• We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
PB-DK said:
3. the ai will ALWAYS build the highest model valued unit regardless of what kind of statistics the unit might have... (e.g. in standard HOI an interceptor has model value 1 and escort a value of 3... and as soon as the ai aquires escorts... ;)) i dont know if it is a problem for you guys as i havn't tested TGW yet...

Known and considered :)


PB-DK said:
5. oh and if you want the british to stop invading the german beaches; add a -1000 (minus) as target prio for the german beaches in the invasion/targets section of the ai, it will stop the landings...
yikes... hope it helps :p

That looks like it could come in useful for other places we don't want the AI to go - should we require that to happen...but I don't think the British AI should be discouraged from attacking the German coast as this strategy was certainly given credence by certain parts of the armed forces (particularly the Royal Navy).
 
Maybe the UK could have some sort of AI only event in which it decides the strategy to adopt against Germany. The first option could be the historical "send the army to france", with the invasion parameter for N Germany set to those -1000, and the second one would have normal invasion settings for those beaches. We could even have a completely insane one with Britain sending large amounts of troops to Russia as option C. :)
 
Gwalcmai said:
Maybe the UK could have some sort of AI only event in which it decides the strategy to adopt against Germany. The first option could be the historical "send the army to france", with the invasion parameter for N Germany set to those -1000, and the second one would have normal invasion settings for those beaches. We could even have a completely insane one with Britain sending large amounts of troops to Russia as option C. :)

Could do....keep in mind that the decision to send the BEF to France was made by the Committee of Imperial Defence in 1912, so when war comes, this would have to be the overwhelming favourite (if we can get the AI to do it at all)
 
If that is made into an event, the chances are, without the "troops to Russia" option, ~95% for the historical choice and ~5% for the german landings. With an option C it changes to ~90% for A, and ~5% for each of the others (Of course you know this, it's in Havard's hacking haven :)).

If those odds are still too good for an ahistorical strategy being adopted, maybe we can add a random = 50 trigger to the event. If the A action for the event is not to change the AI file in use (that 1912 plan you mentioned) then the chances for the ahistorical choices are pretty slim. Any lower, it would probably not be very worthwhile to even have the possibility. :)
 
Gwalcmai said:
If that is made into an event, the chances are, without the "troops to Russia" option, ~95% for the historical choice and ~5% for the german landings. With an option C it changes to ~90% for A, and ~5% for each of the others (Of course you know this, it's in Havard's hacking haven :)).

Okay then, we'll opt for this option! :)

So then.....who wants to code it? ;)
 
The event is quite easy to code, there's only the slight problem of making the AI files. :)

Code:
event = {
  id = xxxxx
  random = no
  country = ENG

  name = "AI_EVENT"
  desc = ""
  style = 0

  trigger = {
    ai = yes
  }

# decide the strategy from the start, to be able to prepare it
  date =  { day = 1 month = january year = 1914 } 
  offset = 1
  deathdate = { day = 2 month = january year = 1914 } 

  action_a = {
    command = { } #do nothing, use default AI
  }

  action_b = {
    command = { type = ai which = "tgw_ENG_invasion.ai" } #prepare to invade germany
    command = { type = build_division which = transport value = 0 } #give the AI a hand. delete this or replicate a couple of times?
  }

  action_c = {
    command = { type = ai which = "tgw_ENG_support_russia.ai" } # stupid BEF-to-Russia idea, delete, replace or keep?
  }
}

EDIT : BTW, anyone know how to make the code tags create a scroll bar instead of bleeding off the screen edge?
 
Gwalcmai said:
The event is quite easy to code, there's only the slight problem of making the AI files. :)

Who'll do that then?

Gwalcmai said:
EDIT : BTW, anyone know how to make the code tags create a scroll bar instead of bleeding off the screen edge?

Press return after every line in the box where you write your message.
 
It is time that we gave serious consideration to the AI setup in TGW – we have not made any significant progress in this area in more than four months, and it is evidently the mod’s weakest aspect. At the moment, we have a few skeleton AIs that chart the overall manner in which each country fought the war. Even these are inadequate at the moment – one only has to have a brief look at the unit building priorities sections of the various AI files to notice that the only aerial units that are built are reconnaissance planes and that no AI run country will build transports (no fear of playing as Britain then).

The first set of AI files that need to be produced – and we’ve had some contributors attempt this already – is the implementation of war plans when hostilities break out. Primarily, we shall require files that attempt to imitate the opening moves of the major powers, who are expected to be involved in the war from near the beginning (whether it be August 1914 or sometime in 1915). Obviously, these countries are Germany, Austria-Hungary, Russia, France, Great Britain, Turkey, Italy, Japan and possibly Belgium and Holland.

Firstly, Germany shall require three different AI files when war breaks out –Schlieffen’s Schlieffen Plan including the invasion of Holland, Moltke the younger’s Schlieffen Plan excluding the invasion of Holland and the ‘East First’ approach as advocated by Moltke the Elder. This is linked to event 21002, which already gives the German player these three choices. I recall that Stephen tested a very basic Schlieffen Plan AI file and observed that the Germans captured Arlon before not advancing any further. This leads me to think that imitating the Schlieffen Plan ‘step-by-step’ may be necessary, by creating a file that covers the projected invasion of Belgium, which leads to another covering the projected movement into Northern France and another that goes around Paris, etc. The ‘East First’ approach should obviously put the bulk of German forces attacking in the East with an appropriate amount to stand on the defensive in the west.

Austria-Hungary’s AI could be quite complicated – the bulk of the army should be concentrated against Russia with a much smaller amount on the border with Serbia. Problematically, imitation of Austria-Hungary’s war plan will require that a small reserve be kept somewhere in the middle of the empire, somehow :)wacko: ). When war breaks out the Austrian AI ought to choose between the threat from Russia or the necessity to fight Serbia – perhaps either choice should lead to the province priorities in either outcome changing: heavily favouring Russian border provinces and heavily favouring Serbian border provinces depending on what option was chosen.

The Russian one should be quite straightforward – it begins with an obligation to attack Germany in aiding France, directed against Memel and Konigsberg provinces, with another attack going into Austria-Hungary.

France perhaps ought to have two variations – firstly one to simulate Plan XVII with an attack full of élan into Alsace-Lorraine, and a different one whereby the French army uncharacteristically goes onto the defensive at the outset of the conflict, possibly shifting the weight of the army’s strength to the northern provinces.

Great Britain can have two variations as well. The first is the transportation of units to France (somehow) to imitate the BEF occupying its place on the left flank of the 5th army opposite Belgium. The second, less likely, but more adventurous plan would be the simulation of Admiral Sir Arthur Wilson’s plans to attack Pomerania. Combined with similar schemes to grab Heligoland and attacking Schleswig-Holstein, setting Germany’s coastal provinces as an attack priority ought to make this one work. This shouldn’t be a problem, seeing as Britain’s default AI has troops land in Germany anyway.

Both British and French AI’s ought to have their colonial garrisons in Africa go on the offensive as well.

Assuming that she joins the Central Powers, Turkey ought to retain a defensive posture around the straits while attacking into the Caucasus, not doing much in Mesopotamia and tentatively (if the AI can move ‘tentatively’ :confused: ) moving into the Sinai. An alternative plan might see the Turks try to recapture Egypt.

Italy, in a straightforward sense would attack the South of France and Tunis if it were in the Central Powers. Should it find itself in the Entente early on, then its AI should obviously be encouraged to attack Bolzano province.

Japan’s territorial priorities would be for the taking of Germany’s Pacific colonies, especially Qingdao province. I can’t see the Japanese deviating from this plan at the outbreak of war.

A lot of the above it an exercise in stating the obvious, but I think it helps to clarify the brief outline of what each country’s war plan AI files should consist of. It would probably be sensible to let these be the norm for three or four months, before having the AI switch to a different file that switches to a more normal conduct of war.
 
In terms of priority, I think the first thing to do in this area has to be the creation of a standard, basic AI for each country.

Writing special AIs to reproduce specific war plans, with lots of options, seems to me to be a luxury; something enthusiasts can play with once the basic mod is up and running.

I noticed that most of the standard WW2 AI files have been left out of TGW. I'm not sure why - while their build priorities might need tweaking (less air and armour) and their attitude to new countries like Austria-Hungary needs to be coded, it still seems to me that Paradox have alredy done most of the work in this area - so why not take advantage of it?
 
StephenT said:
In terms of priority, I think the first thing to do in this area has to be the creation of a standard, basic AI for each country.

Writing special AIs to reproduce specific war plans, with lots of options, seems to me to be a luxury; something enthusiasts can play with once the basic mod is up and running.

I noticed that most of the standard WW2 AI files have been left out of TGW. I'm not sure why - while their build priorities might need tweaking (less air and armour) and their attitude to new countries like Austria-Hungary needs to be coded, it still seems to me that Paradox have alredy done most of the work in this area - so why not take advantage of it?

Exactly. We already have 'basic' AIs for every country - in the mod and the standard HOI, which is why we can afford to start creating ones for war plans....I definately would not call AI scripting for war planning as a 'luxury' - having the war start in as near a historical sense as possible is, I think, obligatory, and the files for which should be written as soon as can be.

The standard files will, as you say, require tweaking in various ways....we just need someone to actually do the adjustments...
 
Shadow Knight said:
But ones that absolutely need to be done now are: Great Britain, Germany, France, Austria-Hungry, Belgium, Russia, Italy, Ottoman Empire, Romania, Bulgaria, Serbia/Montenegro, the US, and Japan.

These were the historical majors and minors.

For GB: We need their navy to concentrate on the North Sea, English Channel, and to a lesser extent the Med.

We need their army to make sure England it self is adequately protected (to keep stupid AI from launcnihg a Sea Lionesque invasion), but to send most of its land forces to France/Belgium. Second priority goes to defending Egypt, and occupying German Africa.

A decent mix between building ships, soldiers, and tech.

For France: Their navy needs to concentrate on the Med, but send some support to GB in the North Sea/Atlantic.

France's armies need to concentrate on the German border, then a few forces to Belgium's border to aid in the defense there, and then finally the least priority is the Italian border and occupation of German West Africa.

France should contruct a lot of infantry with a moderate attention to tech.

For Germany: We have an event for Germany that lets Germany choose whether to follow the Schlieffen Plan or not, so we may need a AI only event that changes its priorities depending on what it chooses.

First the German Navy will be independent of this decision. Its priority should be gaining superiority in the North Sea and the English Channel, but only if it has the strength to do so, else remain in port. Also a strong presence in the Baltic is needed.

For German the Army:
Case A) (Historical path) Heavy presence along the French and Belgian borders with wanting to occupy Paris. Then a good defensive force along its border with Russia. (Probably a small force should be sent to protect those stupid beach landing areas in Germany to keep one division Entente forces from landing there!).

Case B) (Full Schlieffen Plan) Same as A but with troops shifted to cover the border with the Netherlands also.

Case C) (Russia first) Defensive forces along the border with France, and large concentration of forces along the Russian German border.

German should produce a lot of infantry, very few ships, and a moderate attention to tech.

Russia's priority should be their border with AH and Germany, and a small force along the Turkish border. Their navy should try to gain advantage in the Baltic and Black Seas. They should produce infantry/cavalry type forces with a very small tech budget.

AH should keep a small force along its Italian border, a large force against its border with Russia, and a large force with its border with Serbia/Montenegro.

Its navy should concentrate on the Adriatic Sea. AH should build lots of ground units, and with a small to modest attention to tech (bigger than Russia's but not by much).

Serbia/Montenegro: Build infantry/Cavalry like mad, and keep everything along the border with AH (perhaps a small force along the Bulgarian border).

Bulgaria: Build ground units, and concentrate on the border with Romania and Serbia (small force agianst Greece's border too).

Belgium: conentrate on building ground units like mad, and defend the border with Germany.

Japan: Take German posessions in Asia, then it should go to sleep content with its gains there.

US: before war, small tech investment and NO unit buildup. When at war: building units (ships/ground units) at an insane rate. Then concentrate in France.

For the Ottoman Empire: (3 choices) (Same as germany we may need to have AI events only to change OE's AI characteristics)

Choice A) Join the Central Powers (Historical), concentrate on Egypt with a strong defensive force in Iraq, and beach landing areas. Another good size force along the Turkish Russian border in the Caucuses.

Choice B) Neutral (Unhistorical choice 1) small investment in tech, and light buildup of forces, but do nothing but protect borders.

Choice C) Entente (Unhistorical choice 1) concentrate on building units and placing them on the Bulgaria border.

Italy: (Same as OE and Germany, three chocies that may need AI only events to change characteristics)

Choice A) (Historical-Entente) navy should try to break into the adriatic sea. land forces concentrate on AH's border.

Choice B) (join CP's) Navy in the Med. Land forces protect Libya and move into French North Africa and british Egypt. And then protect Italy itself with a modest force along France's border. (vanilla Hoi AI might be good enough for this choice with a modest changes to it.)

Choice C) Neutral. Do very little except modest tech investment and small unit construction with favoring of the navy.

Everyone else left in the world should have a small tech budget, and limited unit construction. But be mostly asleep. (So you could create one or two of these and just copy and past to the rest.)

What does everyone think about SK's ideas for AI arrangements posted a few months ago? I think it's a very good outline to use with regard to fine tuning our 'normal' AI files....

Additionally, some files will need to be produced that deviate from the historical approach taken by the combatants. For example, the debate in the German Command between 'Westerners' who favoured taking the offensive against the French and British, and the 'Easterners' who believed that Germany's efforts should be concentrated against Russia and Serbia. Similarly, the same can be said for Britain - the possibility should remain that they can concentrate on the Western Front, as General Robertson wanted, or on peripheral campaigns such as Palestine, Salonika and Italy, advocated by Lloyd George.

Furthermore, I think that a seperate file needs to be created to simulate a country's attempt to conduct a large set piece battle, with this being in effect for a few months before it switches to the normal AI again, to imitate the closing down of a battle, or the return to normal combat conditions if that achieved some sort of breakthrough.
 
Allenby said:
What does everyone think about SK's ideas for AI arrangements posted a few months ago? I think it's a very good outline to use with regard to fine tuning our 'normal' AI files....

Additionally, some files will need to be produced that deviate from the historical approach taken by the combatants. For example, the debate in the German Command between 'Westerners' who favoured taking the offensive against the French and British, and the 'Easterners' who believed that Germany's efforts should be concentrated against Russia and Serbia. Similarly, the same can be said for Britain - the possibility should remain that they can concentrate on the Western Front, as General Robertson wanted, or on peripheral campaigns such as Palestine, Salonika and Italy, advocated by Lloyd George.

Furthermore, I think that a seperate file needs to be created to simulate a country's attempt to conduct a large set piece battle, with this being in effect for a few months before it switches to the normal AI again, to imitate the closing down of a battle, or the return to normal combat conditions if that achieved some sort of breakthrough.


:eek:

Wow how many pages back did you have to go to root that old post? Anyway since I did go to all the trouble of typing that thing up I of course am in favor of using it. Now if we could only get one of those generous AI coders to come back and help.
 
Shadow Knight said:
Now if we could only get one of those generous AI coders to come back and help.

That might be wishful thinking. Lothos offered to help, but he is generally tied up in other projects. All the people who offered their help have gone absent and attempts to bring them back have been unsuccessful. If we simply put out a plea for more AI coders I suspect that we'll get more of the same - enthusiastic people queuing up to help before completely disappearing.

Besides, I think these AIs will have to be done by people who 'know' TGW, as opposed to contracting out to people who are good with AI, but have only played the mod once and have only very basic First World War knowledge to go with it. Thus, I think the maxim of 'if you want something done right, do it yourself' should apply and that the bulk of the AI should be produced from within the team, not without.
 
Noone contacted me again and my work did not raise a big interest and got zero feedback, only Allenby seemed interested in trying them but for some reason they did not work on his install -did not recognize "strength" and "organization" params...- and noone else tried them, and then everyone forgot about my work. I reworked on them after 1.02b was out and tested them with the suggested mobilization tweaks, but noone contacted me so I assumed you had something going on or were testing my files on 1.02b and they were total crap. I think it's better you handle the AI internally in your group and I wish you good luck.
 
maizenkor said:
but noone contacted me so I assumed you had something going on or were testing my files on 1.02b and they were total crap. I think it's better you handle the AI internally in your group and I wish you good luck.

On the contrary, I contacted you but you either didnt respond or my message never got to you... :wacko: if you still have the files then I would be happy for them to be tested. :) Unfortunately, if they are still afflicted by the same unfathomable problem, then we might have to write new files ourselves, or radically alter the ones supplied until they do work.
 
Well I got no email since the one from 20 Dec, maybe it was just email routing problems and I should have showed up on the forum resubmitting my files instead of waiting. I'll send you the last version I had which is different from the 1st one and don't use the params that caused trouble on your install. I'll join a readme on how to use them and some observations I had after lots of testing, so you could use them as a skeleton to implement your files and try ideas.

The hardest part I found is the German AI for the schlieffen plan. I could not manage to find the proper combination of garisson and front params to reproduce the historical behavior. You can make nice mobilizations and concentration of forces with the garrison AI but when the war starts the front AI takes the hand and tends to undo those concentrations and make a mess of the plan. You can fine tune the garrison AI (put troops there, move more to this frontier etc..) but can't tell the front AI where to attack exactly. You can artificially slow it down on a front (passivity) but that's all, and this is not necessary a good idea. I think you'll have to find a way round and help the german ai, for instance with a Guns of August event to temporary boots their offensive and adding forts on the French and Belgian border.

Good luck!
 
maizenkor said:
Well I got no email since the one from 20 Dec, maybe it was just email routing problems and I should have showed up on the forum resubmitting my files instead of waiting. I'll send you the last version I had which is different from the 1st one and don't use the params that caused trouble on your install. I'll join a readme on how to use them and some observations I had after lots of testing, so you could use them as a skeleton to implement your files and try ideas.

The hardest part I found is the German AI for the schlieffen plan. I could not manage to find the proper combination of garisson and front params to reproduce the historical behavior. You can make nice mobilizations and concentration of forces with the garrison AI but when the war starts the front AI takes the hand and tends to undo those concentrations and make a mess of the plan. You can fine tune the garrison AI (put troops there, move more to this frontier etc..) but can't tell the front AI where to attack exactly. You can artificially slow it down on a front (passivity) but that's all, and this is not necessary a good idea. I think you'll have to find a way round and help the german ai, for instance with a Guns of August event to temporary boots their offensive and adding forts on the French and Belgian border.

Good luck!

I have passed your work onto Patch, who will now test it. Inevitably, there will need to be changes and alterations, so if you are willing to continue working with us, then just say, and I shall advise as to what things need to be done. :)

If you do continue your work, then you shall have plenty of responsibility, as getting some good AI done is the most important task that has to be fulfilled in the coming weeks and months.
 
maizenkor said:
If you have problem using the files mail me but I will not be able to provide more support so:
"I thereby grant you an unlimited license on my files" :)

Ah okay then :) That leaves us with a lot of work to do......I'm sure someone will volunteer to take the lead on this matter...