An AI Player's AE Concerns

[DD696]

PBEM only players, this is not for you.

As an AI only player, I am profoundly impressed with the latest round of changes that are being made to this game. The database expansion hopefully will be more than adequate years after it has been released. The order of battle is being greatly improved upon. The scale of the map will be improved and greater detail available. Many good things to drool over and shout about. And maybe one that isn't.

Reading through the new threads (land, air, naval, map, and general) it appears that virtually any question can be answered - except those that ask anything about how the AI will be improved. To those questions there is a universal theme of "It's being looked at". We can get specific answers to questions about midget subs, oil production, fuel refineries, aircraft performance, etc., but no one will say how the AI will be enhanced. One project member says they have been secretly working on this project for two years. Another says a year and a half. If it has been worked on for this long, then any project leader will tell you that they would have detailed plans drawn up for working on all aspects of the project and that all tasks are laid out, step by step (and yes, I have been a project leader as a Senior Systems Analyst in the data processing environment). It should be known what they are going to do, if anything, otherwise I wonder if anything truly good could come of it. If it is known, why no answers?

I think very specific questions need to be asked regarding AI behavior. For example, will a ship that has been torpedoed proceed to the nearest port of refuge (and not into an area under port attack) or will it continue upon it's merry way across thousands of miles of ocean until it sinks? Will the AI continue to send carrier task forces up against impossible land based air forces until they have been sunk? Look at what the Japanese AI does with its carrier task forces in the Milne Bay/Port Morsby area? It is possible to sink a good portion of Japanese naval assets there with a few squadrons of A-24's and B-25's (operating from at least 10,000 feet). Will the AI actually use it's assets in the latter part of the war? Could Yamato (if still afloat) sortie against forces invading Okinawa? Has the been changes made to the AI behavior so that the Japanese AI is not deathly afraid of six PT boats stationed at Guadacanal? I used to use them there and wondered why the AI would never contest me at Guadacanal. Then I stopped using them there, and elsewhere, and the AI did make some effort against those bases. Will the AI still only send one lone submarine to cruise around Colombo for one solitary cruise? Will the AI lose the vast majority of it's submarines off the coast of eastern Austrailia in it's effort to build up the Great Barrier Reef? Can a player have any hope of encountering enemy AI controlled combat task forces after 1942? Or is the war going to be limited to 1941 and 1942? Will the Japanese economy still be able to collapse for some indetermined reason when it does have adaquate oil, resources and shipping available? Will there be some variety to it's actions? Why could it not have options? Will the AI have some reaction to a player's actions? If I capture Marcus Island in 1942 I would think that Tojo might take some notice of that fact. Will the AI make a random stab at the Alaskan territory? Will the AI always stop at Guadacanal? Note that my questions always come as an Allied player. I am sure those who play as the Japanese against an allied AI will have many of their own.

I seem to recall that when WITP was first released that it was touted that the AI had the capability to take different paths. For example, it may attempt a Solomons strategy, or go for the Indian Ocean, or try for the eastern Pacific area, or try for Austrailia. I recall that there were to be four different opening gambits for it to try, but I could be wrong. I have only seen one. The game was hyped as having a capable AI. After the games release any time anyone suggested improving the AI they were hooted at, snickered over and poo-pooed at, which will happen to me (just note what happened with Herwin's suggestions in the AAR section).

So, the question I raised in the General AE discussion thread still remains unanswered - who is the advocate for the AI players? Who is attempting to address the concerns of the AI player? Where is the AI thread? Who is programming the AI? What are the plans to do so? Who do we ask questions of regarding AI behavior and the improvement thereof? I feel that anyone who is interested in improving the behavior of the AI needs to speak up now and let their specific concerns be known, or else these concerns are going to be lost amongst all the gasping and googling over the pretty new art, aircraft characteristics, ship improvements, and land units (of which the effort make on is truly commendable). Can the same be said of the AI? Speak up now, or wonder why later as you kick yourself in the posterior. I know that I am not the only one who is concerned by this. It is time, now, for your questions and concerns.

[Erik Rutins]

Well, we've responded to a few of your questions already, but I understand you want specifics. We can't give specifics yet - when you embark on this many changes, it's almost like building a new game. The AI is generally the last piece to be put back together after all the other changes (which it needs to adapt to) have been completed. WITP already has a working AI, but we are still in the process of updating it and finding out how it deals with all the new stuff. Our absolute minimum goal is that the AI be no worse than it currently is. I hope it will be better, but testing and development between now and Summer 2008 will determine that. I also expect that most of the many great improvements in the game in non-AI areas will make the game play out much better against the AI as well.

As far as advocates for the AI player, well we all pretty much are (or I'd suspect about 70% anyway). I can say however that the project leadership is very much aware of the remaining AI work to be done and that we all want to see an AI that is as good or hopefully better than that in WITP, along with all the additional improvements.

So, the simple reason you haven't had any concrete answers on AI specifics yet is that the AI is the last piece to finalized and re-tested and so that area is very much "under construction". Feel free to post suggestions and concerns, but realize that the AI is not a low priority, we're just at a point in the project where the other areas are much more solid and can be more easily discussed.

Regards,

- Erik

[Buck Beach]

ORIGINAL: DD696

PBEM only players, this is not for you.

As an AI only player, I am profoundly impressed with the latest round of changes that are being made to this game. The database expansion hopefully will be more than adequate years after it has been released. The order of battle is being greatly improved upon. The scale of the map will be improved and greater detail available. Many good things to drool over and shout about. And maybe one that isn't.

Reading through the new threads (land, air, naval, map, and general) it appears that virtually any question can be answered - except those that ask anything about how the AI will be improved. To those questions there is a universal theme of "It's being looked at". We can get specific answers to questions about midget subs, oil production, fuel refineries, aircraft performance, etc., but no one will say how the AI will be enhanced. One project member says they have been secretly working on this project for two years. Another says a year and a half. If it has been worked on for this long, then any project leader will tell you that they would have detailed plans drawn up for working on all aspects of the project and that all tasks are laid out, step by step (and yes, I have been a project leader as a Senior Systems Analyst in the data processing environment). It should be known what they are going to do, if anything, otherwise I wonder if anything truly good could come of it. If it is known, why no answers?

I think very specific questions need to be asked regarding AI behavior. For example, will a ship that has been torpedoed proceed to the nearest port of refuge (and not into an area under port attack) or will it continue upon it's merry way across thousands of miles of ocean until it sinks? Will the AI continue to send carrier task forces up against impossible land based air forces until they have been sunk? Look at what the Japanese AI does with its carrier task forces in the Milne Bay/Port Morsby area? It is possible to sink a good portion of Japanese naval assets there with a few squadrons of A-24's and B-25's (operating from at least 10,000 feet). Will the AI actually use it's assets in the latter part of the war? Could Yamato (if still afloat) sortie against forces invading Okinawa? Has the been changes made to the AI behavior so that the Japanese AI is not deathly afraid of six PT boats stationed at Guadacanal? I used to use them there and wondered why the AI would never contest me at Guadacanal. Then I stopped using them there, and elsewhere, and the AI did make some effort against those bases. Will the AI still only send one lone submarine to cruise around Colombo for one solitary cruise? Will the AI lose the vast majority of it's submarines off the coast of eastern Austrailia in it's effort to build up the Great Barrier Reef? Can a player have any hope of encountering enemy AI controlled combat task forces after 1942? Or is the war going to be limited to 1941 and 1942? Will the Japanese economy still be able to collapse for some indetermined reason when it does have adaquate oil, resources and shipping available? Will there be some variety to it's actions? Why could it not have options? Will the AI have some reaction to a player's actions? If I capture Marcus Island in 1942 I would think that Tojo might take some notice of that fact. Will the AI make a random stab at the Alaskan territory? Will the AI always stop at Guadacanal? Note that my questions always come as an Allied player. I am sure those who play as the Japanese against an allied AI will have many of their own.

I seem to recall that when WITP was first released that it was touted that the AI had the capability to take different paths. For example, it may attempt a Solomons strategy, or go for the Indian Ocean, or try for the eastern Pacific area, or try for Austrailia. I recall that there were to be four different opening gambits for it to try, but I could be wrong. I have only seen one. The game was hyped as having a capable AI. After the games release any time anyone suggested improving the AI they were hooted at, snickered over and poo-pooed at, which will happen to me (just note what happened with Herwin's suggestions in the AAR section).

So, the question I raised in the General AE discussion thread still remains unanswered - who is the advocate for the AI players? Who is attempting to address the concerns of the AI player? Where is the AI thread? Who is programming the AI? What are the plans to do so? Who do we ask questions of regarding AI behavior and the improvement thereof? I feel that anyone who is interested in improving the behavior of the AI needs to speak up now and let their specific concerns be known, or else these concerns are going to be lost amongst all the gasping and googling over the pretty new art, aircraft characteristics, ship improvements, and land units (of which the effort make on is truly commendable). Can the same be said of the AI? Speak up now, or wonder why later as you kick yourself in the posterior. I know that I am not the only one who is concerned by this. It is time, now, for your questions and concerns.

I know the questions are not directed at me and I know nothing from nothing. Having said that, let's think about it. None of us have heard (I don't think) of any great strides in any games AI improvement. We are just not there. The complexities of WITP were more than the AI (for the lack of a better term) could handle. AE is double or triple the game and complexity of WITP.

DD696, the silence is deafening and I can feel that Matrix (et al) don't want to say the obvious. And, I think you may be looking for answers you really don't want to hear. If you want me to say it "AE is for PBEM"

Hope and pray that I am wrong and that miracles happen, but I think not.
I am an AI only player.

I see you do have a response from Matrix now and I guess my post is unnecessary. However, my nature is pessimistic and I don't see much of a future of me being able to enjoy much of the great things in this new addition. I could always play HEAD TO HEAD, but I can hardly keep up playing one side.

[AW1Steve]

[:)] How ever it works out Erik , thanks. We know you guys are trying your best. We always wish you luck , and you should know that we are your biggest fans.    My only request , and please don't take this as anything critical, is if you can take the time away from this monumentous project for an occassional update to your progress , we would all be very ,very greatfull. Thanks again...[&o][&o][&o]

[Captain Cruft]

I think it might be better to just admit that it's impossible to make an AI for something as complex as this that isn't utterly useless. Then we could have the thing earlier.

That won't happen of course. Although this is a real labour of love, it's still a commercial venture. Though for how much longer this industry can keep selling something that doesn't exist I'm not sure. As long as people keep buying I guess ...

[wworld7]

ORIGINAL: Captain Cruft

I think it might be better to just admit that it's impossible to make an AI for something as complex as this that isn't utterly useless. Then we could have the thing earlier.

That won't happen of course. Although this is a real labour of love, it's still a commercial venture. Though for how much longer this industry can keep selling something that doesn't exist I'm not sure. As long as people keep buying I guess ...

They will do the best they can. Earlier is often not better. Rushing creates an increase in things missed. So no rushing...please.

[Captain Cruft]

I meant ditch the AI. Or just keep the existing one, which will still be in the code though doubtless it would perform even worse.

That would make it arrive earlier.

Don't worry I have no expectation that this will actually happen.

[Captain Cruft]

BTW in case anyone think's I am an ardent PBEMer that's not really true. I have spent far more time playing the AI than PBEM, and though PBEM is inherently better the gamey aspect stopped me playing that too.

I just can't see the point of pretending that you can make a good AI that's all. Though to be fair Erik's last post doesn't really do that. So you might say that in this specific case, where it's not a new game and everyone knows how to play it, that working on the AI is indeed pointless. Just release the thing for PBEM only.

[Mike Scholl]

Something for all you AI players to consider. As long as the game is built on 2by3's original code it's doubtful there CAN be any big improvement in AI performance. Your real hope will be WITP II with new and more flexible code..., and that's still years off.

But with the advent of the "Admiral's Edition", EVERYONE is going to be in "the same boat" skill and experiance wise. Good time to try out PBEM..., everyone will be on a level playing field. And with so many "AI only" players coming out to comment now, you can also find some opponants who have the same type of time constraints you do and are willing to play a slower and somewhat spasmatically timed game.

I'd love a really challanging AI myself for playing whenever I feel like it (I'm retired and keep odd hours). But I don't see how with 2by3's kludgy hard-coding it's going to be possible in this incarnation. My hope is that some of the shorter scenarios which can handle heavy "scripting" will provide a better "solitaire" challange. But in the big campaigns it would seem impossible for the AI to meet the needs of AI-only gamers.

[wdolson]

I understand the explanations about the AI.  One of the programmers weighed in the other day in the AI thread before the announcement.  From what he was saying, it sounded like he had been looking at the AI lately.

One of the biggest limitations with modifying the AI is how much time it takes to do its tasks.  It is technically possible to make the AI much smarter, but players aren't going to be happy if it takes an hour or more to process a turn.  The new computer requirements are one indication of how much more computing overhead there is now, before changes to the AI.  It sounds like the code for all the AE changes is being done right now.  Nobody knows yet how much more time the new routines are going to eat up and how much overhead will be left for the AI to do its thing.

Once the timing issues are known, the team can look at how much smarter the AI can be made.  I've worked on a lot of real time systems.  There are situations when you aren't completely sure how much time various routines are going to take until you write them and test them out.  You can usually get an estimate of how long things will take, but won't know your timing budget until some parts are working. 

I think the reason the development team is being vague is that they don't want to promise something they can't deliver, so they are not saying anything concrete until they have more data.   It may be frustrating to hear, but I understand it.

Bill

[mjk428]

ORIGINAL: Erik Rutins

Our absolute minimum goal is that the AI be no worse than it currently is.

If this absolute minimum goal is not met, will you please do this AI player a favor and release it as a PBEM expansion only. I know I'm not going to be able to resist this if there's even the slightest chance that the game is at all playable against a computer opponent.

I only raise this because with the additional features, it seems plausible that the AI's is going to start off even worse. IMO, it's going to take some expertise and effort to bring it back up to the disheartening, albeit understandable, minimum goal. If not PBEM only, maybe with a disclaimer that the AI aspect is beta & unsupported.

Finally, thank you for being so upfront about this aspect of the game. I might buy it even if only to play head to head solitaire - just for the opportunity to move around the pieces.

[Erik Rutins]

It seems to me that there's a lot of pessimism around without much basis. Can I please ask for some patience? We're not about to release AE right now, so there is still plenty of time for us to finish AI and balance work and give you some progress reports. The fact that we're not ready to discuss the AI now does not mean that there will be no AI - there will be an AI, we're just still working on it.

I'm going to move this into the sub-forum so that it will be next to the rest of the AE FAQ threads, since AI seems to be a big question area.

[aztez]

I already said DD696 that the comments about this issue were harsh and utterly pessimistic.
 
Oh, ..and I'am not PBEM only player by no means. If they can improve the AI than good if not than so be it. I really doubt it is an easy task to do.
 
Keep your fingers crossed and hope for the best. As the development team stated they need to do other stuff first before they can tangle the AI. So, I doubt you are going to see much "specific" answers yet.
 
Witp is a huge game... and with those add onns revealed is going to be gigantic wargame.

[Anonymous]

[Deleted by Admins]

[Anonymous]

[Deleted by Admins]

[Feinder]

I'm sorry, but how can you (3rd person, no one particular) possably complain a release that, hasn't even been released yet?  It's just a discussion of new features and couple of screen shots.

I'll hand it to Erik and the rest of the design staff.  These guys have busted their @sses to offer a major update and they're proud of it.  And yet there's already folks lining up with criticisms.

It's kind of like the a group that criticizes the movie without ever having seen it.

Criticism is fine, it makes for a better product.  But I'm thinking, maybe wait until there's actually a product before trashing it?

-F-

[Anonymous]

[Deleted by Admins]

[Erik Rutins]

Before this spirals out of control - all I'm asking for is some patience. The AI is a priority for us as well and we will keep you all posted. I don't feel anyone is trashing the game, I think these are just concerns, but I don't see that there's any basis for assuming that we will either release without an AI or with one that's much worse than WITP. Just give us the benefit of the doubt and we'll let you know how it goes.

[mjk428]

ORIGINAL: Erik Rutins

It seems to me that there's a lot of pessimism around without much basis. Can I please ask for some patience? We're not about to release AE right now, so there is still plenty of time for us to finish AI and balance work and give you some progress reports. The fact that we're not ready to discuss the AI now does not mean that there will be no AI - there will be an AI, we're just still working on it.

I'm going to move this into the sub-forum so that it will be next to the rest of the AE FAQ threads, since AI seems to be a big question area.

Just trying to manage my expectations. I dearly would love to be completely wrong in my expectations. Which are still cautiously optmistic despite my best efforts to tamp them down.

Again let me add that I appreciate your willingnes to discuss the subject at all. Thank you for that. I really, really mean it.

[juliet7bravo]

Probably the biggest issue confronting the AI are the (to me) game breaking bugs and kludges inherent in WitP.  Human players can use "work arounds", "house rules" ect. to get around many of them...the poor AI can't, is basically porked, and has never had a chance.
 
However, if you look at the PR release, they're addressing most (all?) of the major (and many, many minor) issues we've collectively found, all the way back to UV release.  If these issues are "fixed", then even without an AI "upgrade", it'll be a great stride forward.  It isn't just the additional chrome and the game features we've begged for (for years), or even Brady's midget subs (major though they may be)...it's getting the game "working as designed".
 
Remember that one yammerhead front man for Matrix, whatszizname, with his "working as designed" BS?  Well, if they get the game "working as designed", the AI may have a fighting chance.  That'll effectively be an AI "upgrade".
 
Though I'm not holding my breath since no one evidently knows half of what goes on under the hood of WitP.  Hopefully there'll be a little less "working as designed" toeing the party line amongst the playtesters this time around, and a little more "this is FUBAR, let's re-look it".  Though I am very optimistic since all these guys are serious WitP players.