System for doing away with the AI in games.

[Fred98]

Imagine a game with hundreds, maybe thousands of units.

There are too many units for a player to control. Such a game might take years to play by PBEM.

Solution: Give orders to senior units and have the AI control the junior units. Nope. We all know that would never work because the AI will always be poor.

Better solution: A form of scripting. A player is presented with a list of choices. He makes choices.

As it is scripted, the AI will not play poorly, but instead will control the players units and do a good job of it.

And there are only a certain number of possible outcomes.

At the end of a given time period – a week, a month, 3 months, the computer looks at the outcome, and presents the player, with a new list of choices.

The choices would be like a form where you check boxes. Each box is checked "on" or "off".

The crux of the matter is that you cannot achieve everything because you don’t have the resources. So when you check a box, it might be that other boxes automatically become unchecked.

And so it is, a player could control a large number of units. But without the need to issue orders to each and every unit and without the concerns that the AI would handle the friendly troops poorly.

-

[Didz]

I'm curious as to what game would ever suffer from this problem.

Commanders never had direct control of thousands of units. For instance the French army in CoTD might contain 185,000 men but at most Napoleon only issues orders to about 24 and thats unhistorical.

The sensible approach is to use the historical command and control structure and limit direct orders to those units the players general would normally issue orders too.

To do otherwise is to by-pass the command structure of the army and thus fail to model it correctly in the game.

[OmegaOPS]

ORIGINAL: Didz

I'm curious as to what game would ever suffer from this problem.

Commanders never had direct control of thousands of units. For instance the French army in CoTD might contain 185,000 men but at most Napoleon only issues orders to about 24 and thats unhistorical.

The sensible approach is to use the historical command and control structure and limit direct orders to those units the players general would normally issue orders too.

To do otherwise is to by-pass the command structure of the army and thus fail to model it correctly in the game.

Theres a problem with that reasoning, your asking the AI to, well basicaly reason as a human would, last I heard free thought was hard to code -

Seriously - why any military command stucture works like it does is mans ability to interpet his orders and find solutions to carry out his mission -

[Didz]

ORIGINAL: OmegaOPS
Theres a problem with that reasoning, your asking the AI to, well basicaly reason as a human would, last I heard free thought was hard to code -

I fail to see the problem. CoTD copes fine with this and I don't see why other games can't do the same.

In fact the options available to subordinate commanders get exponentially more restricted as one descended the chain of command to the point where Battalion Commanders and Company Commanders really have little if any free thought options available to them at all.

Certainly in the case of CoTD I have never felt the urge to issue a command to one of the hundred of battalions that form my army let alone one of the thousands of companies and I rarely issue a command to a division.

[Boar]

ORIGINAL: Joe 98As it is scripted, the AI will not play poorly, but instead will control the players units and do a good job of it.

And there are only a certain number of possible outcomes.

At the end of a given time period – a week, a month, 3 months, the computer looks at the outcome, and presents the player, with a new list of choices.

The choices would be like a form where you check boxes. Each box is checked "on" or "off".

The crux of the matter is that you cannot achieve everything because you don’t have the resources. So when you check a box, it might be that other boxes automatically become unchecked.
-

Um, Joe, that sounds about as fun as developing an amortization table!

[Louis Jones]

You forget Boar he's a banker and gets off on all the numerical minutia.
I'm guessing on Saturday mornings, he balances the checkbooks of everyone on his street, and does the pookie dance when he finds them thirteen cents, that was entered wrong and thought lost forever .[;)]
Just kidding Joe but it does sound terribly boaring.[:)]

[EricGuitarJames]

It's the problem of balancing a realistic command structure with gameplay. Joe's idea seems an interesting one but it could reduce the player to the position of being a spectator for large sections of the game. I love HTTR and prefer to play the game issuing as few orders as I can to units lower than brigade level but even here I sometimes feel detached from the action and unable to influence what's actually happening in the game. Whilst this may be realistic it can be very frustrating.

[Fred98]

I knew this would confuse some. Of course I am NOT talking about a game played against the AI. I too prefer games against humans.

And I am NOT talking about a small scale game like HTTR.

But when a game has 1,000s of units, it will take hours to play a turn.

One option is to give orders to senior units and have the senior units control, the junior units. But we all know the AI would place the junior units badly. So that would not work.

I note that, in KP, in a game against the AI, the first turn of the AI is scripted.

Imagine, in a PBEM game, if you give orders to a senior unit and it controls the junior units – but the movements of the junior units are scripted! Then the junior units would move to the correct places.

“Ahhhh” I hear you say. “That could never work because the enemy might be somewhere else, and I want the junior units to move somewhere else”.

Of course the solution is that EVERY move is scripted – then my junior units move to the correct place – to face my opponents junior units.

“Ahhhh” I hear you say. “If the game is totally scripted, what is the use of it?”

Answer: At the start of each turn you, and your PBEM opponent, are presented with a list of choices. Given that you choose item “4” and he chooses “7” a certain script would run.

The choices would be generic, such as “You lot will secure all the ground out to “X” kilometers”

I am aware that it would take many thousands of scripts to account for all the possibilities. But it is the only way we could play a game with 1,000s of units on the map.

I have been following the production of the upcoming War in the Pacific. It will be a great game but I fear it will be too large to actually play the long scenario.

[Fred98]

And a script would have no variables. Example: Given that its turn 7 and I choose option 5 and my opponent chooses option 2, script number “XX44” will run.

There are no variables, no dice, so the result is predetermined in the script.

But there are so many scripts, a player would need to play 100 times before he sees script number “XX44” ever run again. To the player, the outcome would be different in each and every game. To the player the result would be variable.

[dinsdale]

Highway To The Reich manages to cope with subordinates carrying out commands and transferring them meaningfully down the chain of command. Even the old Talonsoft Battlegrounds games managed to do a somewhat decent job 9 years ago.

[*Lava*]

Hi!

As I understand it, from a programming angle, the problem is not subordinate units obeying the orders of senior ones (i.e, corps obeying the orders of an army commander). That has been done and found to be a fairly realistic simulation. The problem it seems is when you start moving lower into the command structure. That is, I as the Army Commander (and human) can give orders to a computer controlled Corps Commander (which he will execute). The difficulty, and as I understand yet unsolved problem, is how to get the computer controlled Corps Commander to properly command units under him (divisions).

Ray (alias Lava)

[Didz]

I can see where you are coming from with games like War in the Pacific. I played UV for a while and found myself spending most of my time playing a role somewhere at the level of logistic's officer.

But as I recall UV never required the movement of individual counters, you merely issued objectives to those counters and I believe War in the Pacific will be the same. So it will be a one off task issuing initial orders and then modifying them as required. Indeed the biggest problem with UV was the stupidity of the AI controlled commanders who I found it necessary to turn off most of the time.

The classic was the twat who decided it was a good idea to sail his CV TF within air range of Rabaul.

[dinsdale]

ORIGINAL: Lava
The difficulty, and as I understand yet unsolved problem, is how to get the computer controlled Corps Commander to properly command units under him (divisions).

Yes that does appear to be difficult to achieve, but it's been done at a lower level for human players, and (though mostly poorly) as the programmed opponent.

AI has lagged years behind other advances as 1) It's not a flashy enough bullet point to put on the box and marketing material, and 2) It's a specialized skill which pays much more in the business sector.

I don't think that the problem is either technology or cost, but rather that commitment to this area is traditionally ignored in favour of other features.

[*Lava*]

ORIGINAL: dinsdale

I don't think that the problem is either technology or cost, but rather that commitment to this area is traditionally ignored in favour of other features.

True indeed.

However, if the publishers could for some reason identify my buying trends, they would see that my buying decisions are no longer based on hardware advances (i.e., video technology - the other features I assume you are alluding to).

As an example, I completely enjoyed playing Max Payne, but I determined that my computer would not run Max Payne 2 sufficiently to buy the game. So I was faced with a decision as to whether or not to upgrade my CPU. That decision was easy. Why upgrade for a product which offers the same gameplay and only offers better graphics? The same can be said for UT 2004. I had a lot of fun with UT 2003, and the thought of having a new version of the game with vehicles (added content, not just better graphics), nearly pushed me into upgrading... but as it turned out, it was not sufficient for me to do so.

I am preparing to upgrade in the very near future. But the fact of the matter is, the gaming industry no longer has the same influence on me as they did in the past because they are not presenting a sufficient "package" of content (graphics, innovation, AI improvements, etc).

Ray (alias Lava)