Artifical Intelligence for World in Flames

[mlees]

Deterministic means it can be determined exactly. The strength of an infantry unit is deterministic.
Probabilisitic means it depends on probability (rolling dice or drawing chits in the case of WIF). The land combat results table is probabilistic.

Ah! Thanks for taking the time to explaining that to me.

So what I was asking is whether under certain circumstances is the player guaranteed getting/not getting intel information? If you do this, you can learn about the enemy units.

The vague "image" I had in my mind was probalilistic, but that's because I am used to dice rolling. (Old PnP 1st ed. DnD alumni here.) I dont have an emotional stake either way.

The advantages of a deterministic system is that a meticulus player can utilise his units in such a way as to "spot" all the desired units, and thus makes for a very controllable and reassuring flow of play. Removes all "chance" from his plans of world conquest.

Also, very easy to see if the FOW system is working. If you place a unit someway as to ensure a "spot", and the computer does not reveal a FOW enshrouded unit, you know the code is goofed somewhere.

The characteristics of a deterministic system is a lot more influenced by the nature of the randomness of the system. Not everything works as planned, and the player will need to adapt his moves "on the fly" more often, as a result, because of failed "spots".

That will make it more frustrating for the overly meticulus player. (The new players should leave such an option off while learning the game.)

But would it add to suspense and surprise, making for a more fun and memorable game? For me it would. But I realise I am not neccessarily the standard mean.

For WiFFCon Tourney purposes, or for those who want the most "pure" WiF, than the option would be "off".

[Neilster]

Falling off the end of the list. Time to bump it for the new people.

Cheers, Neilster

[Nibelung]

biased probabilistic please. Its war, not chess, so most of the things should not take for granted. But we don't want to feel we are playing Russian Roulette too... Open ended dices can be a possibility.

[bredsjomagnus]

Well....
 
I´ve read through some of the posts and think that they contain many good ideas.
 
I do think that fog of war is an important element in war. If a commander knew the exact location and strength of his/her enemy it would be much easier to go to work [8|]. This goes for warfare in all times (Israel had quite some problems with the hizbolla milis just a few months ago, all because of bad intelligence).
 
Another part of me thinks that a game doesn´t get better just because i gets more realistic. Maybe even the opposite sometimes, if it gets to complicated. Just look at games like the Axis & Allies-serie that is both very unrealistic and great at the same time (of course it´s a matter of taste). But since Steve made clear that fog of war would be an opiton I think that it would be great if it was implemented.
 
As said before; one (as commander) should maybe know alot about the front but less about whats going on beyond. Just that simple and clear...[:'(]
 
One should also, I think, be able to activly disorientate with dummy units (like Rommel in north africa [:)]).
 
 
/Bredsjomagnus
-Im from Sweden you know-

[bredsjomagnus]

I wonder if it is possible to create an AI that learns from it´s mistakes och it´s good moves? Sounds creepy I know [X(]...
 
I´ve heard about a chess computer that did just that, and that it also therefore became more and more difficult to beat it.
 
Wouldn´t it be great to play against an AI that constantly evolv depending on earlier expirience?! (hmmm maybe not [:D]).
 

[SurrenderMonkey]

ORIGINAL: bredsjomagnus


One should also, I think, be able to activly disorientate with dummy units (like Rommel in north africa [:)]).


/Bredsjomagnus
-Im from Sweden you know-

But not at this scale. It would be the equivalent of not knowing that Rommel is there at all, which is silly.

[bredsjomagnus]

ORIGINAL: SurrenderMonkey

ORIGINAL: bredsjomagnus


One should also, I think, be able to activly disorientate with dummy units (like Rommel in north africa [:)]).


/Bredsjomagnus
-Im from Sweden you know-

But not at this scale. It would be the equivalent of not knowing that Rommel is there at all, which is silly.

Well of course you are right about that. What I ment is that instead of entire new units (dummie units) you can deliberatly try to fool your enemy that an already existing unit is stronger or weaker than it really is.

[Shannon V. OKeets]

ORIGINAL: Nibelung

biased probabilistic please. Its war, not chess, so most of the things should not take for granted. But we don't want to feel we are playing Russian Roulette too... Open ended dices can be a possibility.

I absolutely agree.

[Shannon V. OKeets]

ORIGINAL: bredsjomagnus

ORIGINAL: SurrenderMonkey

ORIGINAL: bredsjomagnus
One should also, I think, be able to activly disorientate with dummy units (like Rommel in north africa [:)]).


/Bredsjomagnus
-Im from Sweden you know-

But not at this scale. It would be the equivalent of not knowing that Rommel is there at all, which is silly.

Well of course you are right about that. What I ment is that instead of entire new units (dummie units) you can deliberatly try to fool your enemy that an already existing unit is stronger or weaker than it really is.

Fog of war is tricky. There has already been quite a bit of discussion on this topic, I have decided it is not for MWIF product 1 (#2 maybe).

[Nibelung]

oh, no Fog of War? I'm slightly disappointed by the gamey things it will do to the game [:(]

[Shannon V. OKeets]

ORIGINAL: Nibelung

oh, no Fog of War? I'm slightly disappointed by the gamey things it will do to the game [:(]

Well, one of the strong arguments against FOW was that the game has received decades of play in the over-the-board format. During that period innumerable changes were made to strengthen all aspects of the design - based on feedback from serious players (those with extensive experience).

This puts adding a FOW feature into the category of changing a fundamental principle of the game - something that should not be taken lightly. It deserves extensive analysis prior to design and implementation, followed by extensive playtest, and redesign and recoding if necessary.

Given the level of work that entails, and that WIF as a board game won "game of the year" and "game of the decade" honors, a straight implementation of WIF doesn't seem to be a bad idea.

[Arron69]

I totally agree. Wif is a very good game, as it is. And if the FOW can be made in Wif product 2, well that is something to look forward to.

Andi.

[Nibelung]

ok, but what about the rule of not looking at the counters under the top one? Possible, as an option, impossible?

[Shannon V. OKeets]

ORIGINAL: Nibelung
ok, but what about the rule of not looking at the counters under the top one? Possible, as an option, impossible?

This is a fairly common "house rule", but it lends itself to a lot of gamemanship (do you put the weakest counter on top or the strongest?).

Regardless, it would be an extra feature that is not part of standard WIF FE. My policy (to prevent feature creep) is 'No' to all house rules. Players will just have to settle on choosing from the current list of 81 optional rules, 11 scenarios, 4 modes of play, and choice of side/major power.

[ptey]

Recently there have been raised some points, about whatever the ai should do in various situations, is highly dependent upon the options selected. A smart move with some options in play, might not be very good with another set of options, maybe even foolish.

Im therefore thinking that it may be infeasible to create an ai that performs equally well with all possible selections of options.

I would personally much rather see an ai that can do somewhat good with a preselected set of options. Than trying to make an ai that can cope with all options and therefor most likely never performing that well.
There are ofcourse some options that doesnt effect the game that much (and therefore the ai), which the player can use regardless. But perhaps some options could be given less attention in the development of the ai (which the player should be informed, when options for a game are chosen).

Are there being put any consideration into this?

[Shannon V. OKeets]

ORIGINAL: ptey
Recently there have been raised some points, about whatever the ai should do in various situations, is highly dependent upon the options selected. A smart move with some options in play, might not be very good with another set of options, maybe even foolish.

Im therefore thinking that it may be infeasible to create an ai that performs equally well with all possible selections of options.

I would personally much rather see an ai that can do somewhat good with a preselected set of options. Than trying to make an ai that can cope with all options and therefor most likely never performing that well.
There are ofcourse some options that doesnt effect the game that much (and therefore the ai), which the player can use regardless. But perhaps some options could be given less attention in the development of the ai (which the player should be informed, when options for a game are chosen).

Are there being put any consideration into this?

My starting position is that the AIO will play with whatever optional rules the player selects. Note that they can not be changed once a game has begun.

If problems arise, I'll review this assumption. However, I expect to be able to handle most of them by the way calculations are performed (e.g., the value of twin engine and night fighters, combat engineers, and other additional unit types). There are perhaps some 'strategic' optional rules, but I haven't looked for them in particular. Oil is one, but the AIO should be able to cope with that since it is such a popular rule.

[trees]

Check out the WiF discussion list this week ... Presence of the Enemy would be another dual-personality AI.

[ptey]

Ok, I understand that it is desireable to create an ai, that can play well with all options. Since this ofcourse will allow everyone to play solo games against the ai exactly as they prefer it.

I think there are a good amount of options that have a fairly big impact on both tactical and strategic choices. Some of these impacts are fairly obvious and some are more concealed, but still not to be underestimated. I have no illusions that i personally are able to grasp all of them, but i can think of atleast some options that imo will make a noteable difference on the way parts of the game is played. A small example could be in a game i started recently, where italy was able to launce a succeful invasion of malta as it entered the war. This was possible from the combination of playing with divs, limited overseas supply and SCS transport. CW didnt guard it well enough when these options were in play. Sloppy play by the CW perhaps, but in any case, i think there are alot of similar examples which becomes possible with various option selections.

One must recognise that only a finite amount of time is availeable for the creation of the ai. Thus you must to prioritize, do you want the ai to to some extent understand the impact of all optional rules, or do you want it to understand it well for a few.
I once heard a saying from a guy doing microchip engineering going something like "make the common case fast and the rare case work". An analogy for the mwif ai should imo be "make the ai play the common options selection well, and all the strange/uncommon option selections.. not quite as well".

These thoughts naturally implies, that you are able to create an ai where the nuances created by the selection of options actually matters.[:)]

[coregames]

Will I be able have an AI player as my ally?

[Shannon V. OKeets]

ORIGINAL: coregames

Will I be able have an AI player as my ally?

I am not promising this for MWIF product 1. However, I am designing the AI to support both AI Opponents and AI Assistants.

It is the player interface to the AIA that would involve the most work, since players will want to know what the AIA is doing and exert some control over it as well. Merely defining the places where the human and AIA interact in the decision making process would require quite a bit of work. Coding those interactions and then designing reports and control forms/screens for implementing them would be a lot of work too.