AI Opponent - Geography

[Shannon V. OKeets]

2nd and last in the series.

The large swathes of desert in the south are "do not enter" Land Regions for the AIO. That is an idea from Peter. Almost all of northern Siberia will fall into that category too. No sense having the program get operations along the North African Med border confused by worrying about hexes far to the south.

[HansHafen]

Well, I THINK about it. So the AI will THINK like a human. Good deal.

[warspite1]

ORIGINAL: HansHafen

Well, I THINK about it. So the AI will THINK like a human. Good deal.

Warspite1

It depends - if it thinks like me then the AI will be rubbish [:D]

[pzgndr]

So the AI will THINK like a human. Good deal.

How a human "thinks" about choices is basically irrelevant to another opponent who only sees a decision made for whatever reasoning, good or bad. The AI should have some weighted probability distributions or something to "choose" between alternatives. In the end, your computer opponent makes a decision and acts. Good enough, yes? What's encouraging from Steve's descriptions are the different levels of decision making from top down. I don't expect brilliance, but so far it looks good that the AI should make some reasonable decisions at the strategic levels, then operational levels, and then its tactical dispositions. [8D]

[peskpesk]

Here is a very simplified flowchart of the German Grand Strategy and how it could connect to Theaters of operation (TO) and Areas of operation (AO). Note that all options are not shown here and the timeline is guidance.

[Shannon V. OKeets]

Very nice.[:)]

[Shannon V. OKeets]

1st of 3 posts in a series.

After a lot of revisions, Peter and I (with advice from Patrice) have the geographic breakdown for the Eastern Europe Theater of Operations. Here is the northern third.

Sweden is its own Area of Operations (AO) with 2 Land Regions (LR): north and south. The north is shown here.

The Finland-Arctic AO has 4 active LRs: Northern Finland-Norway, Southern Finland, Murmansk, and Archangel. There is a 5th LR to the far east which the AIO consders a "Don't Even Think About Moving There" LR.

The idea of this breakdown is that from the Finnish point of view there is the north and the south. From the USSR point of view there is Archangel (which doesn't see much action), the defense of Leningrad (part of a different AO), and then the rest of Karelia (from which either attacks are made on Finland, or Murmansk and the rail line thereto are defended).

[Shannon V. OKeets]

2nd in the series.

Here is the western portion of the Eastern Europe TO.

In the northwest is the southern Swedish LR.

The Poland-Baltic AO has 4 LRs: Estonia (which has part of Latvia), Lithuania (which has the rest of Latvia and 6 hexes of Poland), Western Poland, and Eastern Poland.

The Balkans AO has 4 LRs: Hungary, Rumania, Bulgaria, and Yugoslavia.

The only breakdown that isn't along country boundaries is the Poland-Baltic AO. There the thought was that which language the people speak isn't as important as the defendable river lines. For example, in defending against Barbarossa, the USSR has the Lithuania LR as it frontline defense and then falls back in the north to the Estonia LR. That translates as a 5 hex front which decreases to a 4 hex front.

When fortunes are reversed and the Germans are defending late in the war, these geographic breakdowns are still fairly reasonable.

[Shannon V. OKeets]

3rd and last in the series.

This was the most difficult portion of the Eastern Europe TO to figure out.

The Turkey AO has two LRs: Northwestern Turkey and Eastern Turkey. The rest of Turkey is part of the Mediterranean TO.

The Central Western USSR AO has 3 LRs: Leningrad, Vitebsk, and Smolensk.

The Southwestern USSR AO has 3 LRs: Kiev, Kharkov, and Crimea.

The Central European USSR AO has 3 LRs: Vologda, Moscow, and Voronezh.

The Southeastern European USSR AO has 3 LRs: Rostov, Stalingrad, and Caucasus.
---
Again the river lines strongly influenced our decisions here. I also wanted to keep the hex count roughly comparable for each LR. If you don't count the relatively unimportant hexes, you'll see that most of the LRs in the USSR are in the 30-50 hex range.

We went back and forth defining these through several iterations and ended up favoring straight lines in a lot of cases. Peter handled all the graphics (mucho thanks[&o][&o][&o]).

The purpose of the geographic breakdown for the AIO is made manifest here. The AIO will coordinate its defense/attack within each AO and LR. You may see a shifting of forces northward or southward across boundaries depending on where the enemy positions his units, but you won't see the complete evacuation of part of the frontline. If opportunities present themselves for controlling the rest of an AO or LR, or expanding control into an adjacent AO/LR, then the AIO will try to make that happen. If defending an AO or LR becomes hopeless, then the AIO will retreat to an adjacent AO/LR.

Of course the boundaries are not sacrosanct for tactical, or even operational, decision making (e.g., historically the Germans sent panzer units south from Gomel to surround Kiev). On the other hand, the AIO will be working with a structured 'knowledge' of the geography of each TO, and not be required to examine each hex on the map as an isolated point.

[Shannon V. OKeets]

1st in a series of 3.

Here is the East Africa TO with breakdowns into Areas of Operation and Land Regions.

The AOs and LRs are:
Egyptian Sudan: Southern Egypt & Sudan
Yemen-Aden: Yemen & Aden
Horn of Africa: Eritrea, Fr-Br Somaliland, Ethiopia, & Italian Somaliland

The rest of the last LR is shown in the next post.

[Shannon V. OKeets]

2nd in a series of 3.

Patrice made a good case for splitting the Belgian Congo between the East & West Africa TOs. The rail line to the Azanian Sea is important tactically for controlling the resource northwest of Elisabethville.

The AOs and LRs are:
Belgian Congo East: Belgian Congo NE & Belgian Congo SE
Central East Africa: Kenya, Uganda, & Tanganyika

[Shannon V. OKeets]

3rd and last in the series.

Peter is doing all the work on these, with advice from Patrice and myself. We are open to suggestions and comments.

The AOs and LRs shown here are:
Southern Africa: South Africa
Madagascar: Madagascar
Southeast Africa: Mozambique, Southern Rhodesia, Nyasaland, & Northern Rhodesia.

The rest of the last LR is shown in the previous post.

Serendipitously, this exercise is resulting in a fresh set of posts of the world map.

[warspite1]

ORIGINAL: Shannon V. OKeets

Serendipitously, this exercise is resulting in a fresh set of posts of the world map.

Warspite1

Which is always welcome - these maps are the S-U-P-E-R-B!

[Shannon V. OKeets]

1st in a series of 2.

Peter has been hard at work. Here is what he and I came up with for China Areas of Operation and Land Regions therein.

AO China Northwest: Sining & Kansu LRs.
AO China North Central: Shensi, Shansi, Honan, & Ordos Desert LRs.
AO China Northeast: Hopeh & Shantung LRs.

We wanted to use mountains and river lines for borders. In most cases we used mountain lines, because they stronger and do not disappear in snow like the river lines do. The borders are primarily for defense, favoring whichever side holds the hexes in 1939. So, the Chinese get most of the best defensive borders in the west and the Japanese in the east.

In the north the Japanese try to capture all the Communist Chinese cities so the CC have no place for their reinforcements to arrive. They jump off from either Hopeh, going through Shansi, or from Shantung, going through Honan. Their objective is to take Shensi, Kansu, and then finish up with Sining and Ningsia.

Anyhow, that is my logic for the AOs and LRs I have outlined.

[Shannon V. OKeets]

2nd and last in the series.

AO Szechwan: Chengtu, Ankang, & Chungking Approach LRs.
AO China Southwest: Yunnan & Kweichow LRs.
AO China Southern Coast: Kwangtun & Fukien LRs.
AO China South Central: Hunan and Kiangsi LRs.
AO China East Central: Hupeh, Anhwei, & Kiangsu LRs.

In the south the Japanese try to capture Chungking (and Chengtu), thereby crushing all hope of the Chinese making any serious comeback. Controlling Hupeh is almost irrelevant, except that Ichang is an important city for Nationalist Chinese reinforcements. From Hupeh, they can push on to Ankang and enter Szechwan from the northeast - but that is difficult terrain and bringing up reinforcement is arduous. The more common attack line is from the southern coastal cities, combined with attacks south from Anhwei. Both Kiangsi and Hunan have a resource as a reward for the Japanese. This makes Kweichow easier to take because the rail lines let the Japanese bring up reinforcements. The final push is from Kweichow into Szechwan.

So, whadda ya think?

[Extraneous]

Will these "Geographical Areas" be visible on the games maps? And if so will you be able to be toggled them on and off?

[peskpesk]

Here is a graphical example of the AIO and how the line of reasoning could be, given the suggested AO and LR.

[Shannon V. OKeets]

ORIGINAL: Extraneous

Will these "Geographical Areas" be visible on the games maps? And if so will you be able to be toggled them on and off?

No, they will not be visible.

[Extraneous]

So the "Geographical Areas" will not be adjustable or seen by the human players?

adjustable - able to be assigned a value or priority by the human players.


Is this going to be displayed by a graphic and described somewhere in the RAC?

[Shannon V. OKeets]

ORIGINAL: Extraneous

So the "Geographical Areas" will not be adjustable or seen by the human players?

adjustable - able to be assigned a value or priority by the human players.


Is this going to be displayed by a graphic and described somewhere in the RAC?

I am not so sure I want to give human players complete insight into how the AIO makes it decisions.