How will AI attacks be performed?

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

How will AI attacks be performed?

View related threads: (in this forum | in all forums)

Logged in as: Guest

Users viewing this topic: none

Printable Version

All Forums >> [New Releases from Matrix Games] >> World in Flames >> How will AI attacks be performed?

Page: [1]

Message

<< Older Topic Newer Topic >>

How will AI attacks be performed? - 6/25/2007 10:43:53 PM

Peter Stauffenberg

Posts: 403
Joined: 2/24/2006
From: Oslo, Norway
Status: offline

I wonder how the AI will move units to prepare for attacks and then select the odds for fighting. Will the AI prefer to make high-odds attacks to minimize bad dice rolls?

Or will the AI take some risks with low-odds (or rather medium-odds) attacks sometimes too? Especially when fighting for a city, resource, factory or a hex to regain supply for own units or cut supply for enemy units. But early in a clear weather turn it can be stupid to make medium-odds attacks because you might risk getting flipped and then you can't move in the next impulse unless an HQ is nearby. So the AI has to calculate the chance of the turn ending and if it's likely there will be more impulses he may have to be more conservative and not risk being flipped (except when the gain is cutting several enemy units, capturing a factory etc.).

I think selecting the number of attacks and odds for them is one of the most difficult parts of WIF. If you play too conservatively and only select the best odds then you won't progress very far. But if you take too high risks then you may have bad dice rolls and lose so many units the enemy can counter-attack and create a lot of problems.

A balance between risk taking and playing safe is probably best. But the Germans in 1941 Barbarossa have to be willing to take some risky attacks because the gain is so high (capturing factories, destroying HQs, surrounding lots of units etc.).

It's also not very easy to calculate when a defending player can go on the offensive. E. g. if an enemy armor spearhead becomes flipped and it can be attacked with good odds then it may be a good idea to attack even at the cost of being counter attacked the next impulse by the enemy.

The Soviet player has to be very careful especially in 1941 (if Barbarossa starts then). Good counter attack opportunities may arise, but it may be smarter to retreat the units instead and form a defense like behind rivers, protecting cities etc. How will the AI do this? Will the AI be "helped" to find good defense lines in e. g. Russia by having them predefined (e. g. in contiguous lines of hexes) or will the AI be able to find good defense lines by analyzing the terrain and where the friendly units are?

The Soviet player will get counter attack opportunities during the Winter of 1941 and must use them to his best abilities. But he has to switch to a new defense line again in time for the good weather in 1942.

I also wonder how the AI will calculate whether a player can be on the offensive or has to be on the defensive at a front. Will the AI add combat factors on the front and look at the odds. E. g. if the odds in favor is 3-1 or better the he can go on the offense many places, if the odds is > 1-1, but less than 3-1 then he can make some selected attacks. If the odds is < 1-1, but higher than 1-3 then he can make an attack or two if the gain is good, but retreat or defend behind rivers or in good terrain is the priority. If the odds is less than 1-3 then the defender should make major retreat if there is enough terrain to give away and only defend in cities and behind river lines if it's possible to form.

How will the AI use ZOC to form defense lines or break defense lines? Using ZOC in an intelligent way is vital to maintain a front. Another aspect is the ability to use a second defense line so the attacking power won't create a big how if one front hex is destroyed. This is particularly important in the phase where the defender has less combat factors than the attacker, but not so low that he has to make a general retreat. If he wants to prevent major breakthroughs then he has to defend in depth. He will lose more units because he doesn't stack them, but he won't lose so much ground. Stacking the defenders and only defending every other hex could create a big gap in the front if this hex is destroyed.

How to use the airpower correctly is also important for the AI to do well. E. g. which hexes to ground strike, which hexes to bomb strategically, which air assaults to escort etc.

Also typical in some strategic WW2 games with AI is that the human player can invade hexes not well defended and create a lot of havoc. E. g. the British player against the German player. This is possible because the AI can't foresee the consequences of not defending the key ports etc. So maybe the AI should rely upon a table made for every major power which tells the AI which tells the priority of which ports to defend and when it's necessary to defend them. The AI has to be able to form a strategic reserve that can be used wherever it's needed. E. g. if the British player makes an assault to a French port then there should be German units not so far away that can destroy these units. The German AI would want to recreate a Dieppe result and not an Overlord result.

I think it's pretty hard for the AI to calculate so many vital things pretty correctly so maybe the AI should be "helped" by guidelines from experienced players that the AI can check when a condition arrives. E. g. like which ports to defend in France, Germany, Italy etc. Where to place reserve units (Paris, Hamburg, Vienna etc.) and what type. I know that this is impossible for parts of the map like America, Africa and Asia. But in Europe some guidelines could help the AI making better results.

One thing we may have to think about is what kind of typical exploits can a human player use against the AI. I'm thinking about abusing special rules to the human gets a benefit that no human opponent would give them, invading places that no human player would allow them because they would see it coming before it happens. E. g. a human player would analyze the production made by the opponent and look for the location of enemy units to foresee invasion etc.

Most WW2 AI's I've seen are too passive and are never able to preempt my offensive plans. I guess it's maybe impossible to make a great AI for a game like MWIF, but I would prefer the computer AI to be active even though it makes some mistakes sometimes rather than having a passive AI that will only make very high-odds attacks and defend whenever possible. I want the AI to surprise me sometimes with attacks I didn't foresee etc. That means I have to be more careful and garrison ports, have a strategic reserve etc. myself too.

Post #: 1

RE: How will AI attacks be performed? - 6/26/2007 3:22:25 AM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

Most of this falls into the category of what I think of as the value of a hex.

It is easy for the AIO to calculate odds for all types of attacks and not too difficult to predict outcomes (the hardest are for the naval combats because of surprise and the 3 difference combat tables).

Overall strength is trivial for the AIO to determine and from that it can decide whether a major power is on the offensive or defensive. The strategic plans will offer different choices to the offensive major powers so they can choose where to attack (e.g., for Germany: Lowlands, Balkans, USSR, Spain, GIbraltar, Suez, SeaLion, etc.). Similarly, the CW, which starts on the defensive, can choose where to defend. I.e., which hexes are most important and the level of paranoia appropriate for defending them before allocating resources to less important hexes (e.g., Gibraltar, Malta, Suez, Great Britain).

Air: I have the air war strategies, operational plans, and tactics fairly well developed. I believe the same code/rules can be used by all the major powers. the use of tactical air units will be up to the Field Marshals to employ where ever they see the benefit the highest. Teh Admiralty will determine when to use naval air units.

Land: Calculating odds permits the AIO to identify weak points when attacking and defending. And from there the AIO can assess a line's overall strength/weakness. With that in hand, the relative strength of two lines can be compared and the stronger one chosen (when defending). I am building into its AIO's strategic plan for each major power the vital hexes for defense and offense (these include vicotry cities obviously). Working from the vital hexes the AIO can build line segments that either: (1) connect the vital hexes, or (2) form a line which protects the vital hexes from capture. For example, every hex in the Maginot Line is a vital hex for France while the segment of the defensive line that lies in Belgium is defending vital hexes to its rear. I am propagating the value of the vital hexes forward to the lines that are defending them. So, the portion of the line that is defending Paris is given a value derived from the value for Paris. The most difficult problem here is to measure the value of holding a weaker line that is farther forward or a stronger line that permits the attacker to get closer to the vital hexes that are being defended. A similar problem is the use of hero cities, versus full evacuation, to minimize losses.

Naval: Here I am coming around to the opinion that sea areas can be treated like hexes - at least in the abstract. A line of sea areas running north to south can be defended by the the USA or Japan. The problem is when to abandon a forward line to reduce losses, and whether it is worth leaving hero islands behind. The attacker has a similar problem when dealing with defending units that get cut off/left behind: whether to devote time and resources to their total annihilation or to bypass/screen them.

As to aggressive versus passive play, I am not too concerned. There are good things that should always be done, bad things that should never be done, and gray areas where random numbers can make the AIO more difficult to predict. This applies directly to your question about what odds to require for making a land attack. The AIO will worry about time pressure and be aware of impending reinforcements, not only several turns ahead, but several years ahead (e.g., Germany will not wait until 1943 to launch Barabrossa).

The thing that worries me the most is when things get chaotic with units from both sides trapped behind the lines, out of supply. Should a large hole appear in a line and the attacker advance into it with fast units, then later reinforcements arrive for the defendrer and instead of a line the battleground looks like splotches of units. In that situation, the AIO could very easily assign importance to something that isn't. Invasions and paradrops are another way to create these positions. Partisans can too.

_____________________________

Steve

Perfection is an elusive goal.

(in reply to Peter Stauffenberg)

Post #: 2

RE: How will AI attacks be performed? - 6/26/2007 3:40:35 AM

SurrenderMonkey

Posts: 124
Joined: 10/5/2005
Status: offline

quote:

ORIGINAL: Shannon V. OKeets
There are good things that should always be done, bad things that should never be done, and gray areas where random numbers can make the AIO more difficult to predict.

This randomization is critically important. Unpredictability goes a very, very long way with an AI. If I am Germany, for example, I need to believe that the UK might try something in France or the Balkans - the simple threat forces me to divert resources, maintain reserves, etc. It's the difference between playability and unplayability, IMO.

_____________________________

Wise Men Still Seek Him

(in reply to Shannon V. OKeets)

Post #: 3

RE: How will AI attacks be performed? - 6/26/2007 6:15:11 AM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

quote:

ORIGINAL: SurrenderMonkey

quote:

ORIGINAL: Shannon V. OKeets
There are good things that should always be done, bad things that should never be done, and gray areas where random numbers can make the AIO more difficult to predict.

This randomization is critically important. Unpredictability goes a very, very long way with an AI. If I am Germany, for example, I need to believe that the UK might try something in France or the Balkans - the simple threat forces me to divert resources, maintain reserves, etc. It's the difference between playability and unplayability, IMO.

Yes.

_____________________________

Steve

Perfection is an elusive goal.

(in reply to SurrenderMonkey)

Post #: 4

RE: How will AI attacks be performed? - 6/26/2007 12:23:35 PM

Peter Stauffenberg

Posts: 403
Joined: 2/24/2006
From: Oslo, Norway
Status: offline

quote:

ORIGINAL: Shannon V. OKeets
Most of this falls into the category of what I think of as the value of a hex.
<snip>

Great. It looks like you have thought of everything. I look forward to seeing the MWIF AI in action when the game is released.

I also agree with the other poster that the AI has to be unpredictable. The human player show know that the computer player excels in finding weak spots and may exploit it. So if you don't defend your cities, ports etc. then you should blame yourself if the AI launches a major offensive against you.

I also wonder if the AI is programmed in a way that it decides upon a strategic plan and sticks to this plan. I'm thinking about priorities for unit production, placement of units and when to launch an invasion. E. g. the German AI will probably always try to take Poland and then France. But he should know already in 1939 or early 1940 whether he should produce airplanes and naval units for Sea Lion, produce lots of land units for Barbarossa or maybe something else like invading Spain, Egypt and later India. It's important that if the AI has decided upon a Barbarossa 1941 offensive that he will continue to produce land units and move units to the front every turn.

But the AI should be able to change his strategic plans too. E. g. if the British player makes an invasion in France in May Mar/Apr 1941 (just before Barbarossa) then the German player should make an analysis of whether he can kick those units out of France with the reserves in the west or he must postpone the Barbarossa offensive to 1942 and move units to the west to regain control.

I also wonder if the different AI players would cooperate with their strategic plans. E. g. Japan and Germany could cooperate so Japan attacks Russia (Vladivostok etc.) in 1941 so the Russians can't sent reinforcements to Moscow to save Russia from the German 1941 Barbarossa onslaught. The Americans and British must be able to cooperate a lot too (in North Africa, Italy, France and even in Asia). Maybe it's possible for the AI to see the British and US forces as a single side after USA enters the war.

How to make the island hopping for USA in the Pacific is not easy to program into an AI either. Air ranges for bombers and fighters are important when selecting an invasion target. The AI has to see several steps ahead. E. g. if I acquire these islands then I will get air range to some new islands that can be invaded eventually leading to some vital territory being captured like New Guinea, Philippines, Borneo etc.

I think it's a good idea to see the sea areas as "hexes" for analyzing where to move the naval units.

It seems to me that MWIF will get a very nice AI. I guess the AI needs a lot of development and testing. So I think it will be awhile until I will have MWIF installed on my computer. I envy the beta testers who will be able to enjoy MWIF with AI already this year.

(in reply to Shannon V. OKeets)

Post #: 5

RE: How will AI attacks be performed? - 6/26/2007 12:40:42 PM

Neilster

Posts: 2890
Joined: 10/27/2003
From: Hobart, Tasmania, Australia
Status: offline

quote:

ORIGINAL: Borger Borgersen

quote:

ORIGINAL: Shannon V. OKeets
Most of this falls into the category of what I think of as the value of a hex.
<snip>

Great. It looks like you have thought of everything. I look forward to seeing the MWIF AI in action when the game is released.

I also agree with the other poster that the AI has to be unpredictable. The human player show know that the computer player excels in finding weak spots and may exploit it. So if you don't defend your cities, ports etc. then you should blame yourself if the AI launches a major offensive against you.

I also wonder if the AI is programmed in a way that it decides upon a strategic plan and sticks to this plan. I'm thinking about priorities for unit production, placement of units and when to launch an invasion. E. g. the German AI will probably always try to take Poland and then France. But he should know already in 1939 or early 1940 whether he should produce airplanes and naval units for Sea Lion, produce lots of land units for Barbarossa or maybe something else like invading Spain, Egypt and later India. It's important that if the AI has decided upon a Barbarossa 1941 offensive that he will continue to produce land units and move units to the front every turn.

But the AI should be able to change his strategic plans too. E. g. if the British player makes an invasion in France in May Mar/Apr 1941 (just before Barbarossa) then the German player should make an analysis of whether he can kick those units out of France with the reserves in the west or he must postpone the Barbarossa offensive to 1942 and move units to the west to regain control.

I also wonder if the different AI players would cooperate with their strategic plans. E. g. Japan and Germany could cooperate so Japan attacks Russia (Vladivostok etc.) in 1941 so the Russians can't sent reinforcements to Moscow to save Russia from the German 1941 Barbarossa onslaught. The Americans and British must be able to cooperate a lot too (in North Africa, Italy, France and even in Asia). Maybe it's possible for the AI to see the British and US forces as a single side after USA enters the war.

How to make the island hopping for USA in the Pacific is not easy to program into an AI either. Air ranges for bombers and fighters are important when selecting an invasion target. The AI has to see several steps ahead. E. g. if I acquire these islands then I will get air range to some new islands that can be invaded eventually leading to some vital territory being captured like New Guinea, Philippines, Borneo etc.

I think it's a good idea to see the sea areas as "hexes" for analyzing where to move the naval units.

It seems to me that MWIF will get a very nice AI. I guess the AI needs a lot of development and testing. So I think it will be awhile until I will have MWIF installed on my computer. I envy the beta testers who will be able to enjoy MWIF with AI already this year.

The AI will analyse the initial position before deciding the initial setup and strategic plan for the Axis. Production priorities are part of this planning. The strategic plan can change if it isn't going well. Strategic cooperation will be very important to the AI. A lot of work has been done on the topics you've raised. Much of it is in the individual power AI threads.

Cheers, Neilster

(in reply to Peter Stauffenberg)

Post #: 6

RE: How will AI attacks be performed? - 6/26/2007 8:53:55 PM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

Whilke I was in Europe I did a lot of work on the AI and have evolved a standard, one-size-fits-all, structure for AIO strategic plans. I am still typing in my hand-written notes/edits from that trip (it is a low priority now and the AIO is scheduled for August/September). Once I get one of the major powers' strategic plans tidied up, I'll post it for comments.

By the way, the forum threads for the different AIO strategic plans drifted to other countries from time to time so when I edit the CW plan I end up transfrerring some of the posts over to the USSR or Italy plans, for example. Once I get all my edits done, I want to revisit each of the AIO strategic plans and implement the standardization I have developed - so don't hold your breath on seeing this soon (version 4.00 is a much higher priority).

_____________________________

Steve

Perfection is an elusive goal.

(in reply to Peter Stauffenberg)

Post #: 7

RE: How will AI attacks be performed? - 7/18/2007 4:51:55 PM

fallgelb

Posts: 28
Joined: 5/8/2007
Status: offline

IMO one major problem would be to get the fluid situation in Russia in good weather under control. The Russians have not enough units to built a stable front from Leningrad to the black sea vs. a 1941 Barbarossa. I defend with a front from Smolensk to the Black sea and with a strongpoint defense (you call it hero cities) between Leningrad and the frontier.
For example:
1st problem: a defensive Russian AI must use the own units as effective as possible, it is better to have a double depth front in the south as a two unit front from north to south
2nd Problem: a offensive german AI have to decide what focus (is there a military term for "schwerpunkt?") it projects. It could go for Leningrad, Smolensk or try to pincer the dnejpr line. And to make it worse, you should have an initial plan (Build Plan and set-up) and be able to alter it dependend on russian unit deployment
3rd Problem: Not withstanding the global Situation and the last combat results, in Russia in snow everything is different and the Rusian AI should attack with white prints as an human player would do (if only to get some GBA)

Not an easy task i suppose

(in reply to Shannon V. OKeets)

Post #: 8

RE: How will AI attacks be performed? - 7/18/2007 7:00:39 PM

Neilster

Posts: 2890
Joined: 10/27/2003
From: Hobart, Tasmania, Australia
Status: offline

quote:

...is there a military term for "schwerpunkt?"

Yep! Centre-of-gravity

I'm pretty sure the points you raise are being addressed. A lot of this stuff was discussed ages ago and some good ideas were developed.

Cheers, Neilster

(in reply to fallgelb)

Post #: 9

RE: How will AI attacks be performed? - 7/29/2007 9:45:49 PM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

I thought about creating a new thread for this question, but the title for this one is quite appropriate. I apologize in advance if this question seems to take the thread in a new direction.

I started work on planning invasions for the AIO and decided it is a question appropriate for resolution by forum members. If you guys can decide how these should be done, then that will save me some time.

As I see it there are several invasion pieces to an invasion:
1 - Softening up with ground strikes made by land or carrier based air

2 - Loading invading units
Infantry corps/army on AMPH or TRS (depending on optional rules)
Marine corps on TRS or AMPH
Division on SCS or TRS/AMPH (depending on optional rules)

3 - Deployment of invasion fleet to sea area

4 - Invasion by land units

5 - Bombardment support by fleet

6 - Ground support by land or carrier based air.

Then on the defensive side there is the calculation of the strength of the Notional unit etc..

Finally, there is the question of what attack odds are acceptable for performing an invasion.

What I would like to do is be able to create invasion groups (land, air, and naval units) for attacking: Malta, Gibraltar, Marseilles, a clear hex behind the CW front line in Egypt, a coastal hex in Syria or Greece, and so on. That is just for the Italians in the Med. Similar fleets are needed by the japanese in the Pacific, the Germans if they invade the United Kingdom, and the Allies when storming Fortress Europe.

If I have invasion groups well defined for taking on different levels of resistance, them I can have the AIO build the necessary units for their composition and assemble them in ports prior to the invasion.

Finally, defending against invasions can then be worked out by checking for what invasion groups the enemy is capable of putting together (immediately, later in the turn, and in the next few turns to come).

Clearly, all of this is vital to the AIO making decisions (especially for Italy and Japan).

===
Oh, and something similar is needed for paradrops.

_____________________________

Steve

Perfection is an elusive goal.

(in reply to Neilster)

Post #: 10

RE: How will AI attacks be performed? - 7/30/2007 9:56:03 AM

Froonp

Posts: 7995
Joined: 10/21/2003
From: Marseilles, France
Status: offline

You can also add, between 1 and 2 for island invasion, cutting the enemy's supply, by using X number of other fleets who will be charged on clearing their area of enemy supply ships, X being equal to the number of sea area that are around the invasion point minus 1 because there will be the invasion fleet here.

They can also as a second mission be charged to clear the area of any shore bombardment factor the enemy may have.

(in reply to Shannon V. OKeets)

Post #: 11

RE: How will AI attacks be performed? - 7/30/2007 10:10:29 AM

christo

Posts: 99
Joined: 11/24/2005
From: adelaide, australia
Status: offline

Sorry if it is just the computer ignorant part of me, but what actually is the question here?
I totally agree with Patrice about isolating the invasion hex (be this by naval units or land units incl paras).

With regards to the question of the attack odds. This depends on the penalty that you are willing to pay for any future lost oppertunaties vs the success of the operation. Anattack on Gibralter is a generally an immediate success or failure with significant rewards, thus people will be prepared for a 10% success rate. An attack on Fortress Europa on the other hand requires significantly better odds (although this depends on a tactical vs strategic operation and the ability to rebuild the lost paras/ marines from prebuilt para/ marine divisions). I suppose this all gets back to the "hex value" that you are assigning to the target.
With regards to "putting the invasion group a clear hex behind the CW front line in Egypt" how does this work?

quote:

ORIGINAL: Shannon V. OKeets

I thought about creating a new thread for this question, but the title for this one is quite appropriate. I apologize in advance if this question seems to take the thread in a new direction.

I started work on planning invasions for the AIO and decided it is a question appropriate for resolution by forum members. If you guys can decide how these should be done, then that will save me some time.

As I see it there are several invasion pieces to an invasion:
1 - Softening up with ground strikes made by land or carrier based air

2 - Loading invading units
Infantry corps/army on AMPH or TRS (depending on optional rules)
Marine corps on TRS or AMPH
Division on SCS or TRS/AMPH (depending on optional rules)

3 - Deployment of invasion fleet to sea area

4 - Invasion by land units

5 - Bombardment support by fleet

6 - Ground support by land or carrier based air.

Then on the defensive side there is the calculation of the strength of the Notional unit etc..

Finally, there is the question of what attack odds are acceptable for performing an invasion.

What I would like to do is be able to create invasion groups (land, air, and naval units) for attacking: Malta, Gibraltar, Marseilles, a clear hex behind the CW front line in Egypt, a coastal hex in Syria or Greece, and so on. That is just for the Italians in the Med. Similar fleets are needed by the japanese in the Pacific, the Germans if they invade the United Kingdom, and the Allies when storming Fortress Europe.

If I have invasion groups well defined for taking on different levels of resistance, them I can have the AIO build the necessary units for their composition and assemble them in ports prior to the invasion.

Finally, defending against invasions can then be worked out by checking for what invasion groups the enemy is capable of putting together (immediately, later in the turn, and in the next few turns to come).

Clearly, all of this is vital to the AIO making decisions (especially for Italy and Japan).

===
Oh, and something similar is needed for paradrops.

(in reply to Shannon V. OKeets)

Post #: 12

RE: How will AI attacks be performed? - 7/30/2007 1:09:57 PM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

Ok, 10% as an acceptable success rate for Gibraltar is a starting point.

Do you want all of the Japanese surprise attacks in the Pacific to be guaranteed (100%)? Or are you willing to accept that some of them fail? What %?

Does this mean that two land units are always required for an invasion or can a marine unit with bombardment and air support be sufficient? How weak does the defending force have to be? When do you attack an empty adjacent hex to set up a normal land attack the next turn on the true target hex?

What I want to decide (so the AIO can decide) is what constitutes a good (or perfect) invasion force? Forum members have stated that under some circumstances building 2 AMPHs is excessive.

Other posts have stated that unlimited breakdown into divisions gives the Japanese too many units with which to invade. How would that work?

I could do all these calculations myself. For example, the Italians do not have any 9-4 infantry, so their strongest infantry division will be 2 factors. Halved for invasion makes it worth 1, so tripled (naval & air support) makes it 3. That division can be placed on an SCS which gives it enough range to reached any place in the Med without suffering penalties for invading from lower boxes. Three such units would generate 9 attack point which could take out a 'clear' hex (no city or enemy ZOC).

If the AIO doesn't build the right units, it won't be able to invade. If it spends lavishly on marines, ATR, paratroopers, etc., then other branches of the service will go without.

If the AIO can't correctly evaluate the threat of an invasion, then it will over/under defend coastal hexes, perhaps retreating from a danger that doesn't exist.

As for the attack behind a CW force in North Africa, I had envisioned cutting off a HQ from its rail link back to supply, or possible cutting off forward elements from their link back to an HQ.

_____________________________

Steve

Perfection is an elusive goal.

(in reply to christo)

Post #: 13

RE: How will AI attacks be performed? - 7/30/2007 8:37:30 PM

composer99

Posts: 2923
Joined: 6/6/2005
From: Ottawa, Canada
Status: offline

Landing behind the CW lines in Egypt can also be a way of out-flanking them or forcing them to split their forces up (fatally, the Axis hope) to face a burgeoning threat in the rear.

In general, I would definitely proceed with an invasion if I can get any attack that, on an average roll of the combat dice, wins me the hex. +12 or better, on an average roll, wins the day without the invading units flipping, so I usually optimize towards that. +13 is even better because then the dreaded 14 cannot come up.

To get those magic numbers, you need 6 attacking factors for every factor of notional plus defending units, plus or minus factors depending on other combat modifiers (flipped units and so forth). Generally, a marine/para corps, another corps, and a div gets you about 9 factors; add in shore bombardment and ground support and you can get up to 27. So taking out a 4-5 factor defence is fairly straightforward.

How much "external" support (ground-striking adjacent or nearby enemy units near the intended target, cutting off supply, and so forth) will be required depends on the circumstances. Invading France? Try to flip out as much ARM/MECH or HQs in the area so you don't get blasted back into the sea. Invading a stacked Truk? Isolate it so it's weaker (and there won't be a notional).

For particularly key invasions (Tokyo springs to mind), be sure to use an O-chit. The massive increase in land factors allows for similar increase in air/shore bombardment support.

_____________________________

~ Composer99

(in reply to Shannon V. OKeets)

Post #: 14

RE: How will AI attacks be performed? - 7/30/2007 11:40:32 PM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

quote:

ORIGINAL: composer99

Landing behind the CW lines in Egypt can also be a way of out-flanking them or forcing them to split their forces up (fatally, the Axis hope) to face a burgeoning threat in the rear.

In general, I would definitely proceed with an invasion if I can get any attack that, on an average roll of the combat dice, wins me the hex. +12 or better, on an average roll, wins the day without the invading units flipping, so I usually optimize towards that. +13 is even better because then the dreaded 14 cannot come up.

To get those magic numbers, you need 6 attacking factors for every factor of notional plus defending units, plus or minus factors depending on other combat modifiers (flipped units and so forth). Generally, a marine/para corps, another corps, and a div gets you about 9 factors; add in shore bombardment and ground support and you can get up to 27. So taking out a 4-5 factor defence is fairly straightforward.

How much "external" support (ground-striking adjacent or nearby enemy units near the intended target, cutting off supply, and so forth) will be required depends on the circumstances. Invading France? Try to flip out as much ARM/MECH or HQs in the area so you don't get blasted back into the sea. Invading a stacked Truk? Isolate it so it's weaker (and there won't be a notional).

For particularly key invasions (Tokyo springs to mind), be sure to use an O-chit. The massive increase in land factors allows for similar increase in air/shore bombardment support.

Very helpful, thanks. I am not sure I understood what you meant by using the Offensive chit though. Is that simply to do a super combined and have a lot of air, land, and naval units available for several simultaneous invasions?

_____________________________

Steve

Perfection is an elusive goal.

(in reply to composer99)

Post #: 15

RE: How will AI attacks be performed? - 7/30/2007 11:56:23 PM

Froonp

Posts: 7995
Joined: 10/21/2003
From: Marseilles, France
Status: offline

quote:

Finally, there is the question of what attack odds are acceptable for performing an invasion.

The question here is best answered by wondering what losses are acceptable in the event of a possible counter-attack. What can the enemy mobilize against my beachhead to put me back to sea. The AIO should check the possible counterattacks, taking into consideration that the enemy may be able to supply troops that are not at the moment of invasion.
Also, the question can be answered by wondering what mobility I want for the next impulses. Do I need to be face-up for exploiting the terrain, as when taking Oran and racing toward Morocco, or do I just need to hold the hex for conquest conditions to come ?

quote:

What I would like to do is be able to create invasion groups (land, air, and naval units) for attacking: Malta, Gibraltar, Marseilles, a clear hex behind the CW front line in Egypt, a coastal hex in Syria or Greece, and so on. That is just for the Italians in the Med. Similar fleets are needed by the japanese in the Pacific, the Germans if they invade the United Kingdom, and the Allies when storming Fortress Europe.

Land : As powerful as it can be, considering the attack odds question's answer. Not necessarily build out the MAR & PARA, but keeping DIVs in places where they can load in SCS and then invade is a good thing. Invading a coastal hex in the Med, outside of France and mountain hexes is fairly trivial to achieve with 3 INF type DIV. Invading Malta generaly needs 2 corps & a DIV, whatever the corps but invasion specialized are better (MAR, PARA). Invading Gibraltar is the same, with the survivability of the invading units being near 0% and the success chances close to 0% too.

Air : Air for domination of the sea is crutial if the invasion is to be useful. You need NAVs (the Italian NAV2 are great here), long ranged FTRs (Range 4 or more -- Italy has a few, and Germany can loan them a couple of extraordinary Fw190). Speaking of the Italians, ground striking the enemy is generaly not an issue as most invasions will be against empty hexes. Now for invading Malta, a couple of bombers are a minimum. A friend of mine has a good rule of thumb : To be "sure" to ground strike everything in an hex (i.e. to give yourself the best chances to succeed), the sum of the TAC factor of your bombers should be 10 or more. So, 3 x 3 strength bombers are OK, 2 x 5 Stukas too, etc... This is just a rule of thumb, but it is quite a good one. For invading Gibraltar, the same rule of thumb can be used, but having bases from them to take of is harder and need more planning (taking Morocco, and Algeria before). Invading Marseilles is not worth the calculation, as you would spend years to build the necessary material (MAR, AMPH and/or PARA) if Marseille is garrisoned, and you'd better attack from Land.

Naval Units : Italy starts with a very good navy, but should in nearly always finish the Roma BBs. They are the best in town, and great value for the cost. The invader should always try to have sufficient shore bombardement for 3 uses : 1) Initial invasion. 2) possible immediate counterattack. 3) possible next turn initial impulse counterattack (taking into consideration that ships will have to drift 1 section lower). Having solid supply ships is also a necessity here, and keeping it at sea under protection for next turn is a concern too. Having a second for replacing the first next turn is a good option to try to have. As the Air units, the naval units are also here to control the seas, but to a lesser extend. They are here for providing supply (if progress want to be made, and if being put back to the sea is not wanted) and shore bombardement. The fleet (TRS & CP included) should also try to compel with the "magic numbers" for a fleet creation, that are : 4 ships, 7 ships, 11 ships, 16 ships, 22 ships, 29 ships. Only the bare minimum should be sailed to satisfy either the 3 shore bombardement needs or only the 2 first. Example, if the invasion force has a strength of 5 (as 3 DIVs have), 2,5 points should be here for the initial invasion, 5 for the possible counterattack, and 5 more for a possible first impulse counterattack, but the last 5 are from the same ships of the 7.5 first ones who have drifted 1 section. 8 shore bombardement wold be sufficient. Putting more ships can then only be a solution to help those shore bombardement ships to survive the enemy possible naval units reaction.

Well, discussing the Italians invasion possibilities is a little frustrating as they have close to none, and have very little way to have more, except by building the necessary assets from scratch, which is a long and costly built plan. The goal has to be worth it, and generaly, except for Gib, it is not. Invading any non occupied shore of the Med is quite trivial with 3 DIVs, and Invading Malta generaly is done when the CW leave it a little weak. Discussing Japanese invasions possibilities is more constructive IMO, as they have more possibilities, and a more adapted force pool, to succeeed here.

< Message edited by Froonp -- 7/31/2007 1:12:02 AM >

(in reply to Shannon V. OKeets)

Post #: 16

RE: How will AI attacks be performed? - 7/30/2007 11:59:26 PM

Froonp

Posts: 7995
Joined: 10/21/2003
From: Marseilles, France
Status: offline

quote:

ORIGINAL: Shannon V. OKeets

Very helpful, thanks. I am not sure I understood what you meant by using the Offensive chit though. Is that simply to do a super combined and have a lot of air, land, and naval units available for several simultaneous invasions?

I think he just meant using a land action offensive chit to double the more invading units possible, as invading Tokyo is a VERY had affair, if it is garrisoned by simply 1 unit.

< Message edited by Froonp -- 7/31/2007 12:34:57 AM >

(in reply to Shannon V. OKeets)

Post #: 17

RE: How will AI attacks be performed? - 7/31/2007 12:31:52 AM

Froonp

Posts: 7995
Joined: 10/21/2003
From: Marseilles, France
Status: offline

Invasion wise, discussing the allied invasion power is even better than the Japanese one. The Japanese nearly only invade islands, that are one shot invasions. when the invasion is finished and successful, then the invasion troops are static and in defense, often replaced by elite garrison troops.

The allies invasions are continent wise invasions. They need to get ashore, and to move on so that more troops get ashore, MECH / ARM troops preferably, to help expand the beachhead the quickest they can.

These invasions need to be from multiple hexes, and to be able to join up quickly and without problems. If not, adjacent hexes are better. 2 hexes are a minimum in invasions of Italy & Europe, but the more the merrier. 4 hexes can be considered as a very large scale invasion. More is possible obviously. But keeping an extra "invasion hex" in reserve can also be convinient to open an immediate 2nd front behind reinforcing enemies.

For that, the necessary number of AMPH & TRS (depending on the number of MAR) need to be ready in ports that will allow them to sail the 3 section at a minimum (in the minimum number of naval moves too if possible). Troops must either start with the ships, or be on shores where they will be able to be loaded from the ships at sea (on HQ or in ports). In an Invasion of Italy for example, the best would be to have all the short ranged fleet and TRS / AMPH in Malta, with all invasion capable units in Malta, Cagliari and Olbia (6 units, or 5 plus a PARA, and the 6th place taken by a reinforcing ARM / MECH). Reinforcing MECH / MAR / HQ-A should either be transported immediately into Olbia & Cagliari as they are emptied, or already ready on Sardinia's shore on HQ. Obviously, more TRS should wait in Malta, wait to sail in the sea when the Beachhead is expanded.

Having an HQ taking part to an invasion can be good, but is also a two-edged sword. It is necessary if a Land Offensive Chit is to be used on the invading units, and if no offensive chit is used it can be used to provide HQ support. But the problem is that it will finish the impulse disrupted thus immobile. It will further hampers the troops mobility as less troops will vacate the beachhead and less troops will reinforce. Also, it is an help for the enemy because this is more chances for him to put you back to sea.

Key for future expansion of the Beachhead, beside the success of the invasion proper where you should aim at trippling each stack (or simply aim at reaching a secure odd ratio -- one that avoid you to get disrupted, and who minimizes losses), is the ability of keeping enough ground support bombers, and escorting fighters in comparison to what the enemy can put in the air, and also the ability of keeping enough shore bombardement. Ground Striking possible enemy counterattack forces (HQ and ARM / MECH preferably) is also necessary and should be done before the invasion if possible. Often, disrupting only 1 units in an hex adjacent to the invasion site is sufficient because this means that this is on less "counterattack space" that the enemy can use. Also, sometimes a convenient PARA drop on a key rail hub can helps preventing any reinforcements to arrive on the field.

In the initial turns of the invasion, and sometimes for a long moment after it, the fleet should be able to garantee 100% supply (which never happens generaly) for the ground troops. TRS should rotate between staying at sea with a small portion of the fleet at turn end, and a new TRS should come with the main fleet as soon as possible. Only when the enemy's forces at sea are sufficiently worn out, or when the battle went far away enough from shore can this be slowed down.

Planes should not hesitate (especially short ranged fighters) to rebase as soon as possible on the invasion beachhead, providing supply is reasonably assured.

Well, this all seems trivial for me when writing all this, I'm not sure it is of any help.

(in reply to Froonp)

Post #: 18

RE: How will AI attacks be performed? - 7/31/2007 1:10:14 AM

Froonp

Posts: 7995
Joined: 10/21/2003
From: Marseilles, France
Status: offline

quote:

ORIGINAL: Shannon V. OKeets
Oh, and something similar is needed for paradrops.

Basically, Paradrops on coastal hexes are the same as invasions, except that they are easier (more powerfull) because the para are not halved and they give a +1 for each para (optional rule). On the other hand, they risk interception by enemy planes, so assessing this risk and the available escort is critical here. It also means that it is 1 or 2, or even 3 less grond strike missions that will be avalable.

This alone, plus all of what I have written on the necessity to keep a fleet at sea, should show everyone that this is a team game that the CW should play with the US, and Italy with Germany, if they want it to merely being possible.

Paradrops on inland hexes are simpler to assess, as this time no shore bombardment will be possible, only air support, and supply can't be opened without a beachhead expanding. The interception problem subsist and should be taken care of. This means that this kind of "invasion" is really harder, but can be quite convinient if it block railways vital for the supply or the bringing of reinforcements for the enemy.

(in reply to Shannon V. OKeets)

Post #: 19

RE: How will AI attacks be performed? - 7/31/2007 2:59:04 AM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

Patrice,

Thanks. This helps a lot.

Yes, I could probably have written something similar myself, but it would have been a distraction from the debugging I am working on. Your comments are quite clear and I can translate them over AIO rules - though I need to do some background work first creating labels for the different unit types, etc.. Together with Christopher's on odds calculations, I believe I can formulate text on conducting invasions from planning through to enhancing the beachhead.

This will provide more details for strategic plans that include: taking Malta & Gibraltar, Closing the Med, and Sealion. And it will apply in many other places too (e.g., taking Norway or Sweden).

_____________________________

Steve

Perfection is an elusive goal.

(in reply to Froonp)

Post #: 20

RE: How will AI attacks be performed? - 7/31/2007 7:14:36 PM

composer99

Posts: 2923
Joined: 6/6/2005
From: Ottawa, Canada
Status: offline

Glad to be of help.

To expand a bit on my Tokyo comments, I would suggest offensive chit use depends on context.

Obviously, if you can get away with tactical/strategic invasions without using o-chits, that is best. Difficult to put it together properly, of course. The invasion of Italy can probably be managed this way, with the CW and US splitting up land and naval impulses to swamp Italy with invaders and then press inland from there.

If you are looking to do a massive 4-6 hex invasion, like in France and NW Europe, you will almost certainly need to do a supercombined.

If, on the other hand, you are trying to get ashore in Japan, which will, by the time you make it there be very strongly defended with land units, then you will probably need to crack out the land o-chit, as you probably will only be able to land in one or two hexes against terrific opposition.

_____________________________

~ Composer99

(in reply to Shannon V. OKeets)

Post #: 21

RE: How will AI attacks be performed? - 7/31/2007 8:49:27 PM

Shannon V. OKeets

Posts: 22095
Joined: 5/19/2005
From: Honolulu, Hawaii
Status: offline

quote:

ORIGINAL: composer99
Glad to be of help.

To expand a bit on my Tokyo comments, I would suggest offensive chit use depends on context.

Obviously, if you can get away with tactical/strategic invasions without using o-chits, that is best. Difficult to put it together properly, of course. The invasion of Italy can probably be managed this way, with the CW and US splitting up land and naval impulses to swamp Italy with invaders and then press inland from there.

If you are looking to do a massive 4-6 hex invasion, like in France and NW Europe, you will almost certainly need to do a supercombined.

If, on the other hand, you are trying to get ashore in Japan, which will, by the time you make it there be very strongly defended with land units, then you will probably need to crack out the land o-chit, as you probably will only be able to land in one or two hexes against terrific opposition.

"The sun comes up over the horizon and casts light upon the darkness; and it is good." - a phrase of my own invention.

_____________________________

Steve

Perfection is an elusive goal.

(in reply to composer99)

Post #: 22

Page: [1]

All Forums >> [New Releases from Matrix Games] >> World in Flames >> How will AI attacks be performed?

Page: [1]

Jump to:

New Messages	No New Messages
Hot Topic w/ New Messages	Hot Topic w/o New Messages
Locked w/ New Messages	Locked w/o New Messages

Post New Thread

Reply to Message

Post New Poll

Submit Vote

Delete My Own Post

Delete My Own Thread

Rate Posts

7.719

Slitherine

How will AI attacks be performed?