected. If two of them have the same builds for the first turn, then the likelihood of that first turn's build being selected is cumulative. Since this calculation is newly performed each turn, the future turns of the schedule are merely hypothetical.

The advantage of this approach is that the AIO can dynamically assess the value of gaining extra build points from the CW (or giving some away). That's done by simply starting with a different # of BPs and performing the process multiple times, comparing the difference at the end. Likewise, should resources be lost earlier or later than expected, or strategic bombing losses occur, the 'best' build schedule can still be worked out regardless of these changes to the # of expected available BPs.

[Froonp]

ORIGINAL: Shannon V. OKeets
I am surprised that no one has mentioned building the 3rd French HQ-I for the first turn's builds.

Hey, CBoehm did, in his "Excl. Blitz bonus" part
[:D]

"Excl. Blitz bonus"
S/O 39: HQ ...then MIL and pilots

[Froonp]

ORIGINAL: Shannon V. OKeets
Currently, my plan is to not create a specific build schedule for the French. Instead I will place the emphasis on defending against land attacks. That has the effect of making the value of good defensive units higher. Besides the obvious strength factors for land units, there are bonuses for anti-tank guns, artillery, armor, and HQs (because of their ability to reorganize units). Additionally, tactical air and fighters get bonuses. Offensive chits are considered too (again because they can be used to reorganize HQs and then units). To a lesser degree, AA gets an value increase.

The net effect is that each available unit in the force pool has an effective Combat Value which is then divided by its Build Point cost to get its CV/BP worth.

The second step is to see how long it takes to build each unit. This is compared against when the crisis is expected to occur. For France that is normally M/J 1940, though it might be earlier if Germany uses a France first strategy, or later if Germany goes off conquering other minor countries (e.g., Denmark, Norway, Yugoslavia, Greece). The build schedule attempts to maximize the total CVs of units arriving before or on the crisis turn.

The third step is to allow for gearing. Since there are unlikely to be very many alternatives possible (at least for France early in the game), a straightforward examination and evaluation of all legal build schedules involving units over a certain CV/BP can be performed. The schedule that delivers the highest CV on or before the crises turn 'wins'.

This process will be performed fresh each turn to allow for changing circumstances. Besides the obvious annual additions to the force pool, there are also casualties to be considered, both past and expected future ones. If the crisis has arrived, then the strategic position is one of "Desperate Defense", which gives greater importance to how quickly builds arrive.

To this mix, I expect to provide some randomness. For example, say there are 3 build schedules that deliver total CVs of 20, 18, and 16 on or before the crisis turn. They might be given 60%, 30%, and 10% chance of being selected. If two of them have the same builds for the first turn, then the likelihood of that first turn's build being selected is cumulative. Since this calculation is newly performed each turn, the future turns of the schedule are merely hypothetical.

The advantage of this approach is that the AIO can dynamically assess the value of gaining extra build points from the CW (or giving some away). That's done by simply starting with a different # of BPs and performing the process multiple times, comparing the difference at the end. Likewise, should resources be lost earlier or later than expected, or strategic bombing losses occur, the 'best' build schedule can still be worked out regardless of these changes to the # of expected available BPs.

Seems good.
Is it possible to make this "formulae" work, outside of the game, so that we can try it and see what production schedule it makes. Some kind of production simulator ?

I wanted to add that for the USA and maybe Japan, the build schedule would have to have a naval part that calculate on the long run, and do not lightly cancel the many CVs built at any occasion.

[trees]

I think the combat value of the MECH outweighs the combat value of the Pretelat HQ. This would be hard to quantify. With 2d10 blitz bonus, the MECH gets a -2 modifier (in classic wargame terms, a complete odds level shift) in clear terrain and clear weather ... even if it is flipped over. That HQ could use HQ support and get a -1 to an attack (and then flip over, giving the attacker a +2 on any future attack, which are likely), or re-org two land units in a land impulse (and then flip over, giving the attacker a +2 if attacked). The French need a strong double line at all times, and the HQs will have to be part of the second line. It is nice to re-org flipped units and make the Germans flip them again, but I think I'd rather have the MECH anchoring a clear hex than two more re-org points. I suppose it would be possible to build both (big CW loans would help), but I don't think I've ever managed it, and it would play heck with getting the infantry gearing up high for the Mar/Apr and future turns to build both, and Infantry gearing should be weighted heavily. I think it is only a feasible build on the first turn.

I do think using the French to work out AI building routines is an excellent way to develop them. China's decisions are even more straight-forward as well.

[CBoehm]

ORIGINAL: Shannon V. OKeets

I am surprised that no one has mentioned building the 3rd French HQ-I for the first turn's builds.

Currently, my plan is to not create a specific build schedule for the French. Instead I will place the emphasis on defending against land attacks. That has the effect of making the value of good defensive units higher. Besides the obvious strength factors for land units, there are bonuses for anti-tank guns, artillery, armor, and HQs (because of their ability to reorganize units). Additionally, tactical air and fighters get bonuses. Offensive chits are considered too (again because they can be used to reorganize HQs and then units). To a lesser degree, AA gets an value increase.

This is all very good if NOT playing with blitzbonus ...but if playing WITH blitzbonus the Mech and the Mech-div is much much more important than the HQ ...not only defensively but also to have a chance to counterattack! (thus also forcing Ge to play sligthly less agressive with respect to flipping/utilizing breakthroughs...)

Also, ART is very nice ...but given that FR already has some, its IMO ...NOT worth the cost if playing with "fractional odds"

ps:
If you read my legthy post you will notice that I DID suggest building the HQ on turn 1 if NOT playing with the blitzbonus ...

[Incy]

The MECH is very good when playing with blitz bonus, but it also messes up france's gearing very bad.

France will usually have 1 or 2 clear hexes in it's line that can be attacked from 2 hexes. The MECH + the ARM stop these hexes from being almost "auto-ovverruns". The capability to hold 1 (MECH + ARM) or 2 hexes (1 each) at less than "auto-odds" is critical for the defence of france.

The typical german attack in a 2-hexside clear hex will be:
odds: ~+5.5 (not counting air support)
flips: ~+4
HQ-support (defensive + offensive): ~+0-2
Blitz bonus: ~+4
In addition, the para can be added for another ~+1

So without french armor/mech to fill the everpresent "clear gap", odds easily end up at around +14, or at least +12 if something goes wrong (missing a flip, defensive ground support clears, or HQ support isn't used). Such odds almost automatically mean dead french units, and a hole in the line. With french ARM + MECH in the exposed clear hex (+ AT if playing with DIVS), typical odds drop to +8-10, which has at least some chance of failure, and a very good chance of having the units survive to fight another day.

Without some french blitzbonuses in those clear hexes, the german spearhead will just tear into the line when the weather arrives, exposing more clear hexes and gobbling up units in the process. The attack will keep on rolling since the units will likely not flip (and will inevitably be reorged if they do).

To get the MECH into the line in time (which normally means MAR/APR), it must be built right away. The big drawback is of course that the french gearing will be totally ruined (especially if playing without DIVs). The next most needed thing for france after MECH/ARM/AT is just sheer volume, france needs to doublestack most of the line, it needs a fence against italy, and it needs a second line wherever the germans can blitz. 1 INF nov/dec, 2 jan/feb, 3 mar/apr is just not enough soon enough. With 2, 3 & 4 inf type the first 3 turns france will often have enough units to trade blood for time. Defend every hex even with bad odds, and just let the germans gobble 2 units at +12-14 odds each clear impulse (they lack enough armor to take 2 hexes at superodds). Germany will gobble the french army, but it will cost a lot of time until the french are so few they have to give up defending in the open (and have to fall back to the seine line to await the inevitable)

[Shannon V. OKeets]

Incy,

There aren't many infantry units for the French to build. To start there is 1 Mtn (+div), 1 inf div, 4 mil, 1 mot (+ div), and 14 territorials. I do not see the Terr as very attractive builds. Another Inf and Gar arrive in J/F 1940, but that only gets us to 11 if we count the Mtn and Mot and all the divisions. The Mtn and Mot take 3 turns to arrive and are too expensive to build in 1939, when there are only 5 BP available. Also these all cost 2 or 4 BP (until the 3 BP Inf arrives), which makes the 5 BPs hard to spend completely.

So, I do not see how you can build 2 + 3 + 4 infantry types in the first 3 turns. Am I missing something here?

[Shannon V. OKeets]

ORIGINAL: Froonp

ORIGINAL: Shannon V. OKeets
Currently, my plan is to not create a specific build schedule for the French. Instead I will place the emphasis on defending against land attacks. That has the effect of making the value of good defensive units higher. Besides the obvious strength factors for land units, there are bonuses for anti-tank guns, artillery, armor, and HQs (because of their ability to reorganize units). Additionally, tactical air and fighters get bonuses. Offensive chits are considered too (again because they can be used to reorganize HQs and then units). To a lesser degree, AA gets an value increase.

The net effect is that each available unit in the force pool has an effective Combat Value which is then divided by its Build Point cost to get its CV/BP worth.

The second step is to see how long it takes to build each unit. This is compared against when the crisis is expected to occur. For France that is normally M/J 1940, though it might be earlier if Germany uses a France first strategy, or later if Germany goes off conquering other minor countries (e.g., Denmark, Norway, Yugoslavia, Greece). The build schedule attempts to maximize the total CVs of units arriving before or on the crisis turn.

The third step is to allow for gearing. Since there are unlikely to be very many alternatives possible (at least for France early in the game), a straightforward examination and evaluation of all legal build schedules involving units over a certain CV/BP can be performed. The schedule that delivers the highest CV on or before the crises turn 'wins'.

This process will be performed fresh each turn to allow for changing circumstances. Besides the obvious annual additions to the force pool, there are also casualties to be considered, both past and expected future ones. If the crisis has arrived, then the strategic position is one of "Desperate Defense", which gives greater importance to how quickly builds arrive.

To this mix, I expect to provide some randomness. For example, say there are 3 build schedules that deliver total CVs of 20, 18, and 16 on or before the crisis turn. They might be given 60%, 30%, and 10% chance of being selected. If two of them have the same builds for the first turn, then the likelihood of that first turn's build being selected is cumulative. Since this calculation is newly performed each turn, the future turns of the schedule are merely hypothetical.

The advantage of this approach is that the AIO can dynamically assess the value of gaining extra build points from the CW (or giving some away). That's done by simply starting with a different # of BPs and performing the process multiple times, comparing the difference at the end. Likewise, should resources be lost earlier or later than expected, or strategic bombing losses occur, the 'best' build schedule can still be worked out regardless of these changes to the # of expected available BPs.

Seems good.
Is it possible to make this "formulae" work, outside of the game, so that we can try it and see what production schedule it makes. Some kind of production simulator ?

I wanted to add that for the USA and maybe Japan, the build schedule would have to have a naval part that calculate on the long run, and do not lightly cancel the many CVs built at any occasion.

I have several reasons against creating a production simulator.

- It needs a lot of information: optional rules in effect, units in the force pool, units in production already, units on the map (e.g., do we need pilots?, is there an ATR for the Para?), strategic plan for production priorities, relative strength of all armies in the game, theaters of operations demand for unit types, and so on.

- Letting the players know exactly what the AIO is most likely to build in advance gives away too much.

- The time and effort to make it usable by the average player.

The strategic plan for the major naval powers (US, CW, Japan) obviously will have to include trade-offs betweeen air, land, and naval forces. France is much simpler, which is why I am starting here.

[trees]

I didn't get the impression that Patrice meant a production AI as a game feature, I think he meant he hopes the AI production code can be run during development to see what it decides, I think you are actually on the same page on this. It also seems like you are maybe close to seeing what the AI picks out on its own? I could understand not spilling the results though.


Steve you are right that the French Infantry gearing and thus their overall build plan is closely tied to which units are in their force pool. A few notes ... actually they start with their one Infantry division on the map. It is an option to break down a corps at set-up, using up the Motorized division. (sometimes if the Italians aren't strong on the French border I might break down the CAV corps to cover two hexes there, but I've been schooled doing that). But the motorized division isn't worth building anyway, and neither the TERRitorials, as you note. As for complete spending in a turn, I don't think very many people would ever play without Saved Build Points, even though it is an optional rule; production and gearing decisions are hard enough already, I couldn't imagine not using it. Anyway the pace of building the MIL sets up the critical Infantry gearing.

[Shannon V. OKeets]

ORIGINAL: trees

I didn't get the impression that Patrice meant a production AI as a game feature, I think he meant he hopes the AI production code can be run during development to see what it decides, I think you are actually on the same page on this. It also seems like you are maybe close to seeing what the AI picks out on its own? I could understand not spilling the results though.


Steve you are right that the French Infantry gearing and thus their overall build plan is closely tied to which units are in their force pool. A few notes ... actually they start with their one Infantry division on the map. It is an option to break down a corps at set-up, using up the Motorized division. (sometimes if the Italians aren't strong on the French border I might break down the CAV corps to cover two hexes there, but I've been schooled doing that). But the motorized division isn't worth building anyway, and neither the TERRitorials, as you note. As for complete spending in a turn, I don't think very many people would ever play without Saved Build Points, even though it is an optional rule; production and gearing decisions are hard enough already, I couldn't imagine not using it. Anyway the pace of building the MIL sets up the critical Infantry gearing.

You're right about the Infantry div starting on the map. There are only 4 mil in the game (2 are reserve and come in with the other reserve units). I have down as one build possibility for the French: Mtn, 2 Mil, Mil + Inf + Gar, Mil + ??? for the first 4 turns.

[Shannon V. OKeets]

I meant to add:

As currently coded, breaking down an infantry corps doesn't make it available to be built again. Instead it goes into a broken down pool. Only after 2 divisions get destroyed, does the corps go back into the force pool.

[Froonp]

You're right about the Infantry div starting on the map. There are only 4 mil in the game (2 are reserve and come in with the other reserve units). I have down as one build possibility for the French: Mtn, 2 Mil, Mil + Inf + Gar, Mil + ??? for the first 4 turns.

Don't forget the HQ-I, if possible. I'd prefer it rather than the MTN.

Also, if playing with blitz bonus, I think that the MECH will pay off.

[Shannon V. OKeets]

ORIGINAL: Froonp

You're right about the Infantry div starting on the map. There are only 4 mil in the game (2 are reserve and come in with the other reserve units). I have down as one build possibility for the French: Mtn, 2 Mil, Mil + Inf + Gar, Mil + ??? for the first 4 turns.

Don't forget the HQ-I, if possible. I'd prefer it rather than the MTN.

Also, if playing with blitz bonus, I think that the MECH will pay off.

Does the HQ-I count as an Inf type? I built the Mtn, so I could build 2 Mil in the next turn.

[Froonp]

ORIGINAL: Shannon V. OKeets

ORIGINAL: Froonp

You're right about the Infantry div starting on the map. There are only 4 mil in the game (2 are reserve and come in with the other reserve units). I have down as one build possibility for the French: Mtn, 2 Mil, Mil + Inf + Gar, Mil + ??? for the first 4 turns.

Don't forget the HQ-I, if possible. I'd prefer it rather than the MTN.

Also, if playing with blitz bonus, I think that the MECH will pay off.

Does the HQ-I count as an Inf type? I built the Mtn, so I could build 2 Mil in the next turn.

Sure, the HQ-I count as an INF type for gearing, as an HQ-A count as an ARM for gearing.

[trees]

I'm used to playing with the Senegal 5-3 MIL from the Politics in Flames counter sheet, leaving me four to build (5 total originally, 2 REServes, plus the Senegalese). Because of the HQ's on that sheet I don't forsee very many people not using those counters, especially on a computer. What to build in Mar/Apr once the gearing is high is a good question. If th sun shines that turn you won't have a problem selecting units, but if it doesn't you've just run into a great question for the AI to decide...when is gearing more important than building the perfect unit arriving quickly with good combat values? This is one of those times if you ask me.

[trees]

Oh and here is another wrinkle ... I always keep the Paris MIL in the hex most likely to be attacked as it is the cheapest 5 strenght unit and arrives the fastest and closest to the front. Of course the Stukas can rather limit French deployment options at times though by flipping everything in sight.

[CBoehm]

ORIGINAL: trees

Oh and here is another wrinkle ... I always keep the Paris MIL in the hex most likely to be attacked as it is the cheapest 5 strenght unit and arrives the fastest and closest to the front. Of course the Stukas can rather limit French deployment options at times though by flipping everything in sight.

I do this too [:D] - another good tactic ...is to try to move your units around so eventually the italian border is held by MIL & RES corps ...so when FR falls by Vichy your regular army is killed while your MIL has survived in the south (instead of the other way around) ...Since MIL and RES dont contribute to the size of the Vichy garrision.