Status of AI

[Ian R]

ORIGINAL: AxelNL

Michael, you are too pessimistic. Netplay was unplayable earlier, and it is now. The bugs Steve are fixing are less and less critical ones. He added a optional rule (kamikazes) recently.

He will start the AI, and the result for us all will be a better convoy/production auto-optimization - which I am greatly looking forward too.

That is excellent news, that is a bit sub-optimal at the moment.

[Ian R]

ORIGINAL: warspite1

It's pretty simple (to me at any rate). Unless production and convoys are fixed then an AI is impossible. An AI will always be at a disadvantage compared to a human in a game like this - and playing with less than optimal resources will hamper the AI even more. The only way around it (that I can see) would be to give the AI a certain no. of build points each turn - and these can be increased or decreased according to difficulty level you want to play at. This would make the convoy game somewhat redundant though....

Give it a % increment instead of a set number of points. The CW seemed to miss out on 2-3 PPs a turn (not BPs) because of convoy routing, out of about 20. So a 15% increment or some such.

[Courtenay]

If the programming language were C or C++, I would be more than happy to help. Unfortunately, all I know about python is that it is associated with the Oracle of Delphi and that it is part of the name of the best British comedy show.

[pzgndr]

ORIGINAL: Shannon V. OKeets
The current # of lines of code for MWIF exceeds 1.4 million.
Good luck on getting someone interested in making knowledgeable changes to a program that large.

Empires in Arms. I rest my case. They're coming to take me away, ha ha, they're coming to take me away... [:D]

[rkr1958]

ORIGINAL: davidachamberlain

ORIGINAL: rkr1958
What I'm kicking around is a community based, python-based open source approach that operates on the game file to produce the desired modifications. There is no action, or desire, to modify the MWiF program. It will be used as developed and packaged by MATRIX as the core of this envisoned community based, open source approach. The most stable, solitary mode of play would be the core. But even with this as the core you could have true multiplayer (3 or more) simulatenous and off-line play. But I think this vision is only possible with true, extensive and distributed community participation.
I

Though I think it is an interesting concept, that ship has sailed a really long time ago. When you think about the number of functions and features involved in this application, unless you can engage a very large number of committed and capable people to participate and collaborate for quite a long time, you will not be able to re-platform such a product.

This is too large of an undertaking to maintain the kind of commitment required to complete it.

Unless there is some really major sponsorship (major investment) to start all over again with a new team, this is just not feasible.

Regardless of Steve's commitment to getting this all done, I believe that what is there and what will get done on the current path is as far as this is going to go.

Dave

Again my vision isn't one of re-platforming but a collection of game file editing apps. But in either case I think you're right at this point. It's just a pipe dream. [:(]

[brian brian]

I think an AI for the Barbarossa scenario will come along and it will be excellent. Players will use it to dial in their own play and then that will increase their skill and interest in play of a full game vs humans.

Then an AI for the Guadalcanal scenario will come along. That one will show only flashes of excellence at first, as no human or AI can really be held accountable for running their fleet into a 1/10 split. But over time the naval AI will also become good. Players will use it to....

[davidc]

The current # of lines of code for MWIF exceeds 1.4 million.

If there are any ladies reading the MWIF forum... "In programming size does not matter."

What matters is clarity and effectiveness.

[markb50k]

Perhaps what rkr1958 mentions is actually a reality. I have been heavily modifying the save game files for years now, at first via a text editor called WinMerge, then via an Excel spreadsheet, then finally embarking on a Windows WPF app written in c#.

I will probably post a separate update thread to discuss the current scenario I'm playing (World War III), but will post a picture of what is possible:

This app (in concert with modified game CSV files) allows me to pretty much run my own WIF-inspired game, where I utilize the actual WIF.exe for things like:
- turn mechanics
- weather rolls
- and the pretty map

But pretty much every thing else is ran through the Windows app, and it simply tells the WIF.exe game what to show via the edited game save file.

It is the most fun I've had modding a game in a long time and I am in complete control of many factors impossible via the original game

[AxelNL]

Impressive

[Admiral Delabroglio]

Hello

CrusssDaddy, I am not trying to start an argument here, but I think the way you stated your opinion regarding Steve's job is not very nice, and I am not the one directly concerned. If it is just due to my poor undestanding of English, I apologize for my statement.

Even without an AI, playing MWiF enables to leave the whole living room free of maps and counters, so with the state it is in, I believe it is already worth it. I wish Matrix had claerly stated that an IA was NOT to be expected during the next decade at least, but I would have bought the game all the same.

As for the lost PP due to convoy management lack of efficiency (or my lack of know how), I sometimes edit the number of saved build points the way Ronnie explained, and that's good enough for me. Maybe as a (temporary) work around, Steve could add a magical "extra BP" button to be clicked during the actual production phase...

Best regards

[BoerWar]

A pox on all apologists. When did this piece of software trash come out? A decade ago. Matrix stole our money. It should have AI by now, it was sold as a product that would have AI. Haven't bought another Matrix/Slitherine game since. Bought several prior.

[paulderynck]

That's a shame.

[Omnius]

Why waste time on an Artificial Ignorance that would be overwhelmed by such a complex game like WiF? Totally inappropriate to call the game software trash just because you don't have a poor AI to trash to make you feel smart. I'm disappointed that we won't get oil tankers for the annoying convoy system but I'm not trashing the game or programmer because of this sad decision. Grow up boerwar and just get over it, play a paradox game since all of them have absolutely pathetically ignorant AI's, that seems to be your speed.

[sIg3b]

Disagree with both sides.

Instead of impossible super projects, they should have started small. Much better than promising an impossible 1939 Global War AI that will never be finished, they should have built a heavily-scripted strong computer opponent for the smallest scenarios (Barbarossa and Guadalcanal). THAT should be doable.

[davidachamberlain]

I expect that Steve will take a progressive approach. I would not be surprised to see the AI evolve and start with Barbarossa to focus on the land action. That would likely evolve into Guadalcanal.

I agree that it might not make much sense to go for the Global War since there will be so much involved in testing it all. However, that will be up to Steve. It is a personal challenge that he wants to take on and is essentially his reward to getting this far. So, let him go for it.

Besides the technology and hardware will evolve a bit during that time, but getting human like performance, both in wait time and decision making, will be hard to achieve.

I would just like to see the full Pacific and European scenarios, most of the essential optional rules, and maybe a 4 player (or more) version of Netplay. I prefer to use my own intelligence in place of the AI as the game is well enough suited to allow for that. Fog of war for Initiative, Weather, turn ending, and my own forgetfulness of some planned moves is enough to keep it challenging.