Random Axis Subs Starting Position

[Edwin P.]

In playing several games as the Allies one always knows where the German wolf packs start from in the Atlantic Ocean. This makes it all too easy to locate and sink them.

To rectify this I suggest three modifications:

1 - Random Starting Location of Subs (when FOW is active)

2 - AI Improvements

3 - Subs Can Transit the Gibraltor Straits if No Naval Units are stationed there.

1>It would be more interesting if this starting position was randomized when Fog of War is active. I would also like to see a chance for the submarine unit in the Baltic sea to be deployed in the Atlantic.

For example:

1. Current Starting Position (30%)

2. Starting Position #2 (15%) - Near Coast of Canada (chance to sink Canadian Army!)

3. Starting Position #3 (15%) - Near Coast of US

4. Starting Postion with 3rd sub unit (10%)

5. Starting Position #2 with 3rd sub unit (10%)

6. Starting Position #3 with 3rd sub unit (10%)

7. Starting Position #4 (10%)- South Atlantic

2> Of course, if time permits the AI for use of the subs should be upgraded as the AI always seeks to move the subs towards Germany and into the Jaws of a waiting UK fleet.

How would I do this? I would have the AI track how many of the UK Carriers and Surface fleet have been sunk. The more UK naval units that have been sunk the more aggresive the AI should be. If no naval units have been sunk the AI subs should adopt a "Hide to Fight Another Day" strategy, perhaps moving to the South Atlantic where they can intercept transports heading to Egypt or forming an anti invasion wall off the Western French Coast.

3> Subs should also be allowed to transit, but not end a turn in, the straits of Gibralter.

I would allow Subs to transit the straits with no damage if the straits are not occupied by an Allied Naval Unit and no land units occupy the rock of Gibraltor.

If land units occupy the Rock of Gibralter but no naval units occupy the port I would allow the Land unit to make a free attack on the sub unit when it is transiting the straits.

[ April 13, 2003, 05:31 PM: Message edited by: Edwin P. ]

[JerseyJohn]

Edwin

Good ideas, especially random placement. I think there should be a number of Axis choices about the U-boats destinations otherwise we're back with the Allied Player knowing what the AI will attempt according to specific circumstances.

The Battle of the Atlantic part of the game still needs to be revised, however. It ought to include U-boat supply ships and replenishment through milk-cow subs. Several players have also opted for surface raiders capable of depleting convoy MPP.

[Edwin P.]

JerseyJohn

I agree that the AI needs to have serveral strategies, otherwise, as you said the human player will know what the AI response is going to be.

The question is what should these strategies be? and in what situation should they be used?

Example:

1> If the Straits of Gibralter can be crossed by subs then the strategy would be;

If UK Navy is sailing the Mediterrean:

% - for German subs to enter Med to help Italian fleet sink UK Navy.

% - for Italian subs to enter atlantic.

2> If the UK Navy is not sailing in the Mediterrean and if the Straits of Gibralter are taken by the Axis then there should be a % that Italy will adopt an Atlantic First Strategy.

Atlantic First Strategy - Send Italian fleet to the Atlantic to blockade the UK and focus production on ships to strengthen its navy in the Atlantic.

-------------------------------------------------

I also like your idea of expanding the war for the Atlantic with supply ships. Possibly with a supply ship appearing on a random hex for X number of turns. The Axis player would know where that sea hex is. Any surface/sub unit that reachs this hex would be restored to full strength. Of course, if the allies find that hex first the supply ship would be destroyed.

[ April 13, 2003, 06:59 PM: Message edited by: Edwin P. ]

[JerseyJohn]

Edwin,

Interesting material.

It would be extremely difficult for the Italians, working alone in the Mediteranean, to reduce and capture Gibraltar, especially if they don't build a few aircraft carriers, which they are unlikely to do. Also, while they were attacking it I think any UK player would move his Eastern Mediteranean Squadron behind them perhaps while bringing a few ships down from the Atlantic, and decimate the Italian ships while they were weakened and low on supplies. Of course, if their were Axis aircraft stationed in force on Sicily the operation would be protected. It would either have to be a large cooperative effort with Germany, or Italy would need to be very strong.

On thing we're never effected by is the need to react to the Japanese. For example, when the Pacific war broke out in late 1941 the British were suddenly obliged to send capital ships to the Pacific, where they had nothing larger than cruisers. They designated a force that was to consist of the BB Prince of Wales, the BC Repulse -- both ships top speed was 38 knots, and I believe the medium sized carrier Hermes. The carrier never quite joined the expedition and we all know what happened to Prince of Wales and Repulse.

In SC there is no Pacific interaction. Of course, as was pointed out by another poster, this also works against the game as most of the troops sent to North Africa were Common Wealth -- Indians, South Africans, Australians and New Zealanders, none of whom got there via England.

Regarding Atlantic subs. As was discussed at legnth in the North Atlantic Forum (please see FAQs, my first entry, for that and other Topic links), a system is needed to move subs and other Axis ships along Greenland to Northern Norway, down the fjiords, in and out of the Baltic, as was done in real life. We all know how hopeless it is to try and move German units past Scapa Flow in the game -- unless you've already, managed -- somehow -- to sink the Royal Navy!

Slipping Italian subs into the Atlantic and German U-boats into the Mediteranean was an interesting maneuver in the actual war, always done by individual vessels as opposed to the groups of subs represented in our unit counters. It probably couldn't have been done without Spanish cooperation; Italian frogmen were housed in a supposedly scuttled merchantman, no doubt they passed information along advising the submariners of when the time was right for such a run.

A German U-boat sank the old BB Barham near Libya about the time Italian Frogmen put two BBs of the same class out of commission in Alexandria (ironically sinking to the deck level, Italian agents watching through binoculars never realized their frogmen, who'd been captured, had succeeded).

Generally speaking I don't know if it's overly important in game terms. U-boats moving to the Atlantic side of Gibraltar can help reduce Mediteranean Convoy MPPs as though they'd passed through the passage. Also, if the port is reduced to zero Axis surface ships and subs can be passed along, for those who really want to make the sea passage. A third way, is if Germany posseses a Mediteranean port (say Toulon or Athens) it can make U'boats and other ships that would be released directly into the Mediteranean Sea.

Germany ran supply and weather ships in the Atlantic until fairly late in the war, I believe it was early 1944 that the last of them, a meteorological station, was sunk or captured. One of them yeilded valuable Enigma inoformation and the loss of weather forcasting ability cost them during the Normandy Invasion when German predictions were for heavy storms throughout the first half of June.

[ April 13, 2003, 09:56 PM: Message edited by: JerseyJohn ]

[Liam]

How subs were accurately portrayed. Make subs in this game weaker towards surface ships period. Although do not have a surprise contact for subs. They're always submerged just in case and they don't have to engage the enemy if they don't wish...or at least make a percentage rate of being caught. How long does it take for a sub to charge it's batteries at night and dive back down? Not long they keep a keen eye out as they're vulnerable on the surface! Focus the sub on a MPP intercept role, a certian # of hexes between the main country port and the resource being blocked by subs cutting MPPs and supply to units abroad unable to get staples for an army to fight. A higher percentage intercept rate for supplies in otherwards. An added tech, anti-sub tech that if Destroyers are not inluded and specialized sea planes; something that allow for convoys to be more elusive and sub hunting ships to detect them better...rather than auto detecting them and destroying them out-right...

[Maximillian214]

U-Boats actually ran on the surface much more than that. They would have been surfaced most of the night. At least during early and mid-war, they would have been surfaced for a significant part of the day.

The batteries just weren't good enough to spend that much time running submerged. They used the diesels as much as possible and snorkel wasn't introduced until mid to late war (IIRC).

[jon_j_rambo]

I like the idea of different set-ups; but NOT the idea of random set-ups. Random set-ups will open up the "you got lucky" thread.

Let countries, & their commanders place their military pieces as they see fit. (See Squad Leader & 3R).

[Edwin P.]

I agree that for multi-player games the commanders should be able to pick the setups of subs, as they would have postioned them before declaring war.

For games against the AI, random setups appear to be the only way to go.

[Edwin P.]

Jersey John

Per your comments ragarding moving subs around Greenland to to avoid the Scapa Flow, could this be done in a similar fashion to the movement of ships via the South Atlantic and Egypt?

Have one hex next to Norway that Ships could enter to appear 2 turns later to the west of Greenland on a random hex along the northern border

To stop this the allies would have to station ships within range of the Norwegian coast. A most dangerous proposition if Germany decides to station air units there, or station 3 ships on the sea to the West coast of Greenland.

[Edwin P.]

Liam

I like your idea of not having a surprise contact for subs. If they saw a surface fleet they would dive, dive, dive.

[JerseyJohn]

Edwin

Yes, that's exactly the way I think it would have to be.

The Bismarck campaign shows how difficult it was to successfully shadow even a large Battleship ranging across the Atlantic. The Germans broke out along this route twice with four capital ships (1940 Scharnhorst and Gneisnau and 1941 with Bismarck and Prinz Eugen [Heavy Cruiser]).

It was also used regularly by outbound U-boats. Once at sea, many of them replenished at bases in France.

Bismarck and Prinz Eugen, while passing through the Denmark Straights between Greenland and Iceland was engaged in a major battle, sinking HMS Hood and severely damaging Prince of Wales, being hit herself several times by Prince of Wales 14" shells. When the firing ceased she continued south and managed to elude two faster, radar equipped, British cruisers attempting to shadow her despite the fact she was trailing an oil slick!

If it hadn't been for Adm Gunther Lutgens' inexplicable and legnthy radio transmission heading SE toward France, Bismarck would actually have reached Brest safely. This isn't speculation, it's a fact, the Royal Navy was looking north instead of south, the Bismarck gave it's own position away literally hours before being beyond striking distance.

Which is not to insinuate breaking into the Atlantic was automatically accomplished, especially late in the war. As a rule of thumb it was nearly always accomplished prior to the activation of Iceland as a naval air station, which I believe occurred in early 1943, following Germany's banner U-boat year of 42.

Once the entire North Atlantic was covered by either land based long range aircraft or escort carrier hunting groups, the theater became increasingly unsafe for U-boats.

Ironically, this occurred after Donitz replaced Raeder as CinC of the Kriegsmarine. Consequently, Germany only began sending large numbers of U-boats to sea when it was least safe for them to do so. Consequently, most German U-boats were sunk without having sunk anything themselves! A comparatively small number of highly skilled captains and their veteran crews were able to survive even under the always worsening conditions, but a very high proportion of the late arriving rookies were sunk on their outward maiden voyage; many in the North Sea.

Getting back to the game topic, I agree with your idea of having it start in the Baltic and exit somewhere in the NW Atlantic. There would need to be a system where Allied ships in the region would cause the emerging Axis ships to appear farther south of their own position. Perhaps the likelihood of this occurring later in the war would diminish, simulating the increase in Anglo/American numbers and anti-submarine capabilities.

[Desert Dave]

As originally posted by JerseyJohn:

Getting back to the game topic, I agree with your idea of having it start in the Baltic and exit somewhere in the NW Atlantic. There would need to be a system where Allied ships in the region would cause the emerging Axis ships to appear farther south of their own position. Perhaps the likelihood of this occurring later in the war would diminish, simulating the increase in Anglo/American numbers and anti-submarine capabilities.

JJ, I think you and a few others are making this a bit too complicated, yes?

I believe Hubert has endorsed an idea proposed long ago, where the subs would have "silent running" and that would allow them to move to & fro with a greater degree of freedom.

Of course, they couldn't do any damage while in that silent state, but at least they could survive long enough to get to the Atlantic in the first place.

I wouldn't favor having special "entry-exit" hexes however. Presumably we will have a larger map in SC2 anyway, and so the Royal Navy will have to cover a somewhat larger ocean.

[JerseyJohn]

Immer, I have no idea what SC 2 covers or looks like or how it will differ from the present game. The only information either myself or the others can go on is the fact that naval movement in the Atlantic is sealed off at Scapa Flow.

That's the only problem we can address and the map we have in the present game is the only one we can suggest sulutions for.

[ April 16, 2003, 04:38 AM: Message edited by: JerseyJohn ]

[John DiFool]

JJ, the problem with that idea is that we aren't

dealing with true submarines here: before the

Type XXI, German subs transited to and from the

shipping lanes on the surface. They didn't

use 'silent running' at such times [that was for

when they were tactically evading escorts

who had already located their general position].

But they tended to transit singly, only forming

a coherent wolfpack once in the combat area. I

hope Hubert remembers my idea about 'dispersed'

wolfpacks, during which time they would take less

damage, be detected less often, and also do less

damage to merchants and warships. It would also

provide for a strategic choice for the Germans:

stay dispersed, but sink fewer MPPs, or go into

wolfpack mode and risk a higher attrition rate

on the subs?

John DiFool

[ April 16, 2003, 12:24 PM: Message edited by: John DiFool ]

[JerseyJohn]

John,

Yes, but we have no way of representing them on this scale as anything but multi-sub units. The assumption would be that they all moved singly along that, shall we call it Artic Rim route, to join other subs along the shipping lanes.

Aside from U-boats, however, we have no feasable route to move capital ships from the Baltic to the Atlantic other than to run them through the English Channel -- a feat that was only accomplished once, and that was only done to bring three big ships back to Germany.

Germany had a very simple method of bypassing Britain, conceal vessels in the innumerable fjords and islands along the Norwegian coast, then move them carefully along Arctic coastal waters till they can run the Denmark Straights. It was done routinely and was, in fact, the only practical way Germany had of exiting it's ships and subs into the Atlantic. In this game it can't be done at all. I'd say something needs correcting here.

Care to silent run a Battleship past Scapa Flow?

[Edwin P.]

One note on adding special entry/exit hexes to the map.

Although coding it would be relatively easy, as it could use the same core code that was designed for the South Atlantic-Suez route, writing an AI routine to take advantage of it would be much harder.

Currently the South Atlantic Suez route is not used by the AI at all, and the AI never attempts to take Iraq (which opens up the strategic opportunity to operate Air Units between the UK and Russia, a major threat to any German conquest of Russia).

[JerseyJohn]

Edwin

Good points, but I have to admit it's been a long time since I've considered the AI in any of my game suggestions. It's a punching bag. The AI won't invade Britain as the Axis, it won't use Cape of Good Hope loop as the Allies, as you say, it won't invade Iraq, nor will it invade Scandinavia. Either the AI needs to be vastly improved within the present game mechanics or the game mechanics have to change to make decision making a simpler process for the AI, or the game has to continue being considered as a person to person game possessing a very weak AI.

On the assumption the AI will remain what it is, I envision the game I'd like to play and don't attempt to fit the AI into it. Personally I'd like to have a very high quality computer opponent, but I doubt we'll be seeing it for a long time.

When it does happen the loudest complaint will be that the AI is too hard to beat.

[ April 17, 2003, 03:36 AM: Message edited by: JerseyJohn ]

[Edwin P.]

JerseyJohn

Its true that the AI is weak (after you have played the game for a while, but I like the game as its simplicity allows you to focus on the big picture while completing a game in one sitting, and now avoid using tricks that the AI always falls for). I guess that I will have to start playing games against a human opponent soon.

Perhaps the only way to get a great AI is to build one that allows for player modifications and easy updates by the programmer.

-------------------------------------------------

Without knowing how the software is designed, I would probably start along this route by adding a simple change such as allowing players to create a research allocation file for the initial research chits and the chance of this strategy being used.

Example:

File Name: RESEARCH020.TXT

Name "antitank"

Country "russia"

Research 30000000

Weight 2

StrategyType "LAND,MIXED"

StrategyName

The file name would always be a text file starting with the word research followed by a number in the range of 000 to 999.

Country is the name of the country. Research is the allocation of chits: 3 in the first position is 3 chits in AntiTank and 2 is the weight assigned to this strategy that could be in the range of 1 to 5. The program would sum up all the weights and assign percentages based on this.

The term StrategyType would allow the AI to link this selection to selected AI strategies. Ie, If the UK AI decided to go for an Air Focused war strategy the AI would only consider starting tech allocations strategies including the StrategyType field value "AIR" and exclude strategies not marked as such.

StrategyName would allow players/designers to link the tech strategy to a particular battle plan.

----------------------------------------------

Then I would probably try adding a more complex French strategy mod file for the initial defense of France. Here, the programmer would provide a series of AI rountines that the players could use/not use to build an opening strategy.

Example:

File Name: FranceOpen001.txt

Name: "ESCAPE"

Description: "Flee to British Ports"

ALGERIA: Custom2546

LEBANON: C5

NavyMed: N2

NavyAtl: N2

NavyD: 2345,2456

FRANCE: F3

FRANCEU: C234-H2234, A345-H2235

Air: A3

Production: Corp, HQ, Corp

OpeningAiLandStrategy: "DefendFrance"

OpeningAiNavalStrategy: "ControlMed"

Weight: 5

Thus for the french colonies the choices would be;

1 stay in place, 2 move to France, 3 move to UK, 4 move to Malta, 5 Move to Egypt, 6 Disband, CustomXXXX. Custom2546 would tell the unit to move to hex2546. The programmer could add more options to the library for the player to select from.

NavyMed would provide choices for the initial French navy movement strategy for its Med Units: N1 - Move to Atlantic, N2 - Reinforce British Fleet in Med, N3 - Nothing

NavyAtl specifies the options for the French forces in the Atlantic: N1 - Move to Atlantic, N2 - Reinforce UK Fleet in Med.

NavyD would specify which navy ships to disband, if any.

FRANCE specifies the general strategy for French units to follow. F3 - Move armies to Belgium border & replace armies in lower two fortresses with Corps, F4 - Evacuate French forces to UK while leaving a delaying force in place.

FRANCEU specifies movement instrutions for particular units such as Move Corps 234 to Hex 2345.

Air sets the initial stategy for the French Air Unit: A1- Stay in Place, A2 - Move to England, A3 - Move to Brest, D - Disband, CustomXXXX.

[ April 17, 2003, 09:18 AM: Message edited by: Edwin P. ]

[JerseyJohn]

Edwin

I haven't touched the stuff except for a little Coball and Fortran programming in the late sixties, the bad old days when holorith (? IBM) cards had to be punched and sent through a hopper. Never got into Computer Inteligence, so I have only the vaguest idea of how to improve something like this. Over the years I've noticed that AI doesn't respond well to either complexity or an overabundance of options. Although this game plays in a simple manner, the decisions involved are often involve very complex reasoning. Much of it intuitive, a task any AI would find difficult if not impossible.

Having said that, I have to state your idea looks very good. Of course, I have no idea what Hubert's AI looks like to the computer. When I do play the AI I leave FOW off for the computer's sake and, after following the AI's reasoning process (a line of writing that runs below the screen) I've come to feel it usually does things on good general principles. The problem is, this form of Artificial Inteligence can't match Human Inteligence at this point in time. Which is good because the day it can I'll have to start watching movies like The Forbin Project, War Games and Terminator and having nightmares about the damn thing electrocuting me as a means of winning the game.

[ April 17, 2003, 03:56 PM: Message edited by: JerseyJohn ]

[Edwin P.]

I agree that good AI programming is the hardest thing to do. Hubert's AI does a good job at general attacks on a wide front but tends to avoid reinforcing a critical area to obtain ensure success and never considers taking Norway, Sweden, Switzerland, Iraq, Algeria, or Turkey.

It also seems to ignore opportunties when they are available but are out of a specified range. Example: Cairo, Suez, Canada are not threatened even if their are no Allied naval forces in range to stop an invasion force.

Sometimes you need to support a good general AI with specific AI that kicks in under certain cirumstances.

Perhaps one way for Fury Software to support an improved AI would be to sell the software with an AI subscription option priced at $12 to $18 per year that would include bi-weekly/monthly AI updates available to subscribers and support a limited amount of software development after the release date.

The key question is - is this a viable marketing option? Does SC have a large enough customer base to make this worthwile from a profit/loss point of view? Would SC players find this a worhwhile investment?

Would such a change increase the market demand for this product? I believe it might, if it was handled in the proper way.

The proper way?

- Email subscriber newsletter highlighting accepted AI contributions by subscribers.

- Subscriber contributor credits included in the software at the close of an AI winning game.

- Press Releases highlighting the upgrade.

- Semi-annual AI design contest for subscribers.

- Online AI design course available for subscribers.

- SC AI design manual/notes for subscribers

Just an idea for consideration.

[ April 17, 2003, 08:10 PM: Message edited by: Edwin P. ]

[John DiFool]

It would be interesting for people to pit their

AIs against each other in AI-only battles! I

wonder if Rambo's would beat everyone elses...

John DiFool

[JerseyJohn]

Edwin

It might well be feasable and cheaper as opposed to having to buy versions 2 and 3 of essentially the same game as was done by projects such as Civilization, especially in it's Civ II series. Though I have to give Cid Meir's outfit credit for changing Civ III with a very good free download.

And, if the $12.00 a year resulted in a game with really great AI it would probably be worth it.

What I find really encouraging is the fact Hubert is obviously a perfectionist, as evidenced even today is upgrade of the V1.07 patch. SC could actually be a game that improves and evolves significantly for years to come. A far cry from a couple of years back, when WW II strategy games were supposedly dead.

[Edwin P.]

Jersey John

I agree with your thoughts, that SC could continue to evolve and that $12 a year would be a bargain. It would also be a bargain for the developer as they would not have to build a new game from scratch every 12 to 18 months.

The key is that they need to increase the income generated by the software. From the size of this forum I would guess that the existing customer base is not very large.

A subscription product would increase their income on each sale by about 100% over two years ($25 + $12 year one + $12 year two). At the same time it would help to attract new customers with the continual publicity and word of mouth. The key (in my opinion) is to make it a community with communication, interaction, competition and lots of free publicity. For if people don't know about your software they can't buy it.

[ April 17, 2003, 11:54 PM: Message edited by: Edwin P. ]

[pzgndr]

The AI is robust enough to handle most of things everyone keeps asking for. It needs some tweaking here and there and some additional strategies loaded, but Hubert's fuzzy logic AI seems to have everything in place to become very good in SC2. Assuming, of course, that SC2 does not become significantly more complicated. Folks forget about that - add a hundred new features and your AI reprogramming task becomes impossible.

Setups could be adjusted in several ways - a purely random setup where the AI gets to position units wherever (for better or worse, like in 3R), a random offset for each unit to shift position into adjacent hexes, or a collection of fixed setups that get randomly selected when needed. Except for a few canned first-move strategies against Italy and USSR, I have not found the setups to be particularly annoying. This is a game of long-term consequences, and the setups tend to be a very minor part of all that.

[Edwin P.]

Bill,

I agree with your comments that as you make a game more complex you really increase the difficulty of programming a good AI.

I also agree that the AI needs tweaking and strategies added to it. Such strategies naturally come to light over time from watching how intelligent humans wage war.

Currently, the AI for SC has not evolved much since the software has been released, although several fixes have been released that improved the basic game engine.

As for the tweaking, I would like to see the AI better concentrate its forces to achieve victory in an area, know when to avoid a trap, and when to sieze an ungarded Canada or Gibralter. For strategics I would like to see canned attacks for Belgium, Norway, Russia, Greece and strategies for taking Norway-Sweden, Iraq or defending France.

ie France:

Strategy 1: Historical

Strategy 2: Disband Navy, Reinforce Army

Strategy 3: Delay while withdrawing troops to England

Stragegy 4: Historical + Send French Navy to Join with British Navy in Sinking the Italian Fleet.