Will AI cheat economically as in EU3?

[unmerged(255)]

forget the AI, can we at least expect a relatively problem-free and stable MP experience?

 

[minority]

I think the technical difficulties in creating an AI that can function without cheats is the biggest factor behind AI cheats.

I'm not a computer scientist or mathematician, but don't chess AIs require quite some processing power to pose a decent challenge? And this is with a lot of academic work devoted to it already.

I wonder whether the same feat can be achieved for an entirely unique environment and system Paradox creates in Rome without allowing for cheats?

cheers

 

[unmerged(76570)]

So if it aint broke dont fix it? i dont have EU3 but it sounds like it did not reach realism by letting the AI cheat. Brings me back to the good-all AI of the Total War series.

 

[OHgamer]

I think the technical difficulties in creating an AI that can function without cheats is the biggest factor behind AI cheats.

I'm not a computer scientist or mathematician, but don't chess AIs require quite some processing power to pose a decent challenge? And this is with a lot of academic work devoted to it already.

I wonder whether the same feat can be achieved for an entirely unique environment and system Paradox creates in Rome without allowing for cheats?

cheers

It probably could be, but at the cost of having a game with minimum RAM and Processor speed requirements far above the vast majority of gamers' computers, thus requiring players who want to play the title to upgrade their systems - something most game developers are loathe to make necessary since that in effect raises the base cost of the game for the consumer.

Consider EU3 - the decision to have all graphics processed by the graphics card alone so that the main HD RAM could be devoted to number crunching and thus improve gameplay performance resulted in a lot of players not having on their system the minimum required video card (128 MB RAM standalone - a situation made worse by the fact that Intel video cards were designed claiming to be 128 but in fact only have 64 on the video card itself and "borrow" the rest from the main HD to get to the 128, making EU3 unplayable to almost everyone who had Intel cards, especially Intel cards in laptops). In spite of the fact that 128 MB Video Cards have been on the market since IIRC mid-2003, a large number of potential players had to upgrade first before they could play EU3 because their systems had older video cards or cards like Intel that reached 128 only by borrowing from the main system RAM, which may very well have depresses a small % of sales of the title when it first came out since buying the game also meant buying a video card to play it.

So it's a double-edged sword. Developers probably could make the game of the strategy players' dreams - but the system requirements to run such a title with all the number crunching involved to get the AI to begin to have a degree of the flexibility the human brain has would likely be well beyond what most consumers currently have in their computers. Maybe in another say five years, with the costs of RAM continuing to drop like a stone, computer power at a sufficienctly high level to achieve what strategy gamers would love to have the AI do will be widespread among the population at large. But for now it's not, and the negative impact on sales that would result from having a title with very high minimum requirements is a strong reason to not try and push the envelope too far in terms of what AIs might be able to do.

 

[minority]

So if it aint broke dont fix it? i dont have EU3 but it sounds like it did not reach realism by letting the AI cheat. Brings me back to the good-all AI of the Total War series.

I think we ca all agree about the differing complexity levels of both games. While the TW series requires a complex battle AI, the battle system is separate from the grand strategic view, which means the battle AI does not need to run simultaneously as the strategic AI and can use all resources available.

OTOH, the TW series usually has fewer countries, less diplomatic actions and a much simpler economic system overall. But most of all, I believe the number of countries to be a great bearing on how much resources are used.

In any case, OHGamer gives a good explanation above. I personally believe some cheating at this point is inevitable if we are to have the complexity of Paradox's games, but certainly stepped improvements would be welcome.

cheers

 

[Jolt]

OTOH, the TW series usually has fewer countries, less diplomatic actions and a much simpler economic system overall.
cheers

Not to mention that it is by turns, and not a continuous gameplay like EU, where every country has to make daily decision at the same time, thus making thousands of calculations per second. TW has 15 countries, each playing on their own turn, basically all using the very same RAM.

Still, the only alternative to the "cheats" is that the developpers spend another year working on AI improvement, which in turn, like OHGAMER said, means the toll on RAM and Processor speed would be greatly affected, thus raising the minimum requirements, making many players have to spend money improving their computers. As far as I'm aware, Paradox still isn't thouroughly known for being a major player inside the gaming industry, one who "forces" players to upgrade a computer to play their games. That would basically mean a serious fall of sales. Then we have a second problem. Since the developpers would have to spend much more time creating a flawless AI, Paradox would have to spend money on their salaries for those months, which would reflect in an even greater cost of the final product. Either this, or put some cheats to help the CPU fare better against a human, thus reducing costs and the minimum requirements. I suppose it's pretty simple to see which one is the wisest solution.

 

[unmerged(59881)]

I can understand your frustration.

However, here is why I think naval attrition was taken out: You a fleet. You have a number of friendly ports and a large number of sea regions. You want to calculate the shortest path between your current location and the location you want to go to, say a port. You want to calculate an attrition while you move so you can decide to rest in a friendly port.

For a human, this is easy. You watch the attrition of your ships and when it's low, you look at the map to find the closest friendly port. Easy.

For an AI, this is a travelling salesman problem. It is NP complete which means that it is the hardest sort of problem that exist! Graph theory research is still trying to figure out a "cheat" for that one.

For more information, have a look at graph theory, road networks and travellers route choice within road networks for some current research.

Don't talk about frustration, it's not important enough to be seriously frustrated.
The situation just demands a completely different strategy. If the AI can't deal with attrition and therefore it's ship don't have any---fair enough. But then the developers should do something to compensate, e.g. a mechanism that makes the AI bring the ships to the next available harbour after a period of time. The length of this period could depend on the type of ship (galleys earlier) and the tech level (the higher the longer). A little random factor would make it more difficult for the player to predict, when the enemy fleet is going to sail away.

 

[Starforge]

To Peter and OHgamer:

Putting this here though it could be plopped into several of the other forums.

As a long time player of Paradox games - the problem with AI cheating is no real suprise to me. AI and cheating is used in nearly every strategy game put out in the last 20 plus years (and argueably the AI has actually backslid over time.) The lack of improvement in this area is sad but also sadly understandable.

All that being said - I don't personally mind the AI cheating as long as the difficulty level or level of cheating can be controlled by the player. As is - it seems that the trend here at Paradox is to make the games a challenge to the powergamer / multiplayer crowd as opposed to someone wanting a casual SP experience. Making this difficulty adjustable through the difficulty settings should be a priority IMO (but then I'm one of those casual SPers so I'll admit to being a bit prejudice here.)

Add to that things such as the DV release wherein I get to roll the dice on whether or not my computer will CTD frequently or not has made me wait on purchasing all of these games / expansions until I know the stability is there as is the playability (for my style - it's my money after all that we're discussing here.)

In short - if the QA is such that testing hasn't improved nor the mindset that satisfying MP / Powergamers be the goal, I doubt that I'll be putting more money into these games and that's a shame. Beyond the Total War series, I doubt I've put more time into any other strategy games beyond CK / EU2.

Please improve the stability of releases and allow me some control over how badly I'd like my AI to cheat.

 

[Peter Ebbesen]

As a long time player of Paradox games - the problem with AI cheating is no real suprise to me. AI and cheating is used in nearly every strategy game put out in the last 20 plus years (and argueably the AI has actually backslid over time.) The lack of improvement in this area is sad but also sadly understandable.

It is my impression that games AI has considerably improved over the last two decades due to truly impressive developments caused both by the availability of considerably more processing power, general increase in knowledge and methodologies, and considerably larger budgets.

I find it hard to understand the "backsliding" argument when one considers just how few cheats games put out today have compared to "the good old games" that typically cheated left, right, and centre, in ways the player could in many cases not easily detect because access to information on what AI controlled factions did was typically minimal.

In my experience, when ignoring the simple and well described game types like chess and instead focusing on more complex strategy games - games that have typically grown more complex over time too - games' AI development over time has been, while a bumpy road, one of increased prowess over time.

All that being said - I don't personally mind the AI cheating as long as the difficulty level or level of cheating can be controlled by the player. As is - it seems that the trend here at Paradox is to make the games a challenge to the powergamer / multiplayer crowd as opposed to someone wanting a casual SP experience. Making this difficulty adjustable through the difficulty settings should be a priority IMO (but then I'm one of those casual SPers so I'll admit to being a bit prejudice here.)

Now you are voicing (in the second part) a concern that has often been expressed on the Paradox forums (catering to MP and powergamers), and one I find it hard to relate to.

1. Looking back on any Paradox game, you'll see that AI changes and cheats are always based on the singleplayer experience, since the AI has no function except being a minor roadblock in MP.
2. The AI changes and cheats are not generally based on what a hardcore powergamer would need to get a real challenge with a major nation but on the cases of weaknesses brought up and widely supported in the general forum population - the change they get may not be the one they really wish for, but it is their concerns attempted adressed.
3. Multiplayer exploits of the game engine that are found out get fixed with direct impact on singleplayer and the same goes for singleplayer exploits that affect multiplayer. However, it is the case that some of the more truly abusive exploits are either discovered or get major coverage in multiplayer while not becoming generally known in a singleplayer environment due to the increased competition.

To take EU3 as an example, when the casual singleplayer finds that playing a major nation in SP on the highest difficulty level is not at all challenging, when he habitually is leading the tech race despite playing extremely suboptimally and not truly understanding what half of the buildings do, while at the same time utterly dominating in the colonisation as well as conquest spheres of the game, when he never faces a challenge except those he create himself, when he posts about it on the forums and gets the support of the other casual gamers, steps get taken to change that equation. As it should be.

If that means that some players have to play on less than "very hard" to have a fun experience with non-majors - if it means they might even have to consider playing on less than "normal" difficulty, or if it means that some players will be utterly unable to play the most backwards nations well on any difficulty setting, well, what's wrong with that?

Catering to the powergamers/MPers? HAH! The vast majority of changes that are not bugfixing are directed solidly at the core segment of gamers, namely those who are not the best, not the powergamers, not the know-it-alls.

Are all changes equally well balanced? D) Are some, in retrospect, not balanced well at all? No to the former, yes to the latter, but the intended recipients are clear in most cases, and it is the common forum-visiting casual SP player.

Add to that things such as the DV release wherein I get to roll the dice on whether or not my computer will CTD frequently or not has made me wait on purchasing all of these games / expansions until I know the stability is there as is the playability (for my style - it's my money after all that we're discussing here.)

Nobody is well served by unstable software. It is hard to imagine anybody who fundamentally disagrees with you on this one.

 

[Starforge]

It is my impression that games AI has considerably improved over the last two decades due to truly impressive developments caused both by the availability of considerably more processing power, general increase in knowledge and methodologies, and considerably larger budgets.

I find it hard to understand the "backsliding" argument when one considers just how few cheats games put out today have compared to "the good old games" that typically cheated left, right, and centre, in ways the player could in many cases not easily detect because access to information on what AI controlled factions did was typically minimal.

In my experience, when ignoring the simple and well described game types like chess and instead focusing on more complex strategy games - games that have typically grown more complex over time too - games' AI development over time has been, while a bumpy road, one of increased prowess over time.

Actually - I did mean that, but maybe not in the way that you took it. Look at some of the older - simpler - games such as Panzer General. Smaller number of units with set and fixed abilities that you can program AI specifically to use / utilize. As the games have become more complex you can argue that the AI has "advanced" to keep up but in reality - while it may be getting "more complex" it hasn't really gotten better since, with the number of variables, it has become commonplace to merely "cheat" your way through to attempt to "simulate" logical behavior rather than actually try and program actual "thought." Example - don't want to actually keep track of ship movement? Easy - remove AI ship attrition and simply move them around at will. Problem solved (kinda) but ignores the underlying issues of actually having to determine an "appropriate" use for your fleets.

Now you are voicing (in the second part) a concern that has often been expressed on the Paradox forums (catering to MP and powergamers), and one I find it hard to relate to.

1. Looking back on any Paradox game, you'll see that AI changes and cheats are always based on the singleplayer experience, since the AI has no function except being a minor roadblock in MP.
2. The AI changes and cheats are not generally based on what a hardcore powergamer would need to get a real challenge with a major nation but on the cases of weaknesses brought up and widely supported in the general forum population - the change they get may not be the one they really wish for, but it is their concerns attempted adressed.
3. Multiplayer exploits of the game engine that are found out get fixed with direct impact on singleplayer and the same goes for singleplayer exploits that affect multiplayer. However, it is the case that some of the more truly abusive exploits are either discovered or get major coverage in multiplayer while not becoming generally known in a singleplayer environment due to the increased competition.

To take EU3 as an example, when the casual singleplayer finds that playing a major nation in SP on the highest difficulty level is not at all challenging, when he habitually is leading the tech race despite playing extremely suboptimally and not truly understanding what half of the buildings do, while at the same time utterly dominating in the colonisation as well as conquest spheres of the game, when he never faces a challenge except those he create himself, when he posts about it on the forums and gets the support of the other casual gamers, steps get taken to change that equation. As it should be.

If that means that some players have to play on less than "very hard" to have a fun experience with non-majors - if it means they might even have to consider playing on less than "normal" difficulty, or if it means that some players will be utterly unable to play the most backwards nations well on any difficulty setting, well, what's wrong with that?

Catering to the powergamers/MPers? HAH! The vast majority of changes that are not bugfixing are directed solidly at the core segment of gamers, namely those who are not the best, not the powergamers, not the know-it-alls.

Are all changes equally well balanced? D) Are some, in retrospect, not balanced well at all? No to the former, yes to the latter, but the intended recipients are clear in most cases, and it is the common forum-visiting casual SP player.

Sorry, as someone who has spent enough time on the forums since the release of EU3 and have played it to 1.3, I know I'm not alone in voicing my concern here. What I fail to understand - and why I stated my comments in the way that I did - is why some of the difficulty settings are not governable by the player. With the number of people who voice this concern not going away as the difficulty gets arbitrarily upgraded in various patches I truly can't understand (save possibly that it's hardcoded and not easily variabled) why some of the more game-tipping changes aren't controllable (letting countries run with 2x supportable troops / ships (approximation) with no maintenance for example.) Instead of the RP experience you could garner from EU2 - you're left with an "expand/exploit or die" experience with EU3. I'm left with an unhistorical 4x game with an interesting economic model, poor social model and weak combat. My opinions of course.

Now - you can argue that some of the changes in 2.1 will be fixed in 2.2. Possibly - and if so or whenever there is a release that makes it less of a 4x (yes - I use / will use MMG for the RP aspects) I'll pick it up. Until then, I can't really justify the purchase personally and will stick with 1.3 and the version of MM I currently have.

Edit: for the one spelling error I noticed

 

[Peter Ebbesen]

As I don't have time for a long post right now, let me just throw in the following as an aside:

It is fun you should mention Panzer General, a simple but truly wonderful game with an exceedingly poor AI, that was fun to play in SP only because your played fixed setups where the AI so grossly outnumbered you in troops that you literally had to advance your units over heaps of dead enemy units, and, even then, you'd rarely lose a unit yourself so long as you used reinforcements sensibly. If the campaigns had been set up to make the two sides fight on anywhere near even odds, it would have been a complete failure as a game due to the total lack of resistance. (And even with that stacking against you on the hardest difficulty levels, perfect victories were the norm once you learned to play the game)

Today, you complain about modern AIs that need more troops than you can easily maintain to provide a challenge and think it is because of the more options, yet, if the EU3 AI played by those old standards, it should need 5-10 times your troops to even dent your advance (which admittedly is much how EU1+2 played - a near absence of a modern tactical AI )

Compexity breeds more problems for sure, but the general AI in games has advanced leaps and bounds beyond what "adjusting for complexity" might bring.

 

[unorthodoxt]

Why do you say the AI cheats. The AI doesn't open up the console and type in codes. The AI performs how it was coded. That isn't cheating. The AI doesn't giggle and say Im gonna really screw over this guy here and cheat!

 

[Uglyr]

IMO the problem is not that AI cheat. The problem appears when AI cheating obvious. When you make your path to victory based on logic and then lose only because the AI do not conform to the rules.
I remember one of my first negative impressions from eu3 - as Muscowy I was at war with Kazan which had larger army. My plan was to weaken enemy by taking out his provinces (they were unfortified), while he siege mine - he would run out of money and manpower - quite logical plan I think. But no matter I was doing it had no effect on him. I played this war endless number of times but was always losing. Then, one time I managed to completely conquer all his provinces, built army greater than his and just wanted to check what will happen as he moved his army to neighbour country. I waited months but his army wasn't decreasing and he haven't run into debt. I've tried to reload as him - debt in the first month end. I think that was my biggest disappointment from any game I've ever played.

Conclusion: AI cheat. And will do this always. When AI will become as smart as human it will destroy humanity. But until that time AI cheating should not be so obvious. And if it is - it should be officialy declared. I prefer to make my decisions based on the "rules". And I prefer to READ rules rather than investigate them through long testing. Let them be diffrent to human and AI but they should be clear. I must be sure (in Rome) that if I move all my archers to capital AI won't drop A-bomb there just because he need some help to be equal to player.

 

[Starforge]

As I don't have time for a long post right now, let me just throw in the following as an aside:

It is fun you should mention Panzer General, a simple but truly wonderful game with an exceedingly poor AI, that was fun to play in SP only because your played fixed setups where the AI so grossly outnumbered you in troops that you literally had to advance your units over heaps of dead enemy units, and, even then, you'd rarely lose a unit yourself so long as you used reinforcements sensibly. If the campaigns had been set up to make the two sides fight on anywhere near even odds, it would have been a complete failure as a game due to the total lack of resistance. (And even with that stacking against you on the hardest difficulty levels, perfect victories were the norm once you learned to play the game)

Today, you complain about modern AIs that need more troops than you can easily maintain to provide a challenge and think it is because of the more options, yet, if the EU3 AI played by those old standards, it should need 5-10 times your troops to even dent your advance (which admittedly is much how EU1+2 played - a near absence of a modern tactical AI )

Compexity breeds more problems for sure, but the general AI in games has advanced leaps and bounds beyond what "adjusting for complexity" might bring.

ROFL

Ah, me. Boy do I feel like a big mouth bass atm. No wonder why you didn't want chess brought up - simply so you could spend time on a post denegrating whichever game I put up as an example. Well, my mistake for falling into the trap BUT let me make a few points here where what you posted was flat out political BS:

1. Umm..playing a historical game like PG, PG2, etc there were plenty of scenarios wherein you were supposed to get a Decisive victory simply to match what happened in history. If you're trying to argue that when assaulting Poland, Belgium, etc that the AI had you outnumbered - then I'll get you my address so you can ship me some of whatever you're smoking. Yes, if you made it to the UK or US you were outnumbered - but then those weren't "historical" representations but what-ifs.

2. Units took advantage of terrain (not simply got a bonus if you happened to catch them there as is what happens in these games.)

3. The AI would make use of supporting units such as artillery and AA.

4. The AI would specifically attempt to finish off wounded units (hurting you in the long run campaign wise.)

5. The time limit on Decisive victories argued against always getting your units set up for well supported attacks. In many instances you had to push forward at even or worse odds and not wait for support units to arrive so that you could meet your objectives. A couple of turns of bad weather grounding your aircraft or a tough assault could lose you a turn when, in many cases you had no turns to give to get a Decisive victory.

6. I don't expect to convince you since the whole point there was a setup. Congrats on your political victory .


As to the EU3 argument, plenty has been said on those forums. Anyone interested feel free to peruse the posts over there. Since I don't expect a debate from you - merely either denial that the problem exists or denial that it should be considered a problem so I'll leave it at that .

I guess instead of waiting for patches from PI I should be keeping an eye on MMG2 - a mod which wouldn't exist at all if PI hadn't made the game moddable (a good thing) and there weren't enough folks out there like myself who enjoyed a different vision than what we get with vanilla. The only unfortunate thing is that some things (like war capacity) can't be modded :/.

Then again - I don't play R:TW anymore without using the EB mod and I understand some of those folks offered to help PI with Rome. Here's hoping if they can't fix it that some of them will mod it.

 

[Daffius]

... Most of Paradox' recent games have had remarkably few cheats implemented (minor economic help leading countries to achieve the economy that any well-played human country can achieve), and only one blatant one (the ever present naval issue)...

Since the whole arguement started from how much the AI cheats in EU3, this certain sentence caught my eye.

I guess we see it differently, what you see is a minor economical help, and what I see is a game breaker. The minor economical help, you mention is that the AI cannot go bankrupt, this can only be solved if it doesn't go below 0, anytime. This on it's own is a minor detail, but it spirals out of control. It doesn't go below 0 therefore it has to mint, but then inflation skyrockets. Therefore it is implemented, so that they cannot mint above an amount. What do we get? An AI thats spends like crazy with no consequences. Yes it does add a challenge, and people may argue pro and against.... the issue is that this is not what the game was designed for. Think of it, what use is trade, naval blockades, embargoes etc... all those options which paradox implemented for our enjoyment. Nothing!! As they cannot be hit economically.

This is not a rant about EU3, nor is it meant to change your views, I'm just saying that these are major cheats, not minor in any sense. And they would be discovered, no matter if it was in easily readable text files or any other format.

So I guess the original posters thread would rather be, "will the AI cheat in a gamebreaking way in rome?". Or will all the options they implement for our use be actually usable?

 

[unmerged(52731)]

The real issues with AI is that Paradox really doesn't have the manpower or the resources to create an exceptionally challenging AI without delaying the game drastically. Financially it would be foolish.

On top of all of that, the AI cheating isn't game breaking because it makes the game interesting. I like to see the AI do stupid things here and there, but if your going to push it into a spiral of debt because its not fair, then I think you should have to ignore all your instincts and reason, as thats not fair to the ai now is it.

Bottom line: the AI will cheat, but the game will as a whole be better for it.

 

[Daffius]

You get me wrong, I'm not argueing about giving the AI advantages (cheats) or not. The point is that giving cheats that confront the main game concept is something that can be considered game breaking. Thats why I brought up this issue... Not allowing the AI to go in the red gives a large challenge, but then whats the point of a number of game features that they implemented and gave us the opportunity to use. Why should you waste time and energy to blockade ports if there is no effect? They won't go bankrupt.

It's not the fact that cheats exist but rather the type of cheats that have been implemented that can create a game breaking experience (for some of us).

I believe I do understand the restrictions of a small company, and this is not a flame against paradox. (If I didn't like it, I wouldn't be here playing with it and argueing , now would I?). It's just rather a question about giving reasonable cheats, or an option to have them enabled. Cause we have to also consider that people don't play the game on a same level. I've played both HoIs, vicky, ricky, the EUs and I find it fascinating that people reach a level of gameplay that I cannot even imagine. But there are a lot of casual players like myself, who like to "play" out all the options given to them in a game and take our time. After a hard days work having an AI that cheats and hands me my butt on a silver plate whenever I try anything IS a game breaking experience.

I went on quite a while, but I hope you get my point. I know it is a delicate balance, but adding cheats should be well controlled and within the games limits.

 

[Starforge]

The real issues with AI is that Paradox really doesn't have the manpower or the resources to create an exceptionally challenging AI without delaying the game drastically. Financially it would be foolish.

Nothing like expecting mediocraty - I guess you'll never be disappointed.

Thats why I brought up this issue... Not allowing the AI to go in the red gives a large challenge, but then whats the point of a number of game features that they implemented and gave us the opportunity to use. Why should you waste time and energy to blockade ports if there is no effect? They won't go bankrupt.

It works agains opponents in multiplayer - as I'm sure you've been told over on the EU forums you should be doing if you don't like the AI. If you really want a RP experience - this likely isn't going to be the game unless you play MP. SP it plays like a standard 4x game - outgrow the AI before it outgrows you. Mods help (thank God) but, sadly, some of the pieces of the engine can't be modded.

In some ways - it would be better (irrationally perhaps) if this wasn't a map of the world but merely a random map with fantasy names, religions, cultures. Then you could treat the game for what it is - a fantasy 4x blob fest.

 

[Fornadan]

So if it aint broke dont fix it? i dont have EU3 but it sounds like it did not reach realism by letting the AI cheat. Brings me back to the good-all AI of the Total War series.

And which Total War game would that be? I always found their decent tactial AI to be held back by an awful strategic one (granted, I've not played M2TW, maybe they've improved)

 

[unmerged(52731)]

Nothing like expecting mediocraty - I guess you'll never be disappointed.

Perhaps, but truth be told most paradox AI is actually quite good, and is suited for most situations. There are simply a number of other things id rather they focused on. On top of that, expecting perfection will lead to disappointment more often than not

In some ways - it would be better (irrationally perhaps) if this wasn't a map of the world but merely a random map with fantasy names, religions, cultures. Then you could treat the game for what it is - a fantasy 4x blob fest.

Yep, and the sooner you do that the more amusement you will milk out of it. There are few games which really deal with historical periods from a history > gameplay viewpoint, and they all, frankly suck. However, more to the point, there are no games that have an AI that can seriously challenge a human. As you and others have said, only a human can do that