Matrix Games Forums

Forums  Register  Login  Photo Gallery  Member List  Search  Calendars  FAQ 

My Profile  Inbox  Address Book  My Subscription  My Forums  Log Out

Fighter escort protection effectiveness questions

 
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] >> Advanced Tactics Series >> The War Room >> Fighter escort protection effectiveness questions Page: [1]
Login
Message << Older Topic   Newer Topic >>
Fighter escort protection effectiveness questions - 11/8/2009 1:27:24 AM   
dominickpa

 

Posts: 49
Joined: 3/9/2008
Status: offline
To provide best protection against fighter defenders, when attacking a city location with level and/or dive bombers, should the fighter escorts be in:

1. The same subformation?

2. Separate simultaneous attacking subformations?

3. Or, should fighters be sent in first to clear out fighter defenses, then bombers?

Which of these three choices do others use, or does it not make any difference in the same turn?

And does your choice depend upon how much flak is defending the city?

When flak is heavy, it would seem that dive bombers first (best to kill and defend against art), then fighters, then level bombers. If you can only reach a city with long-range fighters and level bombers, then choice 3?

Also, does anyone know if there is a difference in merely sending fighters over a city for recon, vs. attacking it? It would seem that a recon mission would keep the fighters concentrated on intercepting fighter defenses, but also subject to losses from flak. Therefore, in the recon role fighters may be more effective at killing fighters than if attacking ground forces at the same time.



Post #: 1
RE: Fighter escort protection effectiveness questions - 11/9/2009 9:57:27 PM   
GrumpyMel

 

Posts: 864
Joined: 12/28/2007
Status: offline
If possible I send in the fighters first to do a fighter sweep and clear the way for bombers. It's a bit of a judgement call. If the spot is heavly defended with Flak you may want to send your air in all in one strike. Note that if you have a good idea where all the enemy interceptor groups are based, you don't neccesarly have to do the fighter sweep against the spot you intend to hit with your bombers. You can hit another, less well defended spot with your fighter sweep to draw the enemy into a dogfight away from thier Flak and then send in the bombers against the real target when the enemy interceptors have been cleared. Of course you want to be carefull about not misjudging where all the enemy air groups are if you do that, as your bombers could run into a nasty surprise otherwise. Usualy best to send a lone fighter in on recon before you launch the bombers just to see what might respond.

Note that recon isn't really that effective for dogfighting, as it only lasts a single round. Thus you can't be assured that the battle will last long enough for you to knock out the enemy interceptors. Generaly you want to do a strike of some type with your fighters to draw the enemy into a dogfight....but it's best to pick a target that doesn't have much flak cover available when doing so.


(in reply to dominickpa)
Post #: 2
RE: Fighter escort protection effectiveness questions - 11/10/2009 1:57:49 AM   
dominickpa

 

Posts: 49
Joined: 3/9/2008
Status: offline
Thanks, for several great suggestions!  Especially using a lone fighter to scout.  So far, it seems that the AI is not much of a match for a strong air force in my user created scenarios.   

(in reply to GrumpyMel)
Post #: 3
RE: Fighter escort protection effectiveness questions - 11/11/2009 2:12:49 PM   
Joshuatree

 

Posts: 507
Joined: 12/30/2007
From: Netherlands
Status: offline
Personally I use separate formations, because upgraded fighters have a much larger reach than divebombers, up to 14 against 10 for the divebombers. If you outnumber the Flak you do have a chance against it, if the Flak outnumbers you you're into a lot of trouble.
Also I've noticed that only when the readiness is low, that's when the divebombers really start to make casualties. So Artillery shoots away a few times, and then I send in the fighterboys.

(in reply to dominickpa)
Post #: 4
RE: Fighter escort protection effectiveness questions - 11/11/2009 6:04:15 PM   
dominickpa

 

Posts: 49
Joined: 3/9/2008
Status: offline
Good idea Joshuatree.  In addition, upgraded fighters usually have good combat experience from previous battles.

I really like using the air force in this game.  In my latest mirrored, normal AI scenario I have destroyed 5000 ground units, with only 10 air and 9 ground casualties while approaching my first enemy city for capture.  Will probably switch over to an A+ opponent next to experience more of a challenge for this same scenario, as this battle is basically curtains for the normal AI. 


(in reply to Joshuatree)
Post #: 5
RE: Fighter escort protection effectiveness questions - 11/11/2009 9:27:05 PM   
Joshuatree

 

Posts: 507
Joined: 12/30/2007
From: Netherlands
Status: offline
Yeah I forgot to mention that; try the AI+. It will use heavy tanks and more planes, and a lot more troops... but it still doesn't do suprise attacks, you know? No amphibious landings, no airborne attacks, AI plus means more and better enemy troops, not an improved AI... now that would be something! A turbo AI Why is it that computer programs can beat experienced chess players, but not wargamers??

(in reply to dominickpa)
Post #: 6
RE: Fighter escort protection effectiveness questions - 11/12/2009 5:00:55 AM   
explorer2

 

Posts: 465
Joined: 11/30/2007
Status: offline

quote:

ORIGINAL: Joshuatree
Why is it that computer programs can beat experienced chess players, but not wargamers??


My guess is it's for one of the same reasons we prefer wargames and not chess - wargames are far more complex. The variety of possibilities in chess, though very many, are limited by the small board and small number of pieces. In wargames, the options are virtually limitless. (And don't forget, the computers that have been programmed to beat humans in chess were supercomputers and had an entire staff of full time programmers working on it!)

(in reply to Joshuatree)
Post #: 7
RE: Fighter escort protection effectiveness questions - 11/12/2009 1:30:23 PM   
Joshuatree

 

Posts: 507
Joined: 12/30/2007
From: Netherlands
Status: offline
You're probably right Explorer. Ofcourse today I do have a supercomputer compared to say, 10 years ago, that's when they started to beat human chess players. But rules in chess are set in stone, whereas every wargame has it own set of rules. And, accordingly, every wargame programmer has to invent the wheel again, that is make a good AI. If only it were possible to make a complex wargame.... and then cut the AI rules/program/code in smaller and smaller parts until there is a resemblance with the rules of a chess game... would it then be possible to make a clever and cunning AI??

I have a cunning plan..


(in reply to explorer2)
Post #: 8
RE: Fighter escort protection effectiveness questions - 11/12/2009 4:37:43 PM   
GrumpyMel

 

Posts: 864
Joined: 12/28/2007
Status: offline
Chess is ALOT easier for a computer because number of variables it needs to take into account is far less then any but the simplest wargame. The board is simple and quite compact. The number of pieces on the board for both sides is small, well defined and the changes in them are predictable . The variety of pieces is also very small and the number of things/moves that any given piece can make is pretty limited as well. On top of that, the results of any given "combat" are 100% perdictable.... a piece can ALWAYS take another piece by moving into it's square.

Chess and games like it are much easier for computers to deal with because of the predictability. The computer "thinks" by calculating out every single possible move from it's current position, every single possible response to each of those moves, every single possible counter-response, etc.... out to as many turns as it's processing capacity can handle... and then calculates which move gives the largest number of "positive" responses from there.

Wargames tend to have so many variables that following that method would be pretty impossible. So instead they tend to follow a "programed opponent" type process. Where the programmer tends to give the A.I. sets of rules to follow (if you see X do Y).  For good A.I.'s they also might program in some variation in response to give the A.I. different personalties.

However it's tough to give an A.I. a good enough set of rules to account for every possible situation it might encounter and which response to give precedence when it encounters multiple different situations....or even tougher, (what humans are very good at) a partial response to those situations that might meet part of each priorty and has synergy. On top of that, a good human player...after a little experience with the A.I. will start to "learn" the A.I.'s rules....and start to be able to predict what it's likely to do....and being able to predict what an opponent will do is a large part of beating them.

That's why I pretty much play AT against human opponents exclusively these days. The A.I. is very decent as far as wargames go.... but still can't even begin to hold a candle to the craftiness, duplicity and adaptability of another human mind


(in reply to Joshuatree)
Post #: 9
RE: Fighter escort protection effectiveness questions - 11/12/2009 6:25:15 PM   
dominickpa

 

Posts: 49
Joined: 3/9/2008
Status: offline
quote:

ORIGINAL: Joshuatree

Yeah I forgot to mention that; try the AI+. It will use heavy tanks and more planes, and a lot more troops... but it still doesn't do surprise attacks, you know? No amphibious landings, no airborne attacks, AI plus means more and better enemy troops, not an improved AI... now that would be something! A turbo AI Why is it that computer programs can beat experienced chess players, but not wargamers??


Great discussion why chess is easier to program compared to war games!

Progress testing report: In the same mirrored scenario with AI+ my kill to loss ratio hasn't changed that much. There are more second and third level researched units coming at me, also seems at a faster pace and in greater numbers. But no surprises. At turn 23 I now have 2493 kills vs. 13 losses. If this continues for a few more turns, AI++ will be tried next. I should mention however, that my scenario is only 30 hexes wide and 60 hexes deep, with lots of plains. So essentially, my one airfield and other city air base can protect my ground defenders from getting chewed up, in this choke point-like defense scheme. My air force continues to dominate.

(in reply to Joshuatree)
Post #: 10
RE: Fighter escort protection effectiveness questions - 11/12/2009 8:47:36 PM   
Joshuatree

 

Posts: 507
Joined: 12/30/2007
From: Netherlands
Status: offline
Quite true Grumpy, there may be more variables in wargames, but the computing power has increased significantly also. So a modern day pc should be able to calculate it fast.... if the progammers are skillfull enough. Chess has seen many centuries of thinkers and highly intelligent players, while pc wargames do exist for... 20 years or so? Guess we're in our early stages now. Take a look at robot footbal for instance: http://site.dutchrobocup.com/

/quote

"The official goal of Robocup is:

By the year 2050, develop a team of fully autonomous humanoid robots that can win against the human world soccer champion team."

/end quote

If they're able to do something like that, maybe we have a skilled AI by that time too. Heck, what does the Pentagon use for an AI in their "games", their training software?

I'm not talking about creating a "supermastermind" in our pc's, that would be near impossible, but I am talking about the possibility of dumbing down a wargame until an AI can "understands" it better. And from there on improving the game, making it more complexe and adding more variables. Oh well till that time one can only dream of an AI making an airborne assault

@ Dominick: AI++ is no fun. It will send masses and masses of troops against you, and it doesn't make the game better, only more tedious. I use multiple AI's, with some a bonus in production.

(in reply to dominickpa)
Post #: 11
RE: Fighter escort protection effectiveness questions - 11/12/2009 9:19:46 PM   
dominickpa

 

Posts: 49
Joined: 3/9/2008
Status: offline

quote:

ORIGINAL: Joshuatree

Dominick: AI++ is no fun. It will send masses and masses of troops against you, and it doesn't make the game better, only more tedious. I use multiple AI's, with some a bonus in production.


You are probably right, I think I played against an A++ some time ago and remember it does become tedious slaudering while I advanced. Maybe it's time to think about playing against carbon units, er ... human opponents!

(in reply to Joshuatree)
Post #: 12
RE: Fighter escort protection effectiveness questions - 11/13/2009 3:41:08 PM   
GrumpyMel

 

Posts: 864
Joined: 12/28/2007
Status: offline
From what I understand, the big trend in A.I. these days (btw, this really isn't my area of expertiese, I just hear about some of this stuff tangentialy once in awhile) is "adpative A.I." That is computers learning the way simple organisms learn instincts.

The A.I. basicaly tries something at random, if it achieves a "positive result" it marks it down as a good response for those conditions and "remembers" it for future use. It basicaly runs iteration after iteration of these sorts of tests, refining it's "behavior" and responses after each one until it essentialy "learns" how to perform a particular function with pretty fair efficiency.

It takes an incredibly vast number of iterations for the A.I. to "learn" to do even simple tasks in this mode. Fortunately computers can process such repetitive tasks VERY quickly. It's still not the equivalent of a human brain but with enough "experience" it can learn to be pretty profficient at things and can even adapt somewhat to changing strategies.

This sort of A.I. has been around for a few years now, I believe, and has even started to make it's way into games a little bit.
(http://www.ercim.org/publication/Ercim_News/enw57/spronck.html)

The huge difference, of course, is in HOW an A.I. learns from it's experience. The A.I. learns basicaly the same way an ameoba learns. It just knows whether something "worked" or not. Human beings on the other hand have the ability to not just understand whether something worked or not... but to be able to understand and form theories on WHY it worked or not. So what we can learn from just a single iteration of something would take the A.I. billions of iterations.






(in reply to dominickpa)
Post #: 13
Page:   [1]
All Forums >> [New Releases from Matrix Games] >> Advanced Tactics Series >> The War Room >> Fighter escort protection effectiveness questions 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


Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

0.270