AI still coming?

[jzardos]

It's almost been a year since m y purchase. Haven't seen any progress on AI for this game. Is it still 'in the works' or has it officially been scrapped? Which would be too bad.

thanks

[michaelbaldur]

ORIGINAL: jzardos

It's almost been a year since m y purchase. Haven't seen any progress on AI for this game. Is it still 'in the works' or has it officially been scrapped? Which would be too bad.

thanks

solitaire and net play are priorities.

don´t think AI will be worked on in 2015, maybe 2016

[Erik Rutins]

Hi Jzardos,

We're still focused right now on issues reported since release and getting the NetPlay issues worked out. We've made a lot of progress across the board, but there are quite a few issues that still need addressing and it will take time. After we get through those and the community is happy with the state of the game, we'll look at adding AI. We very much want to do that, but we need to be realistic about the priority and timing.

Regards,

- Erik

[Centuur]

ORIGINAL: michaelbaldur

ORIGINAL: jzardos

It's almost been a year since m y purchase. Haven't seen any progress on AI for this game. Is it still 'in the works' or has it officially been scrapped? Which would be too bad.

thanks

solitaire and net play are priorities.

don´t think AI will be worked on in 2015, maybe 2016

I agree on that too. IMHO, before the AI can be constructed, the game has to be completed with the missing optional rules and the missing items of the basic game, such as neutrality pacts, peace settlements and search and seizure. Also, multiplayer Netplay is important. And there are still a lot of bugs around too...

[AlbertN]

I am just a user, and in accord to my own experience any AI in such super-massive strategy games involving so many variables, is just something not working.
Defining not working: Simply being unable to cope with the difficulty of the game, and able to react accordingly to new threats and positions and situations properly.

To make parallels - I can bring in Hearts of Iron. A quite excellent serie for a WW2 strategy game, that has been developed through 3 retails already. And frankly the AI has never been decent. Ever.
I can say the same for Matrix Gary Grigsby War in the East. The AI simply does not cut it.

Those games are suited, shaped and made to be played human(s) vs human(s).

Thus I am in the strong belief the AI, whenever will be made, will be disappointing if not even useless; and bore the player to death soon enough once they run amok through the AI-controlled forces.

[Klydon]

I think the HOI series and WitE are poor examples to say that a AI for a given game is going to be bad or whatever. There are other examples with games that are complex that have reasonable AI's.

The AI boils down to a couple of factors usually.

First is the skill of the programmer in designing the AI along with the programmer's skill and understanding of the game. Many times, AI's suffer from "bad strategy" simply because the programmer isn't very good at the game in question.

Secondly is how much time the programmer has to work on the AI along with what decisions are made in terms of acceptable game play (IE, the big one being how long do you make a player wait when letting the AI do its move? The longer, the better generally for the AI, but that is at cost of gameplay complaints from players who are unhappy with how long the AI takes to do the turns).

Something else that comes into play is the game itself. How easy is it to adapt to different difficulty settings for the AI? In most games, the AI would be given a production and/or resource advantage as an example. That is harder to apply to WiF since it is a computer adaptation of a board game. Now, play difficulty can be adjusted by what options are being used and perhaps that may be one way to address it. (Select hard and you are locked into playing with certain options and have some choices on other options for example).

WiF is a terrific game, but for many they just don't have the gaming group and/or space to toss it on a table and leave it for how long it would take to play the game. Being able to sit down and play against a computer opponent would be the next best solution for them. I fit in that category, so I await the arrival (hopefully someday) of when an AI is available.

[jzardos]

ORIGINAL: Erik Rutins

Hi Jzardos,

We're still focused right now on issues reported since release and getting the NetPlay issues worked out. We've made a lot of progress across the board, but there are quite a few issues that still need addressing and it will take time. After we get through those and the community is happy with the state of the game, we'll look at adding AI. We very much want to do that, but we need to be realistic about the priority and timing.

Regards,

- Erik

Thanks for the prompt response. I understand the priorities and challenges involved withj an AI for a game as grand as WiF. I'll mark down Dec 2016 on my calender to get back on WiF.

At least now have WitW to keep me busy for a while [:D]

[Omnius]

I'm really happy that in the old SSG Kharkov game I just bought that it allows for AI vs AI play. I always liked this now defunct setting as I read the rules while watching the AI play itself. That was always a fun way to get into a new game. Really sad that programmers have gone away from this wonderful old feature.

When designing scenarios having the ability to set the AI's playing was a useful scenario design tool for tuning in the respective AI's to make them play better. It would be an excellent tool for MWiF for trying to program the Artificial Ignorance to perform reasonably well for each major power. AI vs AI play allows for rapid completion of multiple test games to see how the AI's do. Far faster than trying to have humans test the AI's. Plus for new players it's a fun way to introduce them to a difficult game system once the second installment is ready to be released. If the AI's stink then you'll never hear the end of it.

Omnius

[Erik Rutins]

Agree - we included that in the new War in the West as well. It's quite useful for some automated (though not definitive) scenario testing.

Regards,

- Erik

[Neilster]

ORIGINAL: Cohen

I am just a user, and in accord to my own experience any AI in such super-massive strategy games involving so many variables, is just something not working.
Defining not working: Simply being unable to cope with the difficulty of the game, and able to react accordingly to new threats and positions and situations properly.

To make parallels - I can bring in Hearts of Iron. A quite excellent serie for a WW2 strategy game, that has been developed through 3 retails already. And frankly the AI has never been decent. Ever.
I can say the same for Matrix Gary Grigsby War in the East. The AI simply does not cut it.

Those games are suited, shaped and made to be played human(s) vs human(s).

Thus I am in the strong belief the AI, whenever will be made, will be disappointing if not even useless; and bore the player to death soon enough once they run amok through the AI-controlled forces.

This has been thrashed out many times in the past. The AI only has to be competent, so as to help ease new players into the game and give intermediate players a game whenever they want. Expert players will seek each other out.

A tremendous amount of work by very skilled and experienced WiF players has gone into the AI already. Have you read the AI threads?

Cheers, Neilster

[brian brian]

An AI for World in Flames has one great thing going for it - the structure of the first two scenarios should be ideal for building an AI. You don't have to start out trying to figure out if you will have enough of that one type of unit for the crucial summer campaign next year, or what grand strategy across the globe will be chosen. On land, the Panzers smash the Russian infantry and at sea the USN & IJN CV fleets maneuver around trying to find an advantageous battle situation. Those two building blocks will get the AI project off to a good start I think.

[Cataphract88]

I think that the order of preference should be: crush as many bugs as possible as the number one priority, then get the A.I. up and running, then and only then finish off Netplay.

[Centuur]

Sorry, but the fact is that Netplay is worked on right now. The writing of code of the AI will take at least a year, the code for Netplay is roughly in place but has a lot of bugs in it which slowly are getting cleared away...

[pzgndr]

ORIGINAL: Cohen
I am just a user, and in accord to my own experience any AI in such super-massive strategy games involving so many variables, is just something not working.
Defining not working: Simply being unable to cope with the difficulty of the game, and able to react accordingly to new threats and positions and situations properly.

To make parallels - I can bring in Hearts of Iron.

It all depends on how robust the AI scripting capability is with the game engine and how much time/effort is spent iterating through many many playtests (AI-vs-AI speeds this up) to continually improve the AI over time (months, years perhaps). If HOI is your only AI experience, you might want to try the Strategic Command series. SC-3 is in development for release by Matrix next year sometime. SC has very robust AI and event scripting to react to various situations.

Based on what I've read about Steve's approach to the MWiF AI scripting structured at a very high strategic level, an intermediate operational level for the various theaters and at an execution level for the various air/naval/land movements/combats, it should (hopefully, eventually) prove to be a competent and challenging computer opponent. For most of us casual gamers, I'm sure it will 'work' just fine.

[bo]

With all due respect to the posts here AI no AI, net play first AI second, I feel heart and soul the AI should have been done first, for several reasons, It gives the casual player something to play against without getting too committed like trying to work out a time element for playing other humans in netplay, he, the player wants to play the game when he or she wants to play not at someone else's whim.

I finally concede that it is better to play a human than the computers AI, as I have always played against an AI with every game I have purchased, out side of COD2 and WOT. Poster Cohen feels that there is no AI that is adequate for a computer war game and in general I have to agree with him. Someday there might be a breakthrough with AI playing but I feel it is not here yet and may never become a reality.

But I have always felt that an AI whether it be great, pretty good, good, just fair or poor, is better than me having to find a player in my time zone with the time on their hands to play me. Most AI's are destroyed by players who work out a solution against the AI after a few games, game in point Strategic command, the German AI was brutal against Russia [me] almost unstoppable until I put a chain of fortified positions across a part of Russia by engineers, the AI could not solve this but a human player would have solved it, so a good AI is possible, could it be in a game with 73,000 hexes, who knows.


To the board game player of WIF who met on a weekend and played for hours that was awhile ago as some of you have gotten married and moved on with other responsibilities than playing computer games, so I can see where netplay is your only salvation, which is a good thing.

I have been following WITW closely, reading all the posts in the last day or two, watched the very interesting film [over an hour]on you tube about operation Husky, the opinion on the AI will come shortly by experienced war game players, the game looks very good but it is not for me [might change my mind] it is not just the micromanaging which I stink at, it is the endless faceless divisions of both allies and axis units, the abstract air war which might be good or bad depending on each ones personal feelings, no real naval commitment that warspite so adequately explained in a post on WITW forums.

But the game was released playable, and at his moment in time that is by far the most important thing for the buyers of the game. There is no question the game will be tweaked in the coming weeks or months and patches will come out to correct problems that beta testers with no fault of their own did not notice in beta testing.

My conclusion is that IMHO if MWIF had delivered what should have been delivered, a viable working AI, a viable netplay program, a solitaire game with only a few bugs, this game would and should have been the biggest selling computer war game in past history and just maybe future history, IMO the game is brilliant, in my mind the finest war game ever made bar none, hopefully this will come to past in the next year or so do to Steve's, the beta testers and the other players efforts of bringing and fixing the game as it should be.

Bo

[Omnius]

I'm glad to see Matrix reviving a good old AI function, having the ability to have AI vs AI play. I hope Matrix makes this a standard feature of all future games with AI's. I sure hope this makes it into MWiF as that will help AI testing vastly.

Omnius

[Omnius]

bo,
After playing the AI some in Kharkov I found that AI stands for Annoying Ignorance. They had to allow their AI to cheat by things like allowing surrounded AI units to take replacements when a human in the same situation can't. My next game of Kharkov will be solitaire. No worry about when my opponent can play and no worry about cheating or stupid moves.

Omnius

[Omnius]

My two cents is to abandon making an AI for MWiF, or make it the absolute last thing to be done. I find that Annoying Ignorances really aren't worth playing. They might do well for a few turns but then they start falling apart. They'll make some good moves then completely abandon some area to make some foolish attacks elsewhere where it's not important. The AI for Time of Fury was the most pathetic of all time, so was that worthless excuse of a strategic WW2 game.

I just don't see an AI being capable of playing MWiF decently. Too many things going on and making decisions about which type of move to make won't go well unless totally scripted. I don't see the AI being able to make decent amphibious invasions unless it attacks an empty hex. Plus just imagine using the oil rules. I say focus on making the program's features work properly first then tackle the AI. I'd rather see the rest of the program's features finished and released before I see a pitiful AI get puked out that will only disappoint players while overloading the staff with too many AI complaints.

Omnius

[AbwehrX]

My two cents is to abandon making an AI for MWiF, or make it the absolute last thing to be done. I find that Annoying Ignorances really aren't worth playing. They might do well for a few turns but then they start falling apart. They'll make some good moves then completely abandon some area to make some foolish attacks elsewhere where it's not important. The AI for Time of Fury was the most pathetic of all time, so was that worthless excuse of a strategic WW2 game.

I just don't see an AI being capable of playing MWiF decently. Too many things going on and making decisions about which type of move to make won't go well unless totally scripted. I don't see the AI being able to make decent amphibious invasions unless it attacks an empty hex. Plus just imagine using the oil rules. I say focus on making the program's features work properly first then tackle the AI. I'd rather see the rest of the program's features finished and released before I see a pitiful AI get puked out that will only disappoint players while overloading the staff with too many AI complaints.

Omnius

Agreed! [&o]

[warspite1]

.