Problems simulating World War II

[techumseh]

I've been playing Curt's latest version of Dictator over the holidays and having lots of fun with it. At least I think it's the latest version - Dictator Epic v.1.3 (I'm playing as Germany, as Curt would want, I'm sure )and it works very well.

I conquered England with some difficulty, and had a devil of a time getting air units past Malta and into North Africa. But In the end, a smallish force got into Egypt as far as Suez. Russia is going well, though I triggered the war unexpectedly by attacking a Soviet enfineer unit that had slipped into Germany. For some reason I thought it was British! Anyway, by mass producing air units, I've now conquered north Russia from Archangel to Moscow, down to Rostov and Sevastapol.

The US has landed in Italy and France - a year early! I managed to defeat the Italian invasion, but I must admit that I lost France to a very effective simulation of Operation Overlord. I'm currently holding a defensive line Amsterdam - Cologne - Milan. Leaving aside the unlikely choice of a Normandy landing with a German occupied Britain in the rear, the events are very effective.

Curt has managed to give good balance playing as the Germans, with lots of the main historical challenges well simulated. There are a few issues, which reminded me of problems I encountered trying to finish my version of WWII, Global War. (more)

[techumseh]

I've been wondering about several questions related to doing WWII strategic scenarios. I ran up against these issues trying to finish Global War, and playing Dictator reminded me that they are challenges to others as well.

1. How can you use aircraft carriers?

Dictator simulates the European war very well - the Pacific not as well. There was some sort of a Pearl Harbor event from what I could tell, but certainly the Japanese naval offensive is spotty and the US Pacific counter offensive is non-existant.

Can you even move units across the terminator line with events? You know what happens when you try to use goto to move a naval unit across the line - it goes the wrong way!

The AI won't use aircraft carriers, so giving them to a human player gives a huge advantage. Kobyashi solved this by making AI carrier forces powerful surface units with no planes. In Global War I decided on a multi-player format, with house rules (no B-29s on carriers, etc).


2. How do you ensure that the right units are built and are used by the AI as intended?

This is a perennial problem for scenario designers. I must admit to being fuzzy on how the AI decides to build which type of unit. I'll bet the Civ2 strategy folks have gone into this in some detail. Has anyone run across any useful threads on this?

This is a particular problem for scenarios which have numerous types available to each civ, as in most WWI scenarios. In Dictator, the Soviets produce mostly air units, with some artillery. No tanks. As a result, all the action is in the air - ground units mostly just occupy cities. I had similar problems with GW.

The use of stackable terrain, ie. fortresses changes the way ground units operate. They are clearly less aggressive or mobile. I'm not sure that a good ground offensive is possible using fortresses as stackable terrain. In my remake of Roman Civil War, I've replaced fortresses with airbases and the results are very good. Mobility is restored, and units even operate in stacks. But that won't help in a scenario with air units.

3. How can you simulate strategic warfare?

Some of the most decisive battles in WWII were often those not dealt with by most scenarios: submarine and strategic bombing campaigns, lend lease, developing communications, and so on. How can these be simulated accurately?

For example, in Civ2 strategic bombers, eg. Lancaster, B-17 are just powerful tactical bombers. They attack other units. In reality, they attacked factories and rail communications. How can this be simulated?

In GW, I played around with using Space Ship components as strategic events triggers. So if you build a module (renamed Strategic Bomber) and an enemy civ loses 100 gold. The problem is that the enemy civ may not have 100 gold, and we don't know how that would affect the war readiness of an AI civ, if at all.

These are some of the problems of doing WWII scenarios that I've encountered. Any suggestions or other issues are welcome.

[Harry Tuttle]

Originally posted by techumseh
2. How do you ensure that the right units are built and are used by the AI as intended?

This is a perennial problem for scenario designers. I must admit to being fuzzy on how the AI decides to build which type of unit. I'll bet the Civ2 strategy folks have gone into this in some detail. Has anyone run across any useful threads on this?

This is a particular problem for scenarios which have numerous types available to each civ, as in most WWI scenarios. In Dictator, the Soviets produce mostly air units, with some artillery. No tanks. As a result, all the action is in the air - ground units mostly just occupy cities. I had similar problems with GW.

The use of stackable terrain, ie. fortresses changes the way ground units operate. They are clearly less aggressive or mobile. I'm not sure that a good ground offensive is possible using fortresses as stackable terrain. In my remake of Roman Civil War, I've replaced fortresses with airbases and the results are very good. Mobility is restored, and units even operate in stacks. But that won't help in a scenario with air units.

I've found that the AI chooses the units built based on the best ratio of attack/defense versus cost. I haven't run any tests on this though.

3. How can you simulate strategic warfare?

Some of the most decisive battles in WWII were often those not dealt with by most scenarios: submarine and strategic bombing campaigns, lend lease, developing communications, and so on. How can these be simulated accurately?

For example, in Civ2 strategic bombers, eg. Lancaster, B-17 are just powerful tactical bombers. They attack other units. In reality, they attacked factories and rail communications. How can this be simulated?

In GW, I played around with using Space Ship components as strategic events triggers. So if you build a module (renamed Strategic Bomber) and an enemy civ loses 100 gold. The problem is that the enemy civ may not have 100 gold, and we don't know how that would affect the war readiness of an AI civ, if at all.

These are some of the problems of doing WWII scenarios that I've encountered. Any suggestions or other issues are welcome.

When I was building Gold Gulch I needed an AI "thunderstorm" unit to attack, plain and simple. Most times it just flew off into the mountains, never to be seen again. However, I discovered that it would attack a unit that had the highest cost in its vicinity. After increasing the cost of each unique "saloon owner" unit, which typically stayed in a city, the thunderstorm unit gleefully ventured to the cities to attack. I think that if you add a non-movement unit to a city and give it a high cost then the AI will try to attack that unit.

I have also seen that the AI will only attack units that it can possibly win a battle against. If the non-movement unit has a attack/defense factor that is too high then the AI will keep its units away from it, if possible.

Simulating strategic economic issues in a scenario is somewhat hard. As you said, you can never really know what the AI treasury has in it. Here are a few possible ways to simulate strategic issues:

1. Use those strategic food routes as I've rambled about. You can set up groups of them. Upon a trigger happening all you'll need is a changeterrain event and the food route is wiped out.

2. Using ToT and the flags that come with the events engine I think it's possible for an event with a Goto command to be switched off. I don't know for certain, you have more experience with this than I do. If a Goto event could be shut off, and then another different Goto event turned on you could simulate a battlefield shift of troop movements. No longer would those AI armies march toward Moscow. Instead they could march toward Stalingrad.

3. Have an event to create settlers, hopefully under the barbarians or a non-used civ, in a blocked off corner of the map. Make the terrain worthless so that settling would be dumb for the AI. Have the settlers home city be a high value city for you or the AI. As the settlers are created they drain food away from the city, making it eventually go hungry, decreasing its population points.

Also, you could try this with generic units. As the units are created they drain shields away from the home city and create unrest if the units are large enough in number. Again, haven't tried this, but it might work.

That's what I can come up with right now. Hope this helps.

[Eivind IV]

Strategic bombers:
This is impossible with civ2, but in the Blitzkrieg scenarios strategic bombing happens when airplanes end their turn on enemy industrial terrain. The ind terrain hence gets occupied for the turn and the city will starv. As a house rule only bomber units can do this.

Like I've already put on the wish list for civ2 changes, the ability to have flying spies would solv this geniously, by having 'improvement destroyers'.

[Boco]

Try an 'Economic Map'!

Not impossible, but very tricky....

Originally posted by techumseh
For example, in Civ2 strategic bombers, eg. Lancaster, B-17 are just powerful tactical bombers. They attack other units. In reality, they attacked factories and rail communications. How can this be simulated?

This is only half-thought out, but I'll spout off anyway. I'd try this on a separate 'economic' map that contains only relevant 'air', flak, and economic units.

• Target cities on this map would be essential for supporting cities on the main map.
  
• Maybe use a clever food chain scheme like those Harry Tuttle concocts. Donor cities on the economic map, recipients on the main map.
  
• Or maybe caravans can only be built on this map. Disable intramap trade to allow only intermap trade. Make trade essential to winning.
  
• Like Eivind, I'd also use spies as bombers. They'd get a choice of improvements to attack. Have to prevent city-bribing. That's no problem if the bomber's human. If we're dealing with AI bombers...
  
• Spy bombers would also get a nuke option linked to Manhattan.
  
• Since spies must be ground units (right?), terrain would need to have one cost. On the economic map, just pretend everything is an air unit.
  
• Simulating operational range with a land spy could be tricky. One key is to make them very vulnerable on defense and high on the AI target list. They'd have to hit their target in one turn or get destroyed.
  
• An alternative (or complement) to this is to use para-spies with mf=1. They could hop far, but if they don't land next to their target, it'll take too long or risk too much exposure not to attack in the same move.
  
• You could also use the CityProduction trigger. Use a unit that can only be built on the economic map. When it's built, trigger a good event on the regular map. This can be very powerful. The job of bombers (spies) is to cripple a cities ability to build this unit.
  
• Maybe use pillaging as a weapon on this map. Squares could be guarded by flak units. When they're destroyed a 'bomber' unit (another land unit) can enter the square an pillage. Perhaps settlers could double as flak units. They could repair and defend with the same unit slot.

Like I say the details are half-baked, but I'd argue that using a separate economic map with dedicated units and cities is the key.

[techumseh]

Good ideas, guys. I'm not really working on this at the moment, I was just inspired by Curt's excellent work and thought it would make an interesting discussion.

Does anyone really know, in any detail, how the AI makes decisions about what to build? Why does it build Phalanxes and Legions, when both have a '1' role? Why does it sometimes move the Legion out of a city to attack, but never the Phalanx? This is the trickiest part of scenario design, and one I've never fully understood. I just use trial and error 'til I get something workable.

[jim panse]

@ Strategical Bombing: I would suggest the same as Eivind & Boco - flying spies

@ Tech: How is this airbase thing working in RCWs? Doesn´t the airbase improvement also add a RR to the square? Could you please explain this further?

[Boco]

I believe in ToT the Airbase loses its RR bonus.

[curtsibling]

Thanks for the mention, Tech!

The carrier AI problem is one that does hamstring so many
decent naval scenarios, I think Koby's idea was ingenious.

I found that giving ground units the 'Air Superiority' role makes
them act interestingly...And can avoid the CIV2 AI build bug...

Boco's spy/bomber/Manhattan idea is tops! Also he mentions many other very
excellent concepts, like settler/AA units...Cool, I never thought of that!



Enticing an AI attacker to use the correct units is a delicate science.
As Harry points out, if high-cost units are present, then the AI will seek them.

Also, tanks are a double edged sword. I usually want them to be vulnerable to
infantry, as is historically correct, but at the same time, want them tobe a primary
build unit for the AI, which will always build the most mobile and strong unit.

Sadly, then the infantry gets neglected by the AI, who always favours defence units and airpower.

On the other hand, if the infantry or bombers are strong, then the armour
never get built either by the AI...Gah! I am going to seek a new way of
mixing unit att/def with hp/fp to create a system of air/land/sea that
entices the AI to make use of traditional unit slots/properties but at
the same time, enables the AI to churn out a varied mix of units too.

I hope the AI has enough smarts to work it out. That is the trick!

Perhaps making defence infantry the cheapest, and armour/infantry the
same in att/def, but with wildy different hp/fp is the answer to making
the AI choose the correct units...And making airpower weak, but with
large hp/fp values is a good way of not over-egging things...

[techumseh]

Curt, I didn't mean to give you a migrane!

Perhaps this is one of the last great unsolved mysteries of Civ2. The ratio of costs to unit strengths is likely part of the equation. In vanilla civ, the AI builds marines, howitzers and tanks (all with a '0' role), and it builds mountain, mech inf and rifle units (with '1' role).

It builds air and naval units more or less in proportion to what's available, ie. if there are three types of naval units that can be built, and three coastal cities, then the AI will tend to build one of each, rather than 3 of one type.

There's a trick to finding out what the AI will build. Use cheat mode and click 'view entire map'. Click on an other civ's city and look at the build menu. Only those units which the AI will consider building actually appear. But if you click 'change human civilization' to the new civ, all buildable units will appear. Just because the human player CAN build them doesn't mean the AI will.

That's pretty much all I know about this. It's been a real problem getting all the units buildable by the AI in Roman Civil Wars. I've been fiddling with it for about a month now.

[techumseh]

Originally posted by jim panse
@ Strategical Bombing: I would suggest the same as Eivind & Boco - flying spies

@ Tech: How is this airbase thing working in RCWs? Doesn´t the airbase improvement also add a RR to the square? Could you please explain this further?

JP: Boco's right, there's no rwy bonus associated with airbases in ToT.

[curtsibling]

Originally posted by techumseh
Curt, I didn't mean to give you a migrane!

Heh!

In fact, these kind of things are what fill my thoughts all the time...!

On the bus, or waiting for FTP uploads to clear, I sit and wonder
at what components the perfect CIV2 WW2 scenario would need!

And if there is any place to uncover great ideas,
it is right here on this thread with you fellows!

[our_man]

For an example of paradropping spies - check out Mike Jeszenka's 2nd Reich scenario. He had a paradropping 1 MP all-as-road unit with spy abilities for the German zeppelin.

[techumseh]

@Curt - where might I find the latest version? Does it have sounds?

@our_man - it would be interesting to see how the AI used it. In my experience, any unit with the paratrooper flag will only move by dropping into an empty enemy city.

[curtsibling]

@Tech:

The Dictator ToT version on the Evolution site is the most current:

http://www.evo-games.net/Home/viewtopic.php?t=678

There are a gallon of SFX with the scenario, but well worth the downloading time!