9 Ways that 4X AIs disappoint us

[MadMaxHellfire]

my point is "boohoo, 4x suck because they don't do this - dude, start playing decent games and you'll see they do this - boohoo, you are trolling".
fuck off.

 

[Ludovic]

Citation needed?

Almost a decade in the games industry and double that as a programmer in general. This statement is part anecdotal, part personal experience, of course. But look at how some of the AAA games have AI much worse than indie games. Civ 6 is a good example, I think. At release the AI was embarrassingly bad.

As an aside, what passes for AI in most games is not really AI per se, it is usually just a combination of a state machine and some custom scripts / game code. One could argue that this is not exclusive to game AI - ie a lot of so-called expert systems are built in the same way.

Most "classic" game AI is built around modelling the game state, possible actions, future game state from these actions, and then via some heuristic assign the future states a numeric value. Now that's massively simplified. But the point is that, as you point out, 4X games have massively complicated state compared to chess. So this approach is not at all viable. This approach can be used to handle sub-problems during strategy game AI processing - ie how to conduct a war or split forces, so I am not saying utilizing classic AI algorithms is not useful. It is just not useful in the same way as for other AI tasks.

Admittedly, real-time games and AI pose some challenge - which makes the Paradox AI pretty impressive, imo. It is much harder to build AI that has to perform in real-time on a wide range of hardware, than an AI for a turn-based game.

 

[a cut of domestic sheep prime]

Almost a decade in the games industry and double that as a programmer in general. This statement is part anecdotal, part personal experience, of course. But look at how some of the AAA games have AI much worse than indie games.

Kewl. Just needed that confirmation.

As an aside, what passes for AI in most games is not really AI per se

Yes, yes. Real AI scarcely exists outside of some Google/NSA lab or an anime series.

Admittedly, real-time games and AI pose some challenge - which makes the Paradox AI pretty impressive, imo. It is much harder to build AI that has to perform in real-time on a wide range of hardware, than an AI for a turn-based game.

Yes, though even as a layman on the subject, I kind of suspect that it isn't really realtime. I think some things are processed in realtime, but other things are refreshed over time. For example, the "AI" decides to build X amount of ships based on X variables, buildings are built to support this. No further processing is necessary unless the variables determining the number of ships it needs change and that could be refreshed every few minutes or so. Just how I think it might work.

Still, it has to be processor intensive to even come close to realtime when you've got 30+ different empires with hundreds of different factors all interacting on the map simultaneously.

This all goes back to my initial question though - ie, is Paradox's AI crap (from a user perspective, though it is interesting that you find it impressive as a programer) because of a lack of processing power to process events in real-time, or is it simply the mechanics or some lack of skill or knowledge on the part of the dev.

In the end though, I think your initial assessment is probably correct. AI is rarely seen as a priority. Even in games where AI is touted as a feature - see Oblivion.

 

[Ludovic]

Well, for Crusader Kings 2, it is a lot more than 30+ entities for the AI, as there are hundreds of counties and thousands of characters, all capable of performing individual actions and having to respond to other characters plotting against them, wars, raiders, events, etc. So that's a lot of processing, even if it happens at preset intervals. As for the general question about their, I don't think it is crap, but making an AI which is capable of handling strategic warfare is very hard. Landlocked, localized wars are not so bad, but if you have a naval element involved and/or several nations - perhaps even multiple theaters, then creating, executing and adapting a strategy is very challenging. It's quite telling that Hearts of Iron tends to have the "weakest" AI compared to the other Paradox games, and that the naval landings are usually bungled by the AI.

Finally, the Paradox games aren't really 4X, which was the original topic of this thread. Now I think the distinction between 4X and Grand Strategy has been blurred over the years, so I think it is very valid to discuss in this context, but I also think it is worth recognizing that due to increased asymmetry, more complicated map topology and a reduced focus on optimization/snowballing, grand strategy games are in many ways a bigger challenge in regards as to avoiding "stupid AI".

 

[RayF]

my point is "boohoo, 4x suck because they don't do this - dude, start playing decent games and you'll see they do this - boohoo, you are trolling".
fuck off.

ok, thanks for clarifying

 

[MadMaxHellfire]

Almost a decade in the games industry and double that as a programmer in general.

and yet you fail to see that

Admittedly, real-time games and AI pose some challenge - which makes the Paradox AI pretty impressive, imo. It is much harder to build AI that has to perform in real-time on a wide range of hardware, than an AI for a turn-based game.

paradox games aren't real time but actually turn based.

 

[Ludovic]

paradox games aren't real time but actually turn based.

If by that you mean 1 turn = 1 day, I think that is being pedantic, and beside the point. In a turn-based game the time spent processing the AI for a turn does have anywhere near the constraints on time that it does in paradox (and other real time) games.

 

[Beastro]

With this in mind, I like you list Ray and how you decided to tackle the problems presenting themselves for the most part. A lot of the things I'd like to see addressed in a 4x game didn't make your list - but rightfully so. Again, we can hardly expect some indie to tackle problems even big players failed to come up with a decent solution to in decades.

Well, as I often say, I am not an indie game developer... I am a professional software developer who has just decided to make a game that he's gotten tired of waiting for game companies to make (a modernized MOO1). Most of my 30 years of so of programming experience lies in a lot of fairly complex engineering applications. And to be perfectly frank, all of the AI issues I listed in my post are fairly trivial to fix. I think the real problem is not their difficulty, but that solving these problems is not a priority for game companies since there's very little return on investment for it. In addition, the constant market pressure to reiterate games as new versions (Civ 3, CIv 4, Civ 5, Civ 6) almost always resets the AI for these franchises back to their starting point which means we always see the same problems.

What I listed are problems I have noticed in my gameplay and that I notice other players complaining about to this day. I'd certainly welcome input on other AI problems that players run into. If I have a list of problems, I can fix them and then open up the code for anyone else in the future.

I'd say offering a better AI would make for a better return on investment if that was mated to a decent game as a whole. About the only 4x game with a AI approaching decent in my experience was GalCiv2, but in it's case it was undermined by much of the rest of the game being shit, especially combat which itself is another thing that needs improvement in most 4x games.

Anyway, all those complains about AI tend to forget the AI is not buying your games. He is only there to entertain the player by losing gracefully.

The most fun I ever get in games is when I lose very ungracefully, fighting tooth and claw in a way that isn't being steamrolled where I keep thinking I have a chance or that it is simply enjoyable to see how long I can last until I am defeated.

About the only game in recent history that encourages such scenarios is Rimworld, but even then those moments of defeat are outnumbered by my successes.

I think a big problem with 4x games is that they fail to realize that there is are stages above an empire at its apex and that is one in degeneration and collapse resulting in fragmentation and production of daughter states that struggle to become the next big thing the big old example being Rome and the shadow it cast over European civilization even until present day.

A big reason for this problem is that these kinds of games have too clearly cut positive feedback loops where strong empires only keep getting stronger when history shows that doesn't happen. Small defeats result in lessons learned and those nations improve themselves while the ones unblemished by losses barrel head first into situations they are unable to overcome (EI: Imperial Japan): Victory is not a great teacher and it is a trap humans continually fall prey to.

My idea would be if you start steamrolling penalties of decay kick in and ramp up, not to handicap you, but to set up the inevitable fall wherein in the challenge if to go all Byzantine Empire and last as long as you can until it fragment and you then get to chose, or are randomly given, a successor state to play as as you struggle from something a square one, only with different tech and more settled enemies.

Pair that with a more drawn out tech tree that is too long for one contiguous empire to research and marry the imperial decay mechanic with the tech (make it so empires can remain stable for longer) so the game isn't simply linear expansion of one faction but factions growing, fragmenting and reconquering that better imitates the flow of history where notable fields of battle that are strategically important find themselves host of conflict over and over again. It is rare to get that experience where you, the player can note that you have fought over for thousands of years because in 4x games of all kinds the borders of your empire always grow outward turning territory into something you fight over once and never again.

paradox games aren't real time but actually turn based.

If by that you mean 1 turn = 1 day, I think that is being pedantic

If we're going down that rabbit hole then couldn't it be said that Bioware games are TB?

 

[Norfleet]

A big reason for this problem is that these kinds of games have too clearly cut positive feedback loops where strong empires only keep getting stronger when history shows that doesn't happen. Small defeats result in lessons learned and those nations improve themselves while the ones unblemished by losses barrel head first into situations they are unable to overcome (EI: Imperial Japan): Victory is not a great teacher and it is a trap humans continually fall prey to.

I think this is fundamentally an issue with most games' tech trees. Games of this nature have several kinds of tech trees that have come to dominate the system: Either the "do all the things tech tree", or the "randomly or choose parts of the tech tree you simply won't get because reasons". This, of course, is not a really realistic tech system at all. WHY, exactly, am I arbitrarily forced to choose between fusion bombs, fusion missiles, or fusion power?

Rather, the tech system should be based on needs. Necessity is the mother of invention. People invent things because they NEED it. And, as above, victory breeds complacency: If your current lazor cannons are doing their jobs right, the desire to invent a newer, better lazor cannon, is somewhat reduced: There's no NEED for it. These needs should arise organically based on what actually happens, so what you DO drives what people think of. If you're sitting there being a complacent lump with no real wars, people aren't going to think of new weapons, they'll instead invent a new system of distributing cat macros.

Similarly, the other thing with these games is that tech is a one-way street: Things become more advanced, and generally more BETTER. The new things don't come with any issues or anything. You never find yourself choosing between the new untested whiz-bangs with their bugs and unreliabilities vs. the thing which is cheap because you've been producing it in quantity for awhile and works and people know how to use it. Technology is never lost: Old things don't fall out of knowledge because nobody uses them anymore. Technology crashes in the real world happen precisely because the old things fall into disuse and people stop remembering how they work because nobody uses them, and suddenly something goes wrong and the new things can no longer be made. In Civ, nobody ever gets bombed back into the Stone Age. In these games, you never fall into a situation where the ability to make something has been lost for so long that the knowledge of how to make it is also lost because everyone who knew how to do this has died.

Finally, of course, player-based empires DON'T generally fall into decline because, in the real world, empires go into decline when replacement leadership proves to be incapable of doing what the old leadership could do. But since the player's empire is always led by the player, it enjoys a continuity of leadership and purpose than no real-world empire ever has.

 

[Ludovic]

Well, for me, that is one of the distinguishing points of 4X and Grand Strategy. 4X is about the rise to power and domination, or the failure to do so. A good 4X ends relatively quickly after the power-snowball has reached critical mass, lingering at the power apex just long enough to let the player bask in it, but not get dragged into into a grindy whack-a-mole towards inevitable victory. Completely absent are the post-dominance phases of stable prosperity turning into decadence and complacency, before finally the empire either is reborn through violent conflict, is conquered by invaders, splits into warring factions or descends into a dark age.

Grand Strategy on the other hand usually chronicles the rise and fall of states and empires, including these phases, and if they're good there are game mechanics actively pushing for these cycles to occur. I think the "end-game crisis" that is being added to many games these days is a move in this direction, and one of the things blurring the line between 4X and Grand Strategy.

I agree with your thoughts on tech trees working against these cycles. Tech is usually a snowball-y mechanic, where the powerful become more powerful. I think what is really missing is the concept of paradigm shifts. The Roman Empire did grow ever more technologically advanced over the centuries, and it wasn't just "Advanced Cavalry" and "Horde Migration Patterns" tech that gave the invading tribes an edge. It was a combination of economic, political and social factors weakening the Roman Empire from inside (events and accumulating maluses can represent these in a game), the tribal neighbors assimilating Roman technology (this kind of tech spread is generally absent in 4X) and a host of other factors, of course. Even if the player provides continuity of leadership and purpose, some of the other factors could be represented - especially the tech diffusion is sorely missing.

In a 4X game when you transition from expansion to exploiting (ie conquest), you're generally rewarded with a stronger economy, stronger tech progression, etc. Conquered planets/cities generally turn docile very quickly, or alternatively you can just raze them and re-colonize with your own people. Also notably absent is any cultural drift or provincialization. Stellaris does have the sector thing - but that's mostly an annoyance due to the AI grossly mismanaging them and not because the sectors start having their own agenda. In Civiliation and many other 4X games you get this weird "unhappiness"/civil unrest thing as you expand, where you basically just have to build more entertainment buildings, have more garrisons and/or accept slightly lower local production (which doesn't matter because your global output is still benefitting from the expansion).

This isn't necessarily a problem - not all 4X players want an empire simulator. They just want to conquer the world or galaxy, and have the biggest spaceships or the most tanks. Many of the mentioned mechanics are in place to slow down the march to world conquest. A lot of people are put off by mechanics that have feeling of inevitability. If my massive empire is bound to suffer civil war and disintegration, and I have to play whack-a-mole with rebels, I can feel the game is taking away my achievements and ignoring my awesome map-painting skills. Late game crisis mechanics are a way to create a post-dominance end-game that does not evoke this feeling: You've conquered the known universe, now fight off the extra-galactic mega bugs, who're an actual challenge unlike these crappy other empires who you snowballed past in the early mid-game. If done poorly, there can be some resentment (ie if they auto-respawn any losses and it becomes better to just abuse game mechanics to stall until some event timer runs out and they can be killed).

 

[Norfleet]

A lot of people are put off by mechanics that have feeling of inevitability. If my massive empire is bound to suffer civil war and disintegration, and I have to play whack-a-mole with rebels, I can feel the game is taking away my achievements and ignoring my awesome map-painting skills.

That is, of course, what happens if the problems in question ARE inevitable, but they aren't. The Roman Empire wasn't INEVITABLY bound to disintegrate, it was a series of growing problems either left unresolved, or resolved in a manner which created another, bigger problem. Presumably, resolving such problems could itself be an interesting game...in most games, they aren't, because managing your own empire is far less interesting than painting the map. This, of course, is the wrong approach if the ultimate goal is to make the game an Empire Manager. Managing the empire itself should be the interesting game, painting the map a byproduct of that. With games like EU and the like, peace is boring, there's nothing you can do during it. If you're not fighting a war with something, or PREPARING to fight a war with something, you're basically just sitting there waiting for something to happen, there's nothing you can proactively do or any issues to resolve.

 

[Ludovic]

That is, of course, what happens if the problems in question ARE inevitable, but they aren't. The Roman Empire wasn't INEVITABLY bound to disintegrate, it was a series of growing problems either left unresolved, or resolved in a manner which created another, bigger problem. Presumably, resolving such problems could itself be an interesting game...in most games, they aren't, because managing your own empire is far less interesting than painting the map. This, of course, is the wrong approach if the ultimate goal is to make the game an Empire Manager. Managing the empire itself should be the interesting game, painting the map a byproduct of that. With games like EU and the like, peace is boring, there's nothing you can do during it. If you're not fighting a war with something, or PREPARING to fight a war with something, you're basically just sitting there waiting for something to happen, there's nothing you can proactively do or any issues to resolve.

I see, I hadn't considered that perspective to be honest. I actually find peace during EU to be equally interesting to war time, depending on the geopolitical situation and my goals during that particular game. I am busy building up the economy, colonizing, cultural/religious conversion, diplomacy, etc. A lot of the time I find the aftermath of a war one of the most interesting parts of a strategy game. I don't really see peace as a time for preparing for the next war. In some games yes, but not always. Empire / nation management can be pretty fun, I think.

 

[Norfleet]

If managing your empire is so interesting, I ask you: How long can you play as Bhutan before you get bored and start conquering the world?

I see, I hadn't considered that perspective to be honest. I actually find peace during EU to be equally interesting to war time, depending on the geopolitical situation and my goals during that particular game. I am busy building up the economy, colonizing, cultural/religious conversion, diplomacy, etc. A lot of the time I find the aftermath of a war one of the most interesting parts of a strategy game. I don't really see peace as a time for preparing for the next war. In some games yes, but not always. Empire / nation management can be pretty fun, I think.

You don't have "cultural and religious conversion" unless you go and conquer another culture's people to convert to yours. You're doing these things so you can...get ready for another war! Because I don't really see anyone playing these games as Bhutan, staying to their fascinating little sandbox for the entireity of the game. Nobody plays like this! Imagine if running your empire was actually so engrossing that the only reason you ever started a war at all was because it was necessary to stabilize your own empire in some way, not because you actually wanted to grab someone's clay. In the real world, there's a lot of wars like that: People fight, no real territory changes hands, and you call it a day. In a game like EU? It's an utterly pointless exercise to start a war without taking something out of it. War is a tool of map painting, and not map painting is boring as fuck. You're either painting the map, or you're preparing to paint the map.

 

[MadMaxHellfire]

Imagine if running your empire was actually so engrossing that the only reason you ever started a war at all was because it was necessary to stabilize your own empire in some way, not because you actually wanted to grab someone's clay. In the real world, there's a lot of wars like that: People fight, no real territory changes hands, and you call it a day. In a game like EU? It's an utterly pointless exercise to start a war without taking something out of it. War is a tool of map painting, and not map painting is boring as fuck. You're either painting the map, or you're preparing to paint the map.

doom mechanics for precolombian nations in eu4.

 

[Ludovic]

I guess I am geeky enough to just enjoy "role-playing" whatever nation, and doing whatever feels like the right thing. Not that I don't find joy in a game of map painting, and spreading Bavarian, Italian or Ugric dominion over all over Eurasia. But sometimes I just want to stay "in-nation-character" and enjoy the unfolding whacky alternate history. Half the fun if such a game is watching the rest of the world, perhaps intervening if I think the AI is painting the map in the wrong colors. It's like when you're playing board games and you don't care about winning, but just want to ruin Bob's day and watch him rage.

But that's one of the great things about strategy games - they accomodate very different playstyles. Open-world RPGs have a bit of the same.

 

[Beastro]

A lot of people are put off by mechanics that have feeling of inevitability. If my massive empire is bound to suffer civil war and disintegration, and I have to play whack-a-mole with rebels, I can feel the game is taking away my achievements and ignoring my awesome map-painting skills.

That is, of course, what happens if the problems in question ARE inevitable, but they aren't. The Roman Empire wasn't INEVITABLY bound to disintegrate, it was a series of growing problems either left unresolved, or resolved in a manner which created another, bigger problem. Presumably, resolving such problems could itself be an interesting game...in most games, they aren't, because managing your own empire is far less interesting than painting the map. This, of course, is the wrong approach if the ultimate goal is to make the game an Empire Manager. Managing the empire itself should be the interesting game, painting the map a byproduct of that. With games like EU and the like, peace is boring, there's nothing you can do during it. If you're not fighting a war with something, or PREPARING to fight a war with something, you're basically just sitting there waiting for something to happen, there's nothing you can proactively do or any issues to resolve.

That's the big lesson the Roman/Byzantine Empire shows. The inevitability comes from the broad fact that "All things die", but they also show that cultures lasts longest against the forces of entropy when they're constantly able to find the ability to adept and renew themselves and their particular empire did so while enduring disasters and collapses again and again, of which, just one would have knocked out most other peoples.

That's the funny thing about the use of "Byzantine", the empire and its people adapted so well over time that to many looking back they'd earned an entirely new name and were considered different people.

I guess I am geeky enough to just enjoy "role-playing" whatever nation, and doing whatever feels like the right thing. Not that I don't find joy in a game of map painting, and spreading Bavarian, Italian or Ugric dominion over all over Eurasia. But sometimes I just want to stay "in-nation-character" and enjoy the unfolding whacky alternate history. Half the fun if such a game is watching the rest of the world, perhaps intervening if I think the AI is painting the map in the wrong colors. It's like when you're playing board games and you don't care about winning, but just want to ruin Bob's day and watch him rage.

But that's one of the great things about strategy games - they accomodate very different playstyles. Open-world RPGs have a bit of the same.

My style is a compromise with the power creep mechanics that make you wind up inevitably conquering large tracts of the world. I mix conquest with RPing and set some broad goals, then once I grow large enough I pick another country to play and effectively emulate imperial decline as the AI bungles while I help dig into it allowing the other AIs to expand again. I rip on Paradox a lot for their decline, but the fact that keep letting players freely pick any country to play as from any save is a huge thing for any dev to do, it frees me from simply sticking to one power and playing "Balance of Power" knocking down the strongest factions over and over until I get border or eventually take over most of the map.

I anticipate the harbinger of the end for Pdox will be their decision to stop adding that feature into their games as a given.

 

[Jeff Graw]

There's also the divide between competitive AI and immersive AI. AI personalities, for example, play a huge part in AI quality, often negatively. Giving the AI personalities makes the game feel more alive and causes it to seem like it has more character, but it also means that various personalities are going to be better suited to playing the game than others

I agree completely with this! To me, the point of the AIs is to create both an immersive and challenging experience for the player. I believe that it's possible to accomplish both of those ends without cheating.

Sure, although both of these are relative ideals. Any instance of role playing is going to weaken the AI, so if your AI is still challenging at the end of the day it's in spite of that particular handicap.

Of course, there's no reason one can't have an AI role play less as difficulty approaches its maximum. I think that's the ideal.

Interesting article, and kind of funny that it uses Master of Orion as an example, since I think MoO1 is one of the few TB strategy games with a decent AI that keeps providing a good challenge for experienced players.

Only because it gets massive cheats, and MoO 1 expert players set up "house rules" (like not being able to change the destination of retreating fleets) to compensate for mechanics that the AI doesn't use or uses extremely poorly.

Despite everything though, MoO 1's two most difficult AIs are still pretty fun to play against.

I think the clue to making a good AI is to not make the game too complex with too many options, at least on the strategic level.

I think the issue is more that some problems are difficult for an AI to solve while some are easy. If I were to design a 4X game to have the best AI possible, I'd just make it the most convoluted and complex math puzzle I could think up, since computers are far better at math than we are. The more complex, the stronger the AI. That would be a rather terrible 4X game though. On the other hand, a plethora of diplomatic options tends to create a very difficult problem for a computer to solve (especially when they involve things like declaring war on a third party).

Ironically, poorly designed 4X games that are little more than math puzzles and rote response, which is where computers should logically excel, in my experience tend to have the most disappointing AIs.

And for single player games I think it may be better to intentionally make the factions not balanced. This creates a better dynamic and less predictability.

I think factions should be unique, certainly, but imbalance itself is not a virtue. Having interesting factions is more important than having perfectly balanced factions though.

The end problem with procedurally generated galaxies has to do with size. At some point, there will be too many stars & civilizations for your computer to handle.

I've worked quite a bit on memory management in Remnants and plan to allow the creation of galaxies with up to 10,000 systems. While I understand that 10K is not galactic size, it's still far more vast than any other space 4X game and frankly unrealistic to expect anyone to complete. It's there for me to do performance and memory tuning (which no one seems to do any more).

Keep in mind that a true AI has its own view of those 10,000 systems, so 20 AI races means 20 views x 10K stars. You can see how quickly it multiplies if you don't optimize.

One idea we're playing with for Dominus Galaxia is to have the game take place on two scales. So we might let the player interact with ~100 stars on the strategic map, but also simulate a far larger galaxy in the background with a realistic number of stars in an abstract fashion that the player can only interact with abstractly.

I mean, the unfortunate reality is that the best way to get good AI is to continually update it over the game's lifecycle to reflect shifting metagame and strategy and the gap between how the game is expected to play and how it actually plays, which involves a lot of unpaid hours. Patches go so far, but obviously there's usually so much they need to cover in terms of bug and balance fixes that the amount of time that can be devoted to improving the AI (on top of the amount of time required to teach the AI new or fixed features) is not going to be a huge amount.

Definitely. We're going to do this for DG though. In fact, we're already doing this as much as possible in the beta!

I think it's very possible to develop a 4X AI that will absolutely crush human players. Given that, it's then possible to tone that down and create competitive AIs that have variation based on personality. Besides 90% of personality is just dialogue, not action, so a good writer will do wonders for that.

Absolutely. Presentation is just as, and possibly more important to conveying an AI personality as is any actual decision making.

As per the first point in the original article - you cannot write the AI if the mechanics keep changing

Actually, for Dominus Galaxia, we've been doing exactly that. But the reason mechanics are changing isn't because we have no vision or don't know what we want to do; they're changing because after teaching the AI how to competently use a mechanic we gain insight on how to improve it; or sometimes we come to the realisation that the mechanic sucks and needs to go away. Playing against competent opponents is a the best way to benchmark 4X design. It's easy to create systems that look good on paper, but the real world is too often a much different place.

To repeat myself: AI requires good programmers with solid knowledge of the game mechanics, lots of testing and iterative changes. Studios prefer to push AI as a technological obstacle, when it is very much a $$$ / project management issue. True, it can be a symptom of crappy or lazy programmers, but if the rest of the game is functional and polished, odds are the problem lies elsewhere.

Yes, but you got the priority backwards. A good AI (for a complex system like a 4X) requires an expert player to create it. They need to be able to program too of course, but actual programming skill isn't nearly as important, and AI logic usually doesn't need to be very elegant to be functional (often, you're left with a huge number of conditional statements). This is why you have all these amateurs making great AI mods that blow official AIs out of the water.

That isn't to say that programming skill isn't important at all. Processing time, readability, expandability, stability, etc. will all benefit from programmer skill.

One issue is that for whatever reason, most game developers are not skilled game players - not even at their own games. Anyone who has played competitive games where the developers openly play as well can likely attest to this generalisation, and competitive games should logically be selecting for developers who are more skilled as game players to boot. I don't understand in the slightest why this seems to be the case, but that's life.

I'd say offering a better AI would make for a better return on investment if that was mated to a decent game as a whole. About the only 4x game with a AI approaching decent in my experience was GalCiv2

GalCiv 2's AI is actually really bad in my own experience, but does a good job of looking like it knows what it's doing when you're a novice. There are so many mechanics that the AI is hopeless at, yet players can exploit to massive effect. When you know what you're doing, you can beat the hardest AI (which also gets many, many cheats) pretty much every time without breaking a sweat.

 

[Ludovic]

Yes, but you got the priority backwards. A good AI (for a complex system like a 4X) requires an expert player to create it. They need to be able to program too of course, but actual programming skill isn't nearly as important, and AI logic usually doesn't need to be very elegant to be functional (often, you're left with a huge number of conditional statements). This is why you have all these amateurs making great AI mods that blow official AIs out of the water.

That isn't to say that programming skill isn't important at all. Processing time, readability, expandability, stability, etc. will all benefit from programmer skill.

One issue is that for whatever reason, most game developers are not skilled game players - not even at their own games. Anyone who has played competitive games where the developers openly play as well can likely attest to this generalization, and competitive games should logically be selecting for developers who are more skilled as game players to boot. I don't understand in the slightest why this seems to be the case, but that's life.

True, that is a good point. In my experience testers are usually a lot better at the game than the developers, at least for large productions. At large studios, the programmers are often not particularly interested in the kind of game they're working on, and if they are, I've heard a lot of colleagues talk about how programming a game makes it very hard to enjoy playing it (and I do somewhat feel this myself, at least for some types of projects). Same goes for level designers and some artists. I think for smaller projects there is a much greater sense of ownership and a higher chance that the programmers are working on a game they actually enjoy playing.

It's also true that programmers are generally not skilled players. I interviewed for position at Riot Games once and they seemed a bit surprised at my decently high LoL ranking, since the majority of the programmers were at the other end of the ranking scale. So I do think you're right about that point. If I had to make a guess as to why, I think programmers are (on average) not very competitive individuals. There's a host of other possible explanations, but that would be quite a detour from the subject at hand.

I do still stand by my original point: the lack of good AI is an issue of prioritization, and studios tend to lie about the technical challenges involved. Also, like you mention with your project, AI can be refined through iterative development, and pairing up a good programmer who is bad at the game with a tester who is great at the game, should be able to produce incredibly strong AI. A trait of good programmers is that they are capable of leveraging the domain knowledge of others in their work.

But you are absolutely right that an AI programmer who is a skilled player, or even just passionate about the game they're working on, has a great advantage over someone who is not.

 

[RayF]

If I had to make a guess as to why, I think programmers are (on average) not very competitive individuals. There's a host of other possible explanations, but that would be quite a detour from the subject at hand.

There are games that I've been very competitive at, but Master of Orion was not one of them. I just loved to play that game as a relaxing, immersive experience. That's why I'm remaking it, but even as a casual player it's not hard to avoid the AI mistakes in the original game. That's why I've concluded that poor AI in production games is as much of a scheduling priority issue as anything else. It's also probably why I am interested in making a more immersive AI (i.e. with personalities) rather than a hyper-competitive AI.

 

[Beastro]

It's also probably why I am interested in making a more immersive AI (i.e. with personalities) rather than a hyper-competitive AI.

whynotboth.jpg

 

[RayF]

It's also probably why I am interested in making a more immersive AI (i.e. with personalities) rather than a hyper-competitive AI.

whynotboth.jpg

Because I'm not interested in making a hyper-competitive AI?

Don't get me wrong; I know a lot of people want this. It's just one of a long list of items that people want in the game that I have no interest in spending time to code up. That is exactly why I plan to open-source the code when it's done -- so people can fork off the repo and make whatever additional changes they want.

 

[baturinsky]

A space combat game with dynamic world would be neat.

I'm making one now. But it took me forever to figure how sole pilot can affect entire world of warring factions in meaningful way.
Eventually I came to idea that he should use ideas. Space fighting/trading/questing/exploring is mostly a way to get fame in general, and renown of particular kind. Which fame/renown in turn can be used to sway leaders and population in some way.

 

[baturinsky]

Well, for Crusader Kings 2, it is a lot more than 30+ entities for the AI, as there are hundreds of counties and thousands of characters, all capable of performing individual actions and having to respond to other characters plotting against them, wars, raiders, events, etc.

Individual agents are not thinking much about strategy. They mostly react to immediate situation according to their stats and random rolls. Having some virtual baron amange individual province is not computantially much different with one AI making decision for each of those provinces action. In fact, it can be even less complex, because AI has an excuse to not coordinate thinking of individual provinces with each other much.

 

[GarfunkeL]

Very few 4Xs get combat right anyway. Tactical combat is cool but if it's poorly done, it just gets repetitive and annoying. Good example: the remastered Master of Orion, in which the tactical battles are a disgrace. Not that MoO1 and MoO2 had spectacular combat either. It's probably best to avoid tactical combat altogether and instead focus on the strategic. That certainly would have helped Stellaris. Space Empires has suffered from this same syndrome despite having far more options that MoO - it requires massive fleets to be able to shine and often the strategic situations makes it a moot point. I know I'm always praising Aurora but it has the best space combat - probably because it resembles Harpoon instead of Space Chess.

Crusader Kings in Space is a neat concept. It has been done in literature and the Battletech/Mechwarrior universe utilizes a similar concept. That would work better for a procedural generation and long-duration game where the goal isn't "just" map painting.

But I dunno, the archetypical 4X has basically three different stages that should all be interesting and exciting and hardly any game has even gotten close. You need to have lot of micro things to do at the start and it's cool to get nitty gritty with the details. Then you get to the middle portion where you don't really want to fiddle with asteroid mines because you have bigger fish to fry. Finally you reach the late game portion where your stellar empire is big enough that you only need to focus on the macro level. I think the best 4X games have gotten two of those right but nothing has gotten all three.

What has this got to do with AI? Well, you can't just take away features from the player mid-game (or can you?) but there is the risk that those micro things that are cool details in the beginning become tiresome busywork late game. Introduce friendly AI that handles those things. Make it capable enough that players actually want to to use it.

 

[baturinsky]

Game can simply limit the number of actions/orders player can do/issue per turn. So, he can command small kingdom/army by himself, but has to resort to AI governors/barons/generals later.