This discussion provides a nuanced look at the delicate trade-off between technical competence and player immersion in game design. It effectively demystifies why "perfect" AI is often less desirable than a well-crafted, predictable challenge.
Install our extension to search inside any video instantly.
AI in Gaming | Single Malt Strategy | Episode 98Added:
Hello everybody. It's the historical gamer once again, but you can call me Matt. And today we're here for another episode of the Single Malt Strategy Podcast. This is episode number 98, and I'm joined by my trusty co-host, Tortuga Power. How are you doing today, Tortuga?
>> You could say that I am expecting a very good podcast. You get that? You see what I did there?
>> Yeah, I I do. I guess I guess we are acknowledging that there's another mini Tortuga on the way in this episode, then.
>> That's right. That's where I was going with that.
>> Didn't know if we were sharing that, but now that we have, >> we just explained the joke. It gets funny.
>> Yeah, I know. Additionally, I am also joined by another co-host here, uh, Wolfpack. How are you doing today, Wolfpack?
>> I am doing great. It's fantastic to be back on the pod. I feel like it's been a while since our last one. We had a little bit of fallout after our last podcast. It really it strained the group. Well, you know, uh, space is space is hard, and there's not many good space games, and I'll leave it at that.
Yeah. Well, I'm glad we managed to figure it out, make amends, and here we are. Kind of beautiful. You sit down and really think about it.
>> Just the three of us today, right?
>> Yeah. Some people were so upset that they're just they they boycotted this podcast.
>> Yeah. Like many viewers said they would.
Uh, there was actually a ride in campaign, ZGu. I don't know if you've heard about that, but viewers were writing into the podcast.
>> A ride in campaign? Really? Where did they write to?
>> Yeah, the letters, man.
>> They were writing to me, my PO box. I It's flooded with uh handwritten letters talking about how KSP1 is actually a fantastic game and that we need to kick you off the podcast.
>> Well, you know, I hear it's very popular again thanks to Artemis and uh, you know, Artemis 2's mission. What's that?
>> Uh, you know, it's the it's the around the moon shot that just uh just just landed actually a couple of hours ago.
We're recording this on uh April 10th, so we'll see when this goes live. It could be it could be June by then. Who knows?
>> I think what you're supposed to do when podcasting is not immediately date your episode in the first, you know, 5 minutes. But that's cool that we did.
>> Don't mention it's April 10th.
>> Why don't we do Is that an AI thing? Is it a Is it a algorithm thing? It's like uh you want your viewers to come across a podcast and be, you know, it's evergreen, as they say.
>> Yeah, I kind of I could see that.
>> I mean, our topic is evergreen.
>> That's true. The topic is is evergreen, I suppose.
>> So, if they didn't uh if they didn't, you know, log off already because they heard me say April 10th and they're like, "Oh, no. Don't want to listen to that."
>> Yeah. It's like It's like, "Oh, this is actually they're going to be listening five years from now and they're like, "Hold up. This is five years old. I ain't listening to this old trash. and they'll go find our new podcast.
>> The podcast dates itself in the feed though. Like it's not >> I don't know. I'm no expert obviously.
>> Okay. Yeah. I mean neither am I, right?
Like we're we're still chilling with uh 33 reviews on Apple. So if you are listening to this podcast and you do like this podcast, you know, drop into your favorite podcast app where you're listening and leave a review because that helps the channel out. And now I've done my uh obligatory podcaster thing.
>> Oh, thank you for doing that first of all. But I also like the way you um phrase that. I heard this from another podcast. They did the same thing.
They're like, "If you like the podcast, go ahead and view." But you know, the the subtext there is if you don't like the podcast, first of all, what the hell are you doing? Listen to it. But second of all, don't rate us. We only want good ratings.
>> I've heard that it doesn't matter though. I've heard that it's just based on like if you have a lot of activity, >> Yeah.
>> it's actually good for the algorithm regardless of whether it's positive or negative.
>> So if you hate this podcast, go leave a review. Keep listening. Go leave a review. Tell us tell us which co-host you dislike the most.
Tear us apart at the seams.
>> No. If you hate the podcast, make sure to write into Wolfpack 345's PO box.
>> Yeah. Yeah, that's a good idea, actually. Why not? I need the the thing uh just sitting there.
>> Have you guys gotten, by the way, ever gotten like deliveries from fans?
>> Yeah, I I actually I do have a P.O. box.
It's in uh I think it's in the description of my videos. It's somewhere. I don't really promote it all that much, but I've had uh a few fans send me some pretty cool knickknacks and stuff, badges, things like that.
>> Yeah, it's pretty nice.
>> Yeah, I've gotten stuff for some people, too. Um some books, which was awesome.
Anyway, >> yeah, that that's pretty cool. I guess I'm less popular with the shareholders.
I've not had uh fan mail before. I've had developers send me stuff. the developers for the, you know, Tannenburg, Gallipoli, Verdon, that like FPS or whatever. Um, they sent me some like World War I pins and stuff.
>> Well, that's freaking awesome. I like their games.
>> They're fun. They've got a Gipoly one.
Is that out yet? I don't even know.
They've got a new FPS coming out.
>> Gipoly. I'm looking forward to that. I think that's a pretty interesting setting compared to uh I mean the all their settings are interesting. I I think the weakest game was Tannenburgg just cuz the variety of the maps. It's like okay, Eastern European forest, etc. But anyh who, >> yeah. Well, I mean, they had just done Verdun, so which like, yes, less trench focusy, but still kind of pretty similar in terms of the gameplay. I feel like >> to go off on a tangent real fast, I'm going to make it quick. That was one of my problems with Tannenburgg is that it it felt more like a regular FPS in terms of gameplay where it's just like Object Capture A where Verdun did a really good job of the back and forth Trench Warfare with their uh main game mode. But anyh who, >> yeah, that makes sense. I think they sent me the stuff maybe part of the Gallipoli announcement promo kit, but I did a I did a sponsorship for the Italian game.
>> Yeah, I like that one too.
>> And that was fun cuz the terrain there was so different. I I enjoyed that.
>> They've added some pretty awesome new game. We should We should all play it.
We should stream it. I haven't stream I've streamed like maybe twice this year. What What's our top What are we talking about today?
>> Well, we're going to be talking about gaming AI. But before we get to that, what have you guys been playing lately?
What about you, Wolf?
>> I have been playing, you know, in my free time. I guess members of the pod or listeners of the pod know I like a good uh RPG game. So, I've been kind of playing through Boulders Gate. Uh, one I've played Icewindale earlier, which is another Infinity Engine game, but kind of playing through that in my free time.
Of course, you all know the usuals that I'm playing uh on my channel, things like Seat Power and stuff like that. But >> for Icewind, Dale, is that the like the 1999 title?
>> Yeah, something like that. It came out after Boulders Gate.
>> Okay.
>> Use the same engine. It was made by blanking on the developer studio, but anyway, it it they were the publishers for Boware at the time and then they they kind of used the the that Infinity engine to make Icewindell. It's fun.
It's definitely more dungeon crawly than story driven like Boulders's Gate, but like a lot of the backgrounds are really well done. Like the dungeons are really well done. It's uh it's quite good. I have a lot of memories of it as a kid.
It's probably the one I played the most.
Although Boulders's Gate 2 is probably my favorite of that group of games.
>> Yeah, hard one to disagree with. It's a good one.
>> I'm playing through Boulders's Gate one now and it just makes me I'm like, man, it's pretty dull compared to two and the dungeons are kind of lackluster compared to Icewindale, but it's cool to see how the series kind of evolved.
>> Yeah.
>> What about uh what about you, Wolf? Or sorry, Tortuga? What have you been playing?
>> If I want to share what I've been doing, what's my free time doing? working on a game. But uh I've been playing a lot of Factorio lately. I've been like looking for something that is a little bit more I mean it's pretty relaxed. You can like basically AFK in Factorio and your factory runs itself. Everyone knows Factorio. I guess I don't really need to describe what it is, but it made me realize like how I don't know. I've been reflecting on how what are the games that I have installed on my desktop?
Maybe that should be another podcast episode sometime. the games that I keep with me >> and Factorio is just is one of those that I will just never uninstall it.
That Kerbal Space Program. Okay, sensitive topic from based on the last episode, but that and Rim World. I mean, these are just games that I will never uninstall because they're so good. And at any point in time, you can go back and play Factorio and just have such a good time. It's timeless. It really is timeless. Like, the graphics don't get in the way at all. That's interesting thinking about the games that I've really never uninstalled on my computer.
That that would be a good uh podcast topic.
>> Total War Show Shogun 2 is on there, too. Just a game I will never uninstall.
>> Yeah, that is fantastic.
>> I don't think I have any games like that.
>> Did you say you didn't you don't have any games like that?
>> Any games that I don't ever uninstall?
>> Yeah.
>> That I still play regularly? I don't know. I don't think so. I'd have to think about it.
>> War in the Pacific. Yeah, I heard that.
>> Okay, fair. more in the Pacific probably. I'm not actually actively playing it right now, so >> that's fine.
>> But you you keep it on your computer.
You whenever it's always there. You never have to go through the rigomearroll of installing a game. Not that that's even a problem these days, right?
>> That feels like a topic I would love to talk about though. That's a good point.
That would be a good future episode.
>> It has like so little to do with the inconvenience of un of reinstalling it and more to do with like comfort food.
It's right there on the desktop. It's almost fun to just look at it. Yeah, >> memories, nostalgia. I don't know. Yeah, we'll talk about it another time. So, Factorio mostly.
>> What about Warhammer Total War? Is that one on there for you?
>> Um, it's not, but that's because I I was going through that phase when I was trying to like beat quote unquote beat games and then uninstall them.
>> And I played through an entire campaign all the way with the dwarves to the very very end where I conquered whatever the enormous requirements are to actually beat the game. And like in hindsight, I have to say Total War games, they really once you just start steamrolling everything, even with the endgame chaos person who comes and like your endgame boss, it's just not as fun. It's so much more fun when you're expanding.
>> Yeah, I I can agree. Most of the time, uh there's games where they make it a focus where you're kind of just trying to survive like a tiller or whatever.
But no, I only asked that because I actually that's that's another game I've recently been playing. Warhammer Total War. I love it. I mean, is it you're doing one and two or you doing three?
>> I'm doing three, but I I bought all of them. So, I have like that big map. I mostly started playing it because I started playing uh Warhammer Fantasy, like the tabletop game, uh building my army and stuff with that. And then I was like, okay, let me let me finally get around to playing Warhammer Total War.
another podcast topic I think because I I would love to kind of drill into that right now even I'd love to ask about how's the tabletop experience like now that you're doing tabletop has it changed your thoughts on Warhammer Total War lots lots of little things I'd like to pick up there >> yeah cuz I was pretty against Warhammer Total War to begin with as a Total War fan I've come around to it >> oh I I love it it's so good it's it's like in its own way it is a really really good Total War game like the balance of the different I mean because obviously when you have a fantasy um universe, you can just add a lot more asymmetry between the different nations >> cuz they can be like totally different.
>> Totally.
>> Obviously, human conflict, I mean, every human nation kind of knows the metag game and they they like move towards it just like any I mean, we're not stupid.
So, when someone's like, "Hey, I got cannons." Not somebody's not going to still focus on like, you know, cutlesses or something, you know?
>> Right. Right. So, you know, you just get this constantly this constant symmetry between armies, which makes something like Warhammer where you can completely defy that pretty fun.
>> Absolutely. Well, what about you, Tishi?
>> Trying to think what I've been playing lately. I've been playing a bit of Ultimate Admiral Dreadnots. I know that's your favorite game, Tortuga. I I originally started when they released the final patch when they like shut Games Lab down back in late 2024. I was like, I'm going to do a new series. and I played one video of it and one or you know an hour of it or so and just never never followed through on going back to it. So I've been playing that a bit. I'm kind of enjoying it. I' It's interesting. I've gotten a number of comments from people like outside of the streams reaching out to me and be like the more I see of this the more it is it's a flawed game. Don't don't get me wrong, but the more I see of it, the more I think it's better than Rule the Waves, which I think is interesting for people to think. I kind of feel like maybe rule the waves just kind of like I don't know if wore out its welcome is the right term but it feels like rule of the waves one like hit such a high note and then the f the subsequent games just haven't iterated enough and I think people have kind of maybe gotten a little tired of it. I don't know. You'd know you're closer to that community than me. Tortuga, >> I think that's a pretty interesting take. Uh I want to also defend myself a little bit. I don't actually have that negative of an opinion on Ultimate Admiral Dreadnots. It's just not a game that I decided I wanted to play a lot.
Actually, I had a copy that worked only from the desktop, not from Steam. I I like, you know, like manual install.
>> Oh, yeah. What they used to do. Yeah.
Yeah.
>> Their own launcher. And then it actually doesn't work anymore. So, I went I tried to play it the other day, not the other day, but like a year ago. And I was like, "Oh, I just can't." Okay. Well, I guess that settles that for me.
>> I'm assuming those servers went down when they closed Game Labs. So, what they used to do, Wolf, is they used to give codes early. So, they did early access directly through their website with their own launcher. And then they would give you a Steam code when it came to Steam if you owned one, but I'm guessing Tortugi probably never did that because you weren't super into the game by that point.
>> When the game first came out, I think I was like one of the most I don't know the full scope of YouTube, but I think I was one of the most popular content creators for that game. So, it did surprise me a little bit that I mean, maybe it's maybe they did send an email and it went to my spam or something, but yeah. I mean, to my knowledge, I don't have a key for that. And I don't know.
Are they still around even? It seems a little late to be asking to >> Well, no. I mean, that's the thing is they went under, which is why their launcher doesn't work anymore, is their I'm sure their servers got shut down.
>> Yeah. I I don't know if they were proactive about it. I can't remember. I remember I had a code with the pre Steam version and then at some point I got a code with the Steam version, but I may have had to ask like their team be like, "Hey, by the way, I'd love a co." I'm not really sure what they did, but it's been it's been years, you know, since since that happened. But yeah, so I've been playing Ultimate Admiral Dreadnots and then I've also been playing a little bit of uh Full Metal Sergeant 2 uh which is kind of a lighter game. You just want to play something that doesn't require too much thinking where you run a basically you run a boot camp type deal.
train your recruits up and then they fight in a in in in sort of a a war or tactical thing after they're they graduate and they either get killed or I don't really know cuz the one time I the one time I played through to combat my dudes got wiped out by a tank so I don't know if they go like back to boot camp for another 12week training course. I know you get new recruits after the battle phase because I didn't lose but I'm not really sure what happens after that. But it's made by Carlos C. So, I think it's like a oneman oneman shop.
Um, and it's the sequel to Full Metal Sergeant 1.
>> This guy made that um Until the Last Plane, that little airport or airplane management game.
>> Oh, I did recognize that.
>> Yes, he did. I actually like Full Metal Sergeant more than the airplane game, but I I didn't think a ton of I didn't I wasn't a big fan of until the last play.
>> Okay. Interesting.
>> So, yeah. So, it comes out on Well, it'll be out by the time this posts. The Full Metal Sergeant 2 comes out on the 13th of April. So, I'm sure we won't be editing this that fast.
>> I don't know. Maybe maybe you'll get your inspiration and it'll hit.
>> Yeah.
>> You'll just knock this out in a night.
>> Well, we'll see.
>> Sorry. Do you you said that you are now playing the sequel to Full Metal Sergeant?
>> Yeah. Full Metal Sergeant 2 comes out on the 13th.
>> Oh, okay. I need to ask for that because that I that's cool. I didn't play the original either. So, um, but I like I did play his the airplane one. Maybe it was a little simplistic.
>> Uh, yeah. Until the last plane.
>> Yeah. But so this one came out after the airplane one, I guess. Right.
>> Well, so Full Metal Sergeant 1 came out first. I think that was his first game.
>> 2023. Did it come out like Wait. Yeah, that that doesn't seem right.
>> Oh, no. You're I'm wrong. Until the last plan was 2021. Man, my timing is all mixed up.
>> Yeah.
>> So, until the last plane was March of 2021. Full Metal Sergeant.
What's the 1942?
He has a full metal. He has until the last play 1942.
>> Yeah, I guess he made a sequel, huh?
>> That one looks a lot more just like the old classic 1942 game. I have to check that out. But and then Full Metal Sergeant uh came out in 23 and Full Metal Sergeant 2 is coming out in 25 or sorry, 26. We're in 26. Oh my god, >> I keep messing up this year too.
>> It's already April. It's like what the [ __ ] Anyway, yeah. So, that's what I've been playing. Not not a lot else. Trying to think. I feel like we're in a little bit of a a doldrum right now of new games that are like inspiring me to want to play right now. I'm keeping a loose eye in Task Force Admiral. I come to back to that every once in a while when they release new updates. Keep meaning to jump back into CE Power with some of the new campaign update, you know, that came out what a month or two ago now, new quotes. Um, but I just, you know, I haven't there's not a lot of games out right now that I'm like, I want to go back and go play that right now for for whatever reason. and there's just not kind of feels like we're in a bit of a a lull right now.
>> I can understand that.
>> I do think there's a bunch of games coming out toward the end of April. So, it'll be, you know, maybe by the time this comes out, I'll be playing be playing a lot more stuff, but >> Okay.
>> I'm intrigued by dictocracy.
>> What is that?
>> It's like a political thing. I don't I don't know. It looks interesting. You take charge as a president of a newly independent country and I don't know.
Looks like he might be bad, but it could be interesting. I don't know.
>> You You got to love that.
>> Yeah. And uh you know, Rasput, we'll see when that comes out. I know that was something that was on Finn's like top top gaming list.
>> Yeah, I guess I kind of agree with some of the the little game like we've it's been an action-packed couple of uh months though, I feel like. So, Task Force Admiral, lots of other updates with >> There's a lot of updates coming to Task Force Admiral and C Power. No, no doubt about that. I'm just thinking like in terms of like new games that come out that I'm like, "Oo, I want to go look at that." I haven't I haven't seen a lot of that the last right >> I don't know month or two. Not like there aren't enough games in my library that I haven't played that I I couldn't couldn't spend some time with.
>> How's War in Spain?
>> I don't know. I really need to actually try to play that.
>> Yeah, I'm just teasing you.
>> That is a game that's installed on my desktop. Does that Does it count if I never play it?
>> Sure, why not?
>> All right. Well, anyway, let's get to today's topic. So, you know, today's topic is AI, but not the Gen AI stuff, just to be clear. We're going to be talking about AI in games. The opponent AI, you know, the opponent that you play AI in games. I think it's a it's an interesting topic to think about like what makes for a good AI, like what you know, what games have good AIs, like those kind of things in terms of we always like to complain about AI. And I I do wonder if there's certain things that are harder or certain games that it's harder to do well. But yeah, that's going to be our topic today. trying to think of the best way to start this start this topic out because it's pretty broad.
>> I was thinking we could start off by reflecting on some games that have had like really good AI or really I mean just some memorable experiences to get us started. And so I remember when I was first playing um War Games, I encountered Advanced Tactics Gold and it has like a pretty good AI. It's not like a M. It's not like the best out there, but one thing I found very interesting about it is that the AI is extremely aggressive and it's not what I am used to seeing in most war games. Um, you get pretty passive AI and this was just on a normal difficulty. It was like really trying to eat up all my flanks. It was like so aggressively attacking me to its own detriment to some degree, but it was just crazy how aggressive it was. And I I was like awe. I was totally in awe of it. So, like it was a game design decision at some point that the developer Vic um he made his AI really aggressive and um I don't look at I'm still talking about to this day that I remember the experiences frightened to hit end turn to know what the AI would do to me and like hunkering down and it was really a beautiful experience. So, what I I guess the way I would start this off is by by saying that the way a person designs an AI has obviously such a huge impact. I mean, it's a no-brainer. It has a huge impact on the game, but it it has not just on the the game experience as in the difficulty, but even like the personality of the AI you can build in so many different ways, and it's worth talking about because it it's such an important part, especially of strategy games, but probably like every game. So, >> that's a good point. Just the different uh styles that AI can have. I'm trying to think. I There was a game that I played a long time ago and I want to say it was a strategy game and I'm blanking on it, but there were different parameters you could set the AI or you could just keep it kind of lucy. It'll do whatever. It'll be random, but you could set it to be like hyperaggressive or more passive. Uh, I guess Task Force Admiral actually kind of has this where in the mission generator you can set like the competency and the how aggressive the AI will be with launching aircraft and stuff like that. And I think having a variety is a pretty important thing when it comes to immersion as well because if it's always just hyperaggressive, you kind of know what to expect at a certain point.
>> You know, you're talking about games that had an impact on you too. I know we talk about a lot on the channel, but Sidmer Myers Gettysburg, it did kind of what you're talking about, Wolf, except it added a layer of personality to it that always stood out.
>> Mhm.
>> When you picked your your AI opponent or whatever in in Sidmer's Gettysburg, they tried to personalize it. There were these generic, you know, very easy, easy, normal, hard, very hard. But instead of just using those terms, they also like they associated each one with a general.
>> Okay.
>> At a different level. Francis Barlo like a division commander might be easy.
Hancock for the if you were playing as a confederate it would always be the other side. So like Winfield Scott Hancock would be the hardest difficulty cuz he's sort of like the best Union general at Gettysburg in a lot of people's accounts. And so like you know if you were playing as the Union the best would be a Lee, the second best would be Long Street. The third best would be I don't know Pender or something like that or maybe Hill. And then like the easiest would be like a division commander like Henry Heath. So, like they tried to personalize it, but then they also had some, you know, if you went into the settings, not just the generic, I'm going to pick my opponent first. If you went into the actual settings, then you could actually change individual traits.
So, you could have a AI be direct, which means they'd launch a lot of frontal assaults. You could have them be flexible, which would mean they do frontal assaults or flanking, or you could have them like focus on flanking.
Um, and I don't remember all of them offhand, but there were like three different traits like that that they had. And then they still tied all of those back to a historical commander based on what you chose. So like flexible it was so it was those flexible direct and that then there was also like aggressive cautious mixed or something and you could still randomize it but it was cool too cuz you could like oh now I'm playing Grant even though Grant wasn't at this battle. Like if you picked certain traits then it would assign historical commanders even if they weren't at the battle which was kind of it's kind of a cool little like flavor piece to it. I will admit I don't know how medium easy hard like that all definitely played that way but I I don't know how I can't remember like oh they really flanked a lot when I chose that trait. I don't remember if it really mapped to actual tactical decisions but it was cool from like I still remember this 20 years later you know type of deal or 30 years later almost.
>> Yeah. A lot of flavor there. And I guess like I this is my problem with I'm thinking of it from a flight simulator perspective I guess because that's kind of when I hear game AI I think flight sims and I think flights sim AI has taken a dramatic nose dive uh since like the times of Isisle 2 1946. Like I I legitimately believe it has gotten worse. And a lot of that is because of developer priorities. And I don't know if that's kind of a similar shift you see in a lot of the, you know, wargaming space where the developers prioritize things like eye candy or coding in new features and uh not so much the AI. Can I double click on that a little more though? Can we um what in what way I mean can you expand on this?
Can you explain how what do you mean when you think that the AI has gotten worse? What what does that mean for a sim game? Does it mean like that you watch airplanes just >> go in circles, don't utilize real tactics, hardly evade or when you play Isisle 2946 for example, or wings over flers fields is the best example. I'll just use 1946 because it's kind of a catchall. The AI does cheat like and it did like visually sometimes you would see it kind of jitter and warp and do really weird things and people hated that. And the Isisle 2 Great Battles team was like, "Okay, what we're going to do for our next game is the AI is going to have all the exact same parameters as the pilot."
>> Whoops.
>> And that causes so much strain on your CPU.
>> Yeah.
>> And it just overcomplicate and like the AI is not good in comparison. They don't evade like they would in 1946, things like this. So, in some ways, it's cool.
They handle their own fuel mixture and you know run their aircrafts efficient efficiently like the player and are under all those same doing all the same things as the player. But it's like at what cost for something that I the player can't see and eventually it gets to be unemersive when planes are just kind of going in circles and not really evading when tracers are ripping past their wings. Beautiful. I I just think that that's so analogous to what you see in strategic games where people are like, I want an AI that, you know, really plays like a human and then you give them an AI that plays like a human that is like a good challenge and they're like, >> "Yeah, >> man. This AI is just so hard. Why did they make this game?" It's different obviously because in strategy games when people talk about a game AI which is good or like humanlike, they really want it to be like good in in theory because they want a challenge. You know, there's a lot of strategy games out there where the AI is >> right. It's just it does very incompetent things and it can lead to a like a too cheap of a feeling when you have a victory. So it sounds like in sim games it's actually the opposite though.
>> Yeah. Do they Is that true though Tortuga? Do like Yes. I'm not saying people don't want a challenge. But I also think there's an element of like people want the appearance that the AI is like they want the feeling of oh it's a real human and I'm playing against them or it's using real tactics or whatever. But no one wants to fly against in a video game very there's exceptions obviously but no one wants to fly against Baron vontoven because they don't want to die every single time they're flying because you're not as good as a real pilot was you know in in that conflict probably >> okay >> and no one wants to fight against Napoleon because you will probably just lose unless you're the 0.1% of people who are better than them.
>> Um I I I understand what you're saying.
I think the problem with modern flight sim AI is it's not good and it's not compelling to fight against. Also, I probably have more virtual flight hours than Manford von Richtoven. So, there is that component as well. Players will do things and prolong dog fights longer than any real pilot would, right? I mean, you hear about World War II dog fights or World War I dog fights. I mean, they were a lot of the time quick affairs because the person who was shot down didn't know where it came from.
They never saw him coming.
>> Yeah.
>> That's not how it is in like, I don't know, flight sim. So, like the the thing that an AI has to do is deliver a challenging experience. I think that's very important. It has to be challenging um even to experienced players, but also just like believable, like compelling.
If a plane's shot up, they run home. And it doesn't it can't be cheating. Like I think that's the problem with Total War AI, right?
Everyone knows that if you up the difficulty on a Total War game, how they make it more difficult is it's all economic and they can recruit, you know, however many more units and people that's not fun. That's not interesting.
>> Yeah, that's a good example, too. You know, a lot of times it's there's nothing that you really change about the programming of the AI when you go to higher difficulty levels is just getting unfair advantages, >> unfair buffs. Yeah. Yeah.
>> Yeah. I think the cheating like I think but I think the cheating piece comes back to the I'm not saying you're necessarily wrong, Wolf, but I do think there is at least there is part of this is like the whether you really want it super realistic or not. You want to you want it to look or feel like it is. And if if the AI is constantly cheating, obviously that's not going to feel authentic to, you know, to the experience of the player.
>> That's weird though, cuz I think you have to >> I think you can hide the cheating.
>> Yeah, exactly. You have to do the cheating in a way where the player doesn't observe it as much or minimize the because it's if you don't do the cheating, then the AI is not going to be good. If you do the cheating too much, you there's just a tighter walk there for the for the game designers. Well, I was just going to say that a if you think about it too much as a player, any AI is going to fall apart under, you know, scrutiny, you know, at a certain point. That's just the reality of it.
They're still going to do something weird that's not immersive and may take you out of it a little bit. Um, and I guess I've kind of come to accept that with AIS in video games, but yeah, there's still a way to deliver a mostly believable experience a lot of the time.
>> That that was my thing with the Advanced Tactics Gold AI, though, because for some reason this like ultraaggressive whatever. I mean, I knew how to play against it. I knew how to like thwart it, but it it it always felt pretty I don't know. It it never felt dumb. I was actually incredibly impressed that this oneman development team was able to put on this AI, which I really don't know how he did it. And I I probably could just reach out and ask him. Just give me a tutorial on how you made this beautiful AI. And I I don't know if I'm the only one. Maybe it just for some reason that AI just clicked with me in a special way.
>> Did you feel that way with his later games?
>> Um yeah, I think so. For the most part, I'm His AI is is really good.
>> Yeah. Cool. And I don't understand why or how. I I wish I understood the mechanics behind it because I feel like it would be really interesting to study um from like a game design perspective how I mean I'm sure under the hood this is all about building the right cost function for your AI to observe and you know do its die rolls over. But how you do that well I just that's a that's like a it's black magic how he does it to me.
I don't know how >> I suspect the aggression is playing a big part of that. Like I I'm not, you know, not dismissing the AI being good compared to other games, but I do think if the AI is generally aggressive and it doesn't give the player time to think, it does become more challenging in of itself because the player doesn't have as much time to think or act. I think a lot of games fall apart where the AI isn't aggressive enough. And now I, as the player, like in Grand Tactician for example, I can always find the enemy flank. If you give me enough time, I can always find the enemy flank and roll it up. And the only way I'm not going to be able to find the enemy flank. That's a real-time Civil War tactical combat game. The only time I'm not going to be able to find the enemy flank is one, if there's a time limit, which you know those type of games often have or don't have, but you can usually it's not like pressing on the gameplay. Um, or two, if the AI attacks me while I'm trying to find the flank, right? Like if I'm trying to go around the enemy army, but they're coming straight at me. Well, now I can't really I can't get into the ideal position and so now I'm reacting rather than, you know, the AI not reacting to me. Um, and things like Grand Tactician, they try to improve the AI. Like the AI will refuse its flank to make it harder for you to flank, but generally you can still always get around it. And so I think that's like one of the one of the limiting factors is if the AI is too passive, it lets the player dictate too much and that just makes the AI inherently easier to thwart.
>> I you know I I'm thinking about this more. I I definitely agree with that point cuz back to flight sims the 2 AI is passive. The DCS AI is its own nightmare. That's just bad because it's it cheats in unrealistic ways and it's not even good. So just this kind of UFO stuff, but Wings Over Flanders Fields has different levels. Some pilots are very passive. Some flights as a whole will be very passive. But there will be times where like a flight will trick you. They'll drag and bag you. They're hyper aggressive. And like you said, whenever someone's actually shooting at you in one of those games, like there's not there's no time to think and it definitely puts you under pressure and makes it seem a lot better than maybe it might be. Yeah, that's a that's a good point. And I even think about that.
>> I also have this one scenario for you all for for just flight sim talk. I kind of imagine that like when you have inexperienced pilots, those are the ones who are less likely to know where to look and like how to they're the ones you're going to catch like flying straight and you just shoot them down.
So I I have this um funny question that cames comes to mind like do you think if you make the World War II German campaign for the start of Barbar Roa we know that the Lufa just just decimated the road. Yeah. So I mean if you made a video game about that would it even be fun? I mean it would be like yeah the AI is so stupid. I hate this game and game develers didn't do a good job. But in fact they like correctly modeled. I think there's a way to do that where it's still compelling. Like the German AI should be better as a whole, but this is something that kind of flight sims and I'm sure a lot of other games, historical games run into.
It's like people want the Battle of Berlin to be done in a flight sim. And it's like, okay, if you're flying as Germany, >> Yeah.
>> or even the Soviets, it's not going to be very compelling at a certain point realistically. But of course, you kind of abstract it. You have them flying more sorties than they were in reality.
And >> yeah, to because at the end of the day, it's still a game.
>> Yeah.
>> It's not like a onetoone exact recreation in that way. Um, none of these games will be. I mean, they don't model things like I guess some do, but fuel, you know, interruptions, ammunition problems, sabotage, things like that. I mean, that's that's kind of hard to model in these simulators and a lot of sim devs just decide not to go with it because it'll probably create a lot of complaints on the forms.
>> Yeah. What Falcon 4.0 is the one of the few that has done like a compelling like we're doing a big campaign that's not just you. I feel like >> Yeah. But I I actually kind of like the way that Red Baron I played a lot of Red Baron 3D growing up and I remember that was like the ground combat was it wasn't really a thing, right? Like the front line moved as it should historically.
you got like news reports of stuff that was going on and maybe that would influence some of the missions that were being planned and things like that, but it's not like you were like helping anyone drive forward on the front line.
And I think that's one of the maybe one of the reasons that air combat in both World War I and Korea get so much sort of romanticism is the front line kind of bogs down for large parts of those conflict. And so the only area that gets a lot of like attention because things are happening in kind of a flashy nonmud grindy way is the air war. That's exactly why it took off in Korea and World War I. But yeah, I mean, newspapers were looking around and like there's nothing exciting happening right now. Let's look to the sky where, you know, there's a race of aces and stuff like that, >> right? I'm curious, and this is, I think, relevant for both strategy games and simulation games. In my experience, some of the best AIs that I have run across, I think, are in like tactical war games that are scenariobased.
And so I wonder how much of the opposition is really quote unquote AI and how much of it is a script because you've got a defined period of time you're fighting in a defined sector of the battlefield, a defined number of units, a defined, you know, the maps have specific objectives you're trying to attack and and take in a way that like Total War doesn't have, right?
There's no magical hill you're trying to take in a Total War battle. a battle about Gettysburg, you were trying to take little roundtop. And so I have to imagine when you're programming an AI, if there's a defined objective and a limited time to do it, it's easier to like script an AI out in a way that is more challenging and comp and competent because the variables are less, right?
There's less uncertainty about what's going to happen or who where you're going to fight or what over. I imagine in a flight sim that's really hard to it's so big and open that maybe that's why you've got some of those issues with planes not knowing what to do because there's so many unknowns as opposed to you know attacking Cemetery Ridge at Gettysburg. I guess the same's probably for any of these real like you said real big, you know, grand strategy games. Um I guess the problem you might you may run into with those more curated scripted experiences is replayability.
Yeah, I mean that I think those at least the war games tend to benefit from. I want to refight um the battle at Litter on top a million times over. I don't want to fight a battle over a no-name ridge that I have no idea why it matters, right?
>> At least for like a lot of the player bases. But that being said, I do think you're you're right. They do tend to build in varants. So, like when I was working with the Scourge of War stuff, you might have one scenario, but it could have like six or seven different potential scripts that would fire >> based on, you know, just a die roll that I think is determined by, you know, the historical outcome would be the most likely, but then there'd be like different sort of flavors of that. Um, I know the Waterlue scenario at Scourge of War Waterloo, there was always like a percentage chance that Grus would arrive with his reinforcements, so it wouldn't just be Napoleon getting stopped by the Prussians at the end of the battle. the the French reinforcements could actually arrive and so it kept the player on their toes a bit to be like which which case am I going to get you know >> oh interesting >> what about what about you Tortuga what are your thoughts like and I'm thinking like big strategy games right cuz like I do think the tactical ones have an advantage that the scope is more narrow but like have there been any grand strategy games that have done AI right or compelling and >> well I don't know that's that's pretty tough I think most of my examples of like AI when I think about it is is more tacticalbased I I think that yes, when you have an AI that is uh scripted, it's more likely to perform well. A lot of campaign editors that even they allow the players to access, usually it also allows access to some kind of scripting to get the AI to do what you want.
>> I know scripting is huge in like the command modern oper enable operations um community.
>> Yeah. Not only that, but there's the flashoint campaigns. They have a a pretty reasonable AI if it's scripted. I don't know if I would trust it that much if it's just left to do what it wants.
It's probably true for like combat missions, all these games. It's just, you know, you have so many different choices. It's a very complicated, complex problem. Like when I think of grand strategy games, the Hearts of Iron series is one of the first ones that comes to mind. Maybe Europa Universalis.
And these are kind of plagued by really bad army AI. I guess it's a little easier to like guide a nation through what to do economically because the good news is that those problems are a little bit more constrained. I would say you have like simpler choices to make like what to build, which um resource thing resource building to select next or what research, those things are a little bit easier. But reacting to a player's army, I mean, the Battlefront, the battle line AI in Hearts of Iron 4 was just, it was so bad for so long. And it's probably a problem that just even persists today through Vicky 3. I don't know. I just first of all, I should um disclaim that I'm not a fan of it. They make the player now use the battle line AI. So, in some ways that's good because, you know, now it's an even playing field, but it's the AI is really incompetent.
So, they make you use it so that you're both fighting with one hand behind your back, but it's very like I think that a lot of these games that involve a battle line, a front line, and war and all that. I think most players, at least I'm going to pretend a lot of people are like me, we like to see the battle. We like to control it. you know, even in a time a lot of people would do in Hearts of Iron um 4 or Hearts 3, they'd put a lot of things on automated, but they'd at least control like, you know, the major armor division. And I think you can do that kind of stuff, but when the front lines are just behaving in such an awkward way. I'm just realizing I'm talking a lot about this game and, you know, basically berating it. Um, this is, I guess, what I would file as a bad example of AI that you didn't get things quite right and then because you're you're kind of like trapped in this game design system where you set up this Battlefront thing, the only way to save it is to force the player to adopt it as well. So that at least the player, you know, it'd almost be like what Wolf was talking about earlier for Sim Aai and stuff if like, you know, the if the the plane was like really performing poorly under AI hands. So they just the only way they could resolve it is by forcing the players plane to also have really poor performance. It's like, okay, that's not the experience that we wanted, but it is a way to make the fight fair. I guess I'll try to think about a good example of grand strategy.
That's a tough question. and THG. Maybe um the Gary Griggsby games are the best ones that come to mind.
>> Yeah, they're different though because I think they do a I think they do a I don't know if they're good or not. Gary Griggsby's War in the Pacific is sort of known as being the AI will be competent as Japan for the first 6 months >> because it's scripted >> because they know roughly what the J even if there's variance in there, they know roughly what the Japanese objectives were, right? And there was sort of a a playbook you can use up to Midway. And as I understand it, what will happen a lot of times if the kiddobai is not destroyed around maybe, you know, maybe up through Guad Canal because they probably have a way to figure out what to do with it for that that type of stuff. But if it's like 1943 or 44 and the main Japanese carrier fleet still there and Japan has taken way more territory than they did historically, then the AI doesn't really know what to do. And I've heard like or I've seen a ARS where people like find it out of fuel at some random island just doing nothing for the rest of the war.
>> I mean that's kind of realistic though, right?
>> I guess but not for the right reasons.
Um but yeah, maybe it is realistic, I guess, now that we say that. But I, you know, I think that's an example of you have a you know, it's a more limited game in many ways. So you're as the allies anyway, you're not managing the economy. As the Japanese, I think they can honestly just script the economy, right? Like you can edit it as a player, but the the computer just says, "All right, this is what you're going to do.
Define number of turns, define number of, you know, whatever, and just basically produce what was produced historically, and then you don't really have to worry about the economy. There's no diplomacy. It's really just like what bases should I take? What unit should I assign to that? What ships should support it?" And in that way, it's a much more limited limited scope for the AI to cover. I think it does a decent job. I think it does a decent job in the War in the East games, too, in in uh there. But again, I think that comes back to well, we kind of know the path that the War in the East took. And it's really hard in War in the East for the Germans to win if you're not a human like finding exploits in the system or just being that much better than like a human opponent.
>> I guess I kind of like that. Personally, >> I think the Griggsby games would lean into like, yeah, well, Germany had a lot working against them. They probably weren't going to win in the Eastern Front. And so, like, >> yes, we make it really hard because it's supposed to be.
>> Yeah. Yeah.
>> You know, like same same for Japan in the war in the Pacific. Like, realistically, could Japan do better than they did historically? Absolutely.
Were they realistically going to overcome that production sledgehammer that the US had going and that they were, you know, pumping into the Pacific? I mean, probably not. They probably could have delayed him for longer, but that's probably about the best they could have hoped for, you know, is >> not with that attitude.
>> THGS here in the bunker would have been executed.
>> The bunker. What Japanese bunker is there? Um, but but anyway, I'm not They had They had bunkers. They had a They had bunkers.
>> Yeah, just go to Euima and ask me about Japanese bunkers. This is what Okay, I thought you were making a reference to Downfall in Hitler's bunker, but you know, >> Herito definitely had a bunker, like an air raid bunker. You know, >> really, we didn't bomb the emperor. We were we were noble fighters.
>> Okay. Yeah, but you know, all it takes is one strike. You don't think he was rushed to an air raid shelter every time a bomber was going overhead?
>> Now, I want to know. You're probably right, but I want to know. Let's ask a >> Finnish Jagger told me I actually asked this recently to our local expert and he said yes they they did have a air raid complex for the man.
>> Okay. All right. That makes sense.
>> It was uh built with Korean slave labor and then they killed the Koreans once it was built.
>> Real.
>> Yep.
>> That's Wow.
>> Yeah. Japan and Korea not not not big pals in uh in >> Yeah. Japan during World War II and anyone though. It's just like, holy cow, the atrocities.
>> Yeah.
>> Well, we got off topic on the AI stuff.
Anyway, but Grand Strategy, I I want to say that in the vein of War in the Pacific, Gary Griggsby's, there was um another man, I'm I'm afraid to say it's another game by Vic um VR Designs, Decisive Campaigns, Barbar Roa. It's it's like a you can think of it. It's just another hexbased game. So, it's I mean any hexbased thing that's in my opinion, it's basically it's tactical.
It really is at the strategic level though I think well I guess you would call that the operational level probably but that one and and war in the war in the east are basically at the same scale and I think that decisive campaigns Barbar Roa has a really good AI as well.
Um I'm not sure how much that I'm pretty sure that that is not due to scripting as well. They just probably because it's super aggressive.
>> Is it aggressive if you're the Well, I guess the Soviets did a lot of counterattacking so that would work.
>> It is. Yeah. And I honestly when you play as the Soviets, it's so difficult because that game only goes for one year. So there's no like turnaround point. It just kind of sucks. I I don't understand the purpose of having I mean no most people play as the Germans.
Probably 90% of people who play decisive campaigns Barbarosa are playing as the Germans. It's more fun to play that way too. But you can play as the Soviets.
You can do multiplayer. I don't understand the point of it because basically the the Soviets is just it's about like trying to not die as quickly as possible, you know, slow down your own death because you never get the turnaround. You never get the comeback moment, you know, you just get beaten and beaten and beaten and you win basically if you weren't totally destroyed and you somehow managed to hold on to Moscow after giving up, you know, hundreds of miles of territory.
>> Victory is not losing.
>> Yeah. Yeah. Grand strategy games. Let's get to one. Let's get a grand strategy game with good AI.
>> Grand strategy. I was going to go with Cauldron of War Barbar Roa, but I don't know that the AI is good in that so much as like >> No, it's just good game design >> game. You're playing the environment more than you are, you know, the AI is almost the environment.
>> I would say that that game has is not AI like it's not about the AI in that one.
It's really just about good game design.
>> Yeah.
>> Which is definitely a way you can support AI by giving it a good design structure. I mean, a game design structure to function within.
>> Yeah. Because I mean the it's it's like you're fighting the distance. You're fighting the logistics. Yes, the AI the enemy is there, but it's it's more of like you're fighting everything around you and the enemy is part of that, but it's not that's not the key thing holding you back. Like clever flank attacks by the Soviets or something.
Speaking of another game where it's like I don't know why you'd want to play as the Soviets, but you can.
>> Which game?
>> Cauldron of War Barbarosa.
>> Actually, I was thinking about that game. I actually like that game quite a bit. Yeah, I think I'm not as big of a fan of the Stalenrad one, the sequel, but >> I like the Stalenrad one a lot.
>> You do? Okay.
>> I like the more I guess fighting over the city and just com, you know, I I I enjoyed the Stallingrad one quite a bit. I was also, believe it or not, that's the first time I've ever uh played by email. Played Finn in that one and uh it was quite enjoyable. But yeah, he liked it too. I like that game because it's just overreaching decisions. very broad that a general would do. You have your, you know, underlings figure out how to execute your orders.
Now, I want that in the Battle of Berlin and I want my orders to be incredibly unreasonable if I'm playing as the Germans and nothing happens. Nothing good happens ever.
>> Can you issue orders to units that don't exist?
>> Yeah, exactly. I want to be moving paper units around. Not not saying I would want this for the Battle of Berlin or anything with World War II, but it would be interesting if you were playing if they had a game >> where you were the commander of an army that was like besieged or beset and you were like increasingly losing your hold on reality. You have to try and decipher like what is real and what isn't. Like almost like you're an un unreliable narrator, the player.
>> So this guy just wants to play as Hitler but wants to remove Hitler from the equator.
>> Oh, I do not.
>> Yeah. No, I mean not a historical context, but let's just pretend there's a situation where a person is losing their mind as they're starting to lose the war and they may or may not have a suicide pill they could take at the end of the interview. Who knows?
>> This is all [ __ ] you added. Let's be clear. I said playing a game where you have an unreliable character as your leader in a battle would be interesting.
>> Could call it the hike rag.
>> God damn it.
Oh man. Yeah, that's good. I like it.
>> I want to be clear on this podcast. I do not want to play a game where you are playing as >> well person.
>> You you could spin that in a let's let's bring that back to a place where we can actually talk about it. I think that that's a game where essentially it's about intelligence, right? So you can have that same game design without a person going crazy. You can just have very inaccurate intelligence. It's the same difference, right? You just don't know what to really trust.
>> I think Ruse by Eugen Systems tried to do that.
>> What's the game called?
>> Ruse by Eugen Systems.
>> Yeah. Yeah, true. I actually think that the better >> I don't know that they did a good job of it, but they tried to have like, oh, there's radio signals over here. Are they going to attack here or over here?
You know, like that kind of a thing.
>> I think a better version of that is actually what War in the Pacific does. I have not seen a game which has a better fog of war than War in the Pacific. I think that that one does the best job.
And the reason why I think most games don't do that, okay, separate podcast topic, I know, but I feel like the real battle in in wars is like logistics and and intelligence.
And that's just not fun. It's not fun to like act on incorrect information and lose because that sucks. But that's reality. That's like how the wars are fought and won is, you know, you got to have intelligence, recon, all that. Um, and it's really, especially on in naval games, I think something like War in the Pacific, just think how important it is to know where the enemy fleet is, like how what what a huge advantage the US had in the Pacific because they had better radar, they had they had the code cracked and all that. So anyway, >> and the intelligence report in War in the Pacific will tell you like when an AI does something or the human player you're playing against when they issue certain orders, there's a segant section that will tell you certain snippets of what they might be doing. Um, so like if a unit's preparing to attack a certain base, if a carrier is spotted in a certain port, if a unit's preparing to land at a certain location, it'll tell you that stuff. Now, won't tell you all of it. The Americans get a much better sense of those orders than the AI does.
I assume if you're playing with a human player, they know this, so they will be like, I'm going to prepare for stuff I have no intention of doing to try and fog, you know, muddy the waters a bit.
But it is a pretty cool little feature.
>> I love that.
>> Yeah, it's it's does work. playing as a playing as a player against another player and doing like like Masarovka or whatever like just moving units around just to trick the other player. That's I love that.
>> Of course, there's the risk that the intelligence report that gets passed to your adversary is the real move you're making and not the clutter you're creating to try and, you know, mask it.
But yeah, I think it also updates like it gets more or less effective depending on what part of the war it is and how well the other side was reading that the communications because doesn't Japan at some point change some of their ciphers and then the US kind of gets less accurate until the rest of the way.
>> Yeah, that that's something when we talk about World War II, it's always like, oh, and then we cracked their codes and that was it. That's kind of you know the story. But yeah, it was a constant struggle to >> continue to be reading their mail for both in Germany as well.
>> Yeah. And I I know didn't Japan they had like they very rarely used it which is probably why it never got cracked. But I thought they had like a a super secret code and it was very rarely used and we never cracked it. But we cracked like the more common one. But then I want to say it was like after the reporting on Midway they like leaked that we c we did something and then Japan like found out and maybe I'm confusing this with something else and then they changed their codes.
>> Got all of that. Interesting though.
Well, did we ever land Is there a good strategy AI game? Is there is there one?
>> I mean if you guys have ideas, please chirp in. But um there's two examples I've been thinking about as this episode progressed. Um one is the Total War AI.
I think that that's like a like a lightning rod for people. to talk about for probably a lot of bad reasons, although I think it does a pretty good job of what it should do. And you know, there's definitely two layers to that, strategic and and tactical. I won't comment that much on the tactical one, but I almost like I like to observe the difference in AI because I I think that uh how do I say this? I feel like a strategic AI can be worse and you can just I don't know boost it with bonuses or something like the way that the Warhammer actually does those things and you can like skate by with that more where obviously in the tactical battle um if the if their troops are just like getting flanked but that they get some AI specific bonus so they have like 50% more hit points and you start losing even though you you flank them just to make the battles more fair. People would just that would be a non-starter for people. So, it's interesting that I really like Warhammer. I mean, sorry, Total War, but I think that they have I'm not sure if I've ever seen them have like a really good strategic AI. It's just another example of bad AI, but it's somehow it's okay for them to have bad strategic AI cuz what people are most interested in is not having like a totally incompetent tactical AI. Is that just me though?
>> Yeah, the strategic AI is a means to fight the the tactical battles.
>> Yeah. Okay, then maybe that's just game specific. I was trying to think if that's true of like all strategy games.
The other thing I wanted to talk about was the tycoon games. Okay, I know it's a bit of a stretch, but like we've played these games where you're technically competing against other people. And I feel like this is like the one example of a game where I really do not want the AI to be that smart. That probably in a tycoon game, the systems are not so complicated. You could probably teach an AI. there's not like you're not usually going head-to-head against a player as much. So, the AI, if they're good, it's really just means that they're optimizing their business well. And I would say that that's actually not fun to to to play against because you want to give the player an opportunity to like make fun decisions or do things like suboptimally, just play the game. They're just playing, right? And if you have an AI that's just like a super company that does everything like with machine precision and, you know, just gets everything right. Is there a particular game you're thinking of right now?
>> Well, as far as like good AI know, but like the one that I'm thinking about in my head when I'm talking about this is um industry giant um what's the what the hell's the name? I really >> Let me know when you'll find it because I have an idea of one.
>> Oh, go ahead.
>> Railroad Tycoon 2.
>> Good enough. Rise of Industry. I got it.
Rises of Industry. But yeah, that I think any of these tycoon games, they're all going to be similar. Um did you have one in mind?
>> No, not not one in particular. I was just kind of curious, you know. I think Railroad Tycoon 2's AI was interesting in that one, you didn't have to play with it when you're playing like the sand sandbox missions. You could like turn it on, include it, have more, have less uh opposing companies. But one thing I thought the AI was generally good at, like I think the game maps, depending on what you chose, you could always make sure there was enough room for you to have success without the AI.
But where things got interesting was like if you were really trying to take over certain routes, the AI would have a way of often poaching a route from you and that could be a very profitable route and you could still go there, but you might have to pay fees to use the railroad that would fund their operations. You also could charge in theory fees against the AI that to use your railroad. But the stock market element of it was always very interesting. like trying to acquire an enemy company would be interesting and challenging and you could sometimes run yourself into a hostile investor which would be an AI kicking you out of your position in the company. So like I thought as a kid growing up I remember playing this and thinking like wow this is so frustrating and it sucks and I'd rather just never have the AI. As an adult when I play I'm like yeah I want to play with this sometimes. It's kind of an interesting challenge. It's it's a very different kind of a game when you play it that way but like I I think it depends on what you're looking for out of the game. But I definitely think it was competent. Usually the AI companies would be pretty pretty soundly run. They would still go bankrupt from time to time. If you picked like four AI companies, two or three of them would do really well and one would like flame out into either stagnate or or die. And then if you bought them out, they would come back like the, you know, the the chopping off the head of a snake, but more the you know, more come or something of what is that that uh that um mythos. But >> hydra.
>> Yeah, the hydra. It was very interesting because you'd like buy them out, but if you buy them out and take their company over, guess what? you just bought all their stocks. Now the founder has all that money that you jacked their stock price up so you could buy their company and now they're going to go find found another company, you know, somewhere else. So it was it was a very atypical way of playing. And maybe it's just normal for Tycoon games. Maybe Industry Giant does something like that, but I didn't I haven't played Industry Giant.
So well, Rise of Industry is the one I I was referring to earlier. But yeah, you could I mean, Railroad Tycoon works for this. Um, Transport Fever. I'm just trying to like take a step back because we were I couldn't come up with a good discussion about the grand strategy stuff. So, I started thinking about how AI is handled in other genres. When I was thinking about tycoon games, I was basically coming to the conclusion that it's probably true for a lot of people, but it's at least true for me that I don't want an AI to be as involved in my tycoon games. I want them there like as kind of like a background presence whereas, you know, you live or die on a good AI with a good AI in like a tactical um battle game or war game.
Total Warhammer.
Why can't I say the damn name of that?
Total War Games. If they had a really bad tactical AI, nobody would play them.
It would just be like that's not worth playing. That's stupid.
>> What about Civ, like the original Civ 1, Civ 2? Was AI any good in those? Well, I was going to come to Forex next, so that's perfect segue to that. I think that it has to be competent, but good question. Yeah, that's a lot of people play Civ on like one of the hardest difficulty levels. So, I guess people really do want a challenge there.
>> Are the Civ games known? I didn't play the early ones. Are they known for having a a good or competent AI?
Well, you think I was playing colonization recently and I remember the AI being pretty competent in col colonization. Whether it was the home country AI that you're like eventually trying to break away from or whether it was the which I think maybe that's not as good, but I think the opposing colonists were generally pretty good.
Like if you're dealing with the Dutch or whatnot, I'd go into their cities and they were generally profitable and doing a good job. And maybe they were more passive, but I think I had the settings set to be that. um you know when a war happens. But I was I I always wondered if the civ games had good good strate you know grand AI if you will.
>> Yeah I finally came to a conclusion about that. I think that Civai is not very good. And the reason I'm coming to that conclusion is because I know the way that people play that game, like you play it a lot, you play it on the hardest or the second hardest difficulty. That's to have like a good challenge. And the reason why the what it happens at those is that the AI just gets insane bonuses. So, it gets to like Warhammer status where the AI is just cheating so much to be able to keep up with you. And to me that means it must not be playing the the game itself very intelligently. I mean it's defin if it needs the like 200% bonus to give the player a challenge. Obviously it's not playing very well as a player itself.
>> Okay.
>> Yeah. But I think the 4X's are probably a game where you do need more of a challenge. So I would liken them a little bit more to tactical AI. I don't know. I just thought I'd go off on this tangent cuz um we talked about this a little bit before the podcast, TH, but I was thinking about first person shooters and like what AI involves for those. And I think that that basically boils down to programming an AI for firsterson shooter. It does have to do path finding and that can be a challenge. But besides that, it's really about how do you [ __ ] the AI to miss enough that you don't like if you don't miss enough that you're just going to annoy the player.
They're going to be like, "Oh, this thing is cheating. it's, you know, no scope me from half a mile away. So, I feel like the real challenge of like building a a nice AI in like a lot of action games is just making it perform worse to the point where it performs worse than the player.
>> Convincingly dumb.
>> Yeah. And and again, most people play a game to win. And a lot of people are like accepting of losing in PvP or you know a lot of people also not accepting of losing in PvP based on my experience.
But if you lose to a bot then people I feel like there's just something that people consider like insufferable about that. So you really don't want you want to have an option so that people can play your game and not and not lose. I just have seen so many complaints. You know you ask people what they want and they want I want a really good AI. make sure this game has a really good AI and then they put a good AI in there and people are like I hate this game. It's so damn difficult. So, I don't think people know what they really want >> or they'll just instantly say it's cheating whether it is or not, >> right? Yeah.
>> If we want to talk about that, I don't I do not care if an AI cheats. I personally would like I really don't either.
>> I would like the challenge myself if I design my own game of trying to build it in a way where it doesn't cheat, but ultimately the it's about giving the player a good experience. So, I don't I think that to do that it may be too difficult to create a good representation of an AI. I know that we're not talking about Gen AI, but that might actually be a reasonable application for Gen AI is just having it simulate what a human would do. Um, that might be interesting to see.
>> I think I think cheating just you mentioned it earlier, but I I think it's one thing if the game cheats, but I don't think you want the player to think it cheats, right? Like I if if it's cheating and it's rubbing it in your face that it's cheating. Oh, look, logistics don't apply to me. I think that's really annoying as a player.
>> Yeah, actually that's one thing I can say negatively about the VR design games. Um Shadow Empire is like very famous for this, but it's also true with like Advanced Tactics Gold. The AI was not good about building its roads or its um keeping its supply networks. So you would see like really kind of goofy things like there'd be a road, a railroad or whatever, a road that just started building from the in the middle of nowhere. It wasn't connected to any logistical point and just started going somewhere and it just seemed like it was like a random die roll for an engineer to code to this hex and just build a road there because it made no sense. But the player the um AI units were never out of supply. And I guess it's just really a hard thing to teach an AI is how to stay in supply. So the you I've seen this in a lot of games, heartspiring and all that, but yeah, you typically don't try to penalize the AI for having units that are out of supply.
And people always get mad about this. I don't know what the what like what's the alternative though? I don't think the option is to build a competent enough AI that yeah, I mean, if that was an option, people would just do it, right?
build a better AI that could actually support the complex decision- making that humans do to like plan ahead for logistics.
But that's not that's not easy. It's not realistic. So, what's the alternative?
If you don't want the AI to cheat, do you just want them to only defend?
Change the aim rules to make it easy.
That way, the AI can actually understand them. I don't know. I don't I don't like any of those options. I'd rather the AI just cheat. But Shadow Empire did have like really stupid situations where you'd cut off an enemy and it would never run out of supply and it's like okay that's well we we gota we have to have like some kind of limit of the cheating.
>> Yeah, situational cheating.
>> I don't know I don't know how to answer that question like how I would do that because that just seems like a really difficult problem to solve. Yeah, we got to we should ask somebody on who knows about this stuff. And it'd be really interesting to hear if they were like willing to confess like, you know, the inner trade secrets of how you build an AI that actually functions.
>> I almost think you'd have to get a former dev because I feel like current devs are never going to publicly say, "Oh yeah, our AI cheats." Cuz >> yeah, >> it's just a bad PR. Whether it's true or not, it's a bad thing from a PR point of view to say.
>> That's true. You're right. I don't know that I have a lot else to say about it.
Like I know it's kind of it's a hard topic, right? Like >> Well, please can we somebody just give a good example of ground strategy AI? Is there a good one out there?
>> I guess my my only like additional kind of little talking point out there is I know I was just saying, oh, AI's gotten worse and some reason. I feel like we hear about people complaining about AI more now than we did maybe 10 years ago or whatever. And how much of that is just gamers are getting better? You know what I'm saying?
>> I don't.
>> Is there is there any merit to that idea, you think, or no?
>> Keep up devs.
>> Okay. Okay. Um, yeah, I guess so.
>> You don't think gamers are getting better, Tortuga?
>> I don't think so in tactical or strategy games. Um, but I think that might be true in flight simulators or sim simulators.
I'm not sure though. But I don't think that um people playing tactical well especially these the turn-based games for sure I think any turn-based game people could play just as competently 20 years ago. It's not like I at least I can't see and maybe you can enlighten me Wolf how like people playing games 20 years ago like what where do you come up with this idea? Can you give some supporting evidence or something? I think people play more games, you know, and are exposed to more games today than they were maybe 20 years ago. And, you know, people who are playing 20 years ago have been playing for 20 years, right? I I don't know. That's that's kind of my thing.
>> Okay.
>> And you kind of know all the tricks as a player after a certain point.
>> Um, and maybe this is a side effect of the genres we play generally appealing to older people. So, I don't know. But that that's where I got the the general idea because I used to think about some of these flight sim AIs and I was like as a kid they were hard. I go back now and I'm like, "Okay, this is it was hard because I sucked, you know."
>> I don't know if players are generally better. I think in certain genres they might be, especially in Sims.
>> Yeah. FPS's.
>> Yeah, I would say that for sure. Or like League of Legends. Exactly. I think sometimes players attention spans are worse now than they were 30 years ago for games or 20 years ago for games.
There's just so many alternatives. It's not even like, oh, oh, people don't think anymore. It's more like, you know, you've got so many things vying for your attention.
>> I think I think there's some merit to that that other point you said. But yeah, >> but I do think I do think what may be the biggest difference in perceptions about the equality of AI is the exposure. And what I mean by that is YouTube.
>> You know, like if you played a game 30 years ago, 20 years ago, hell, 15 years ago, and the AI did something funny or weird, you might be like, "Huh?"
>> Yeah, >> that's weird. That looks really dumb. No one else saw that. you know, to the other people who it happened to. But everybody probably thinks they're like a oneoff, right? Is that is that happening every time everybody plays a certain way? Now that gets packaged, clipped, thrown on YouTube, and everybody sees it who's looking. And so, like, it is easier to expose the bad AI >> today than it was growing up. I also think as a culture, gaming continually has this scope creep. We want more. We want bigger. we want, you know, more complex.
And I I think like especially in war games, you've seen a shift away from scenario-based games like Sidmer's Gettysburg to things like, you know, even like take uh Task Force Admiral, right? That's scenario-based. But there sure is a heck of a strong push for a campaign in that community. I don't think that was originally what I I remember talking to the the lead guy on that when they first announced it. And at least what I thought I remember them saying way back when and and they you know I think by the time the Kickstarter came out maybe they changed their mind but I thought I remember them saying they weren't really planning on a campaign. It was going to be scenario- based, >> right?
>> But now like I think it's like the last thing in the road map. I don't think it's their primary focus but like even now they're saying there like a campaign will come eventually. Yeah.
>> And I think that's just an element of like everybody wants these openw world sort of dynamic things, but that makes it harder to make a good AI as we've discussed.
>> Dynamic campaign AI is a whole another >> can of worms, especially when it's real time with these games. Y >> I mean, it was like that total war problem you were talking about earlier.
You have your tactical AI and then you have to have a competent strategic level AI like games like War on the Sea.
Frankly, I don't think either of the AIS are very good, but you can pick up on what the strategic AI does very quickly, and it's one of the reasons why I I really don't like it. Yeah, definitely makes it more difficult. And yeah, the push for campaigns is kind of funny.
It's like every game that comes out, people want dynamic campaigns. And I'm guess I'm not really interested in playing a yubot simulator or a flight sim and changing history as a single boat or plane, but it seems like a lot of folks want that. So, you know what's interesting about that is that like Silent Hunter 3 gave like to me what felt like the illusion of a not really a dynamic campaign, but it was an open world.
>> Yeah.
>> Right. Right. So you had the big world you could sail around in. And so it kind of felt dynamic. But like one U boat skipper, no matter how good, even if you sink a million tons, you're not changing the course of the war. Hate to break it to you, the allies got a lot more ships where that came from.
>> What if it was a million tons of warships?
>> I mean, even so, they had more than a million tons to spare. For sure.
>> What if you snuck into Scappa Flow as a single submarine >> and you sank?
>> You're sounding like a YouTuber, a YouTube comment right now.
I I this is my and I think this is because I come at it from someone who's very interested in the history and where gamers want to win like you were saying previously. Um and playing a game where you're playing as a hubot and the war just ends and you still lose in the end.
I can see how that's like oh well. But I like I don't know. That's what it was.
No matter like you were saying, no matter how good how many ships you sing, you aren't changing the course of the war as a single boat, a single crew, an entire, you know, fleet arm at that point. Like I don't know.
>> Not with that attitude.
>> That's true. That's true.
>> I don't even know if the Germans had enough torpedoes to change the course of the war. I I don't think we need to engage in this kind of hypothetical, but I will just hypothetically just throw out there. What if there was like a perfect yubot that operated and like just sank every British ship that came in and out of British port?
>> That would that may do it actually.
>> Yeah. Anyway, but not whatever.
>> You could have given them a fast attack nuke boat and they still would have lost. Sorry. I like chasing history and video games, not necessarily changing it. That's kind of my >> That's a nice way of putting it. I wish I could play Devil's Advocate for that, but I really can't. Yeah, I don't. When I played Silent Hunter, I never I never considered my actions actually changing the course of the war. So, I guess I'm I'm in the same boat as you. Get it?
Boat.
>> Your objective is to survive.
>> You know, that's a that's kind of an interesting thing. This maybe goes back to like AI behavior and what are people really looking for? Especially in flight sims is I also do think there's an element of players probably don't want it realistic because if they did if it was realistic they'd get a lot fewer kills, right? Like if they played the way a real pilot does, they'd probably bug out way sooner, way more often.
>> Oh yeah.
>> You know, like most pilots never or I don't know if it's most, but a lot of pilots never even pulled the trigger.
Like they never got behind an enemy to be able to shoot him down.
>> Yeah.
>> What was it? There's some There's just some absurd stat that like something like I I don't know. I don't It's like a really high percentage of the kills came from an extremely small percentage of pilots.
>> Yep. Yep. For sure.
>> Well, I have one different angle um on this. Uh I if you want a really competent AI, I've mentioned a few times, but I I think that you you can actually build your game design to support that. One example I can give is aptly named AI Wars. Do you guys remember that one or played it? It's Forex game where or maybe it's called AI War, but there's one and two, AI war 2.
And the the whole premise of the game is that it's almost like the Matrix has come true. you're up against this super AI that the whole point of the game is that you're trying to take over little things, but kind of like do it in a way which is the least threatening as possible so the AI doesn't send like everything at you and just stomp you out of existence so that there's like kind of like this threat meter >> and if you do things in like a slow enough manner like obviously at some point you cross some threshold and then they know it's go time but if you've done enough to like secure a few good systems and got enough resources is you can survive the like the attacks that they send your way. But the the idea is that the AI is just solving so many other problems right now. It's dealing with so many other threats that if you can just kind of fly under the radar, you won't be attacked. So you can like go by the enemy ships and like they don't really care about you. And you one person if you were playing in a different game would be like, "Oh, that look at that AI ship didn't even try to attack you." But that's all built into the premise of the game. So, it's one like a great example of using game design. Yeah. To support your AI or like to give your AI a great playing field because it it can operate so cleanly in this game design structure. So, yeah, just an example.
>> Oh, that sounds pretty cool.
>> And I think that like a counter example to that is when you give something too many choices. If you have like a really complicated system, maybe this is like a shadow empire problem where they the AI can't handle logistics. Well, if you make complicated logistics, you can't really expect your AI is going to know how to deal with that that like that future planning or whatever it is that get makes humans somewhat able to think about like logistics and planning ahead.
So, I mean, if you put that in your game, you are you have to have like a probably a fallback for the AI is not going to do this well.
>> Interesting.
>> Yeah. almost give your AI as narrow a focus as possible and it'll be more competent >> or scripted >> to design your game as such, right? Like >> Yeah, I like the scripting focus. I really do. I mean, I think that that's we were talking about this um not today, but like previously we've talked about how uh there's some like scripting has fallen out of favor or like scenarios have fallen out of favor. Maybe it's even playing into this open world dynamic campaign thing. But like I I like the idea of a command and conquer where I go my campaign is to follow through this series of missions that they have scripting and you know the AI is just perform so much better in that situation than if you have this a more illdefined scenario where like you the dynamic campaigns I feel like that can be a problem or if it's a um a situation where the AI is supposed to really take on something bigger than just like a single map. I don't know. I guess I'm running out of steam on things to say, though. That's That's about it. That's the empty tank.
>> Uh, yep. An hour and a half. It's been a good pod.
>> I I don't know. I don't know that I have anything else to say. I don't know that we came up with a thesis of like this is why how you make a good AI. I think we definitely discussed a lot of li limitations, challenges, perhaps identified that there is no good grand strategy AI out there. I don't know if anybody has any closing thoughts on AI, but I would love to hear people's thoughts in the chat, in the comments.
What makes a good game AI? Hey, if you're not on YouTube and you're reviewing this on Spotify, you can leave your thoughts in the review, too. Not that that would be super. I don't know if that would be good or bad, but treating uh treating reviews like comments.
>> Does Spotify have allow reviews? Spotify allows reviews, too, for podcast.
>> Gives you stars. I think so. I think it does reviews.
>> Okay, good.
>> Just, you know, leave reviews. Leave good reviews. Leave many reviews. Vote.
Vote often.
>> The various platforms want engagement.
Let's give it to them. If you listen this far into the podcast, ask ask chat GBT what you think of the single malt strategy podcast and just put copy paste, baby. Put that right in there.
>> Anyway, um all right. I don't have anything else, guys. What do you have anything or should we wrap it here?
>> Wrap.
>> Cool. Well, thank you again, guys, for coming on. Um, Tortuga, best wishes with uh with your, you know, expectations and, uh, thanks for coming on again, Wolfpack. It was great to have you. I hope you guys all enjoyed the single malt strategy podcast, episode number 98. We're getting close to 100. I don't know what we're going to do for that yet. Um, but uh, until next time, this is the historical gamer for to I was going to say Eric, but for Tortuga Power and uh, Wolfpack. Until next time, we're out.
Related Videos
OpenHuman VS Hermes AI: Who Wins?
JulianGoldieSEO
285 views•2026-05-29
Long-Running Agents — Build an Agent That Never Forgets with Google ADK
suryakunju
142 views•2026-05-30
5 Mind Blowing Omni Uses Cases
PaulJLipsky
1K views•2026-06-02
This computer is made from real human brain cells. And you can buy it.
Talktmsmedia
3K views•2026-05-28
BREAKING: Microsoft’s New Image Generating Model Beat Out GPT 1.5 and Nano Banana 2
aimmediahouse
122 views•2026-06-03
I Made the Same Anime Fight Scene in Every AI Video Generator
NobleGooseAnime
295 views•2026-05-30
Nvidia Bets Big On AI PCs | New Chip To Power Windows Laptops | Technology | AI Updates | N18S
cnnnews18
3K views•2026-06-01
I Tested NEW Opus 4.8 on Four Projects (Updated LLM Leaderboard)
AICodingDaily
298 views•2026-05-29
