-
We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
- Status
The discussion of AI between Johan and the experts of AI is really interesting. People can learn a lot of things regarding AI that people usually misunderstand.
Absolute and complete nonsense
The strategic mode in paradox games is FAR, FAR, fAR< FAR, FAR more complicated than in TW games
I've played almost every paradox game and EVERY TW game and that's a fact
TW games are great though for what they are... the tactical battles rule.. I was very disappointed in the ETW AI however... Johan is right that on release it was totally busted... but if they can patch is the game has a very solid foundation and could really turn out to be a gem
Different target audience? Given the number of people cross-registered on both forums it's hard to tell...(and note the number of comments about the qualities and deficiencies of Paradox AIs on the other forum)
By the way, HoI2 sold 500,000 copies. Not too shabby for a wargame. Plus, I wouldn't call a project successful just because a large number of people are conned into believing that it's successful only long enough to fork over the cashI openly admit to being very biased here, but the reaction "over there" to Johan's remark doesn't reflect very well on the quality of ETW. I don't mean harm to CA, by the way. I think it's a great concept (well, except for Hellenistic Egyptians looking like they took a time machine trip from the Middle Kingdom...).
Think about it - if someone over at CA said something like this about HoI2, and then someone opened a thread here agreeing with that, they'd be flamed to death (justly or unjustly) and Thracid would have to salvage the situation by closing the thread. But "over there" the remark got mostly total agreement for the first few pages. And I've been curious about this and did some googling - seems like ETW's AI really is pretty bad...
I'm guessing that those 500,000+ copies of HoI2 were INSANELY profitable for paradox given their modest development and marketing budgets.
I can tell you the AI in vanilla ETW was TERRIBLE, but it has been updated, and is considerably better than it was. Granted it still has tendency's to do stupid wacky stuff, but now they actually build large, well composed army's and fleets.
Comparing the AI in Supreme commander to TW or HOI doesent make alot of sense. Its a Tactics game, not a strategic game. The main thing that wrong with the TW AI is the strategic AI, not the Tactical one. (which isnt all that great either, but Im the kinda of player who fights 3 tactical battles then gets bored and lets my generals handle it
I was using Supreme Commander as an example of a real-time game having thousands of entities on very large maps and nevertheless managing to have a decent AI. The difference between the game types concerning AI isn't great because even in strategy games it usually does very little planning.
I have to disagree. Having free movement in a Real time game probably wouldnt be possible with our current technology. I think the only reason it works for TW is that its a turn based game. I find it amazing that Paradox games can have an AI as good as it is given that its all happening in real time.(although It could always be Improved
Try and imagine HOI or EU (or whatever paradox games you have played) with a free movement map with 1000's of divisions moving around and the AI making on the fly calculations... I just dont see it happening.
I didn't mean to say that Paradox should switch to a free-form map - in fact quite the opposite as the gameplay differences are small at best and the cost is pretty high (especially performance-wise).
Nevertheless, I disagree with your assessment. If I may use Supreme Commander as an example again because it's the only game of that type that I recently played: it has thousands of units on a big map and the AI does on-the-fly calculations and even simulates projectiles. And it seems to run fine while being fairly CPU hungry; however I think the game was released at a time before multi-core CPUs were widely spread so if you make a properly thread-safe game I think it'd be possible.
Sorry.
Just was irritated after paying for the game at releasedate and basically winning every single battle first try.
Then having the AI being unable to build armies that were challenging (even if you had no clue about the gamemechanics, which the manual did not explain.)
Basically I was able to knock out denmark in 1 year as Sweden by just marching the troops i had at start to christiania and kobenhavn.
It felt like "build something random, and move armies randomly at weak positions". The AI in eu1 was a bigger challenge and that one could not handle more than 2 targets per country.
We've been slaughtered for bad ai in games as well, but I'd expect that if you got a budget over 1M USD you'd be able to have at least 1 guy at AI for a year
Just have in mind that you are talking to PR person that is stating how things you are mentioning above is actually something that is "working as envisioned "(his own words):rofl:.It is really saddening that posts like Kieran's can be found anywhere on internet at all, while they would heavily censor(delete) something like this on their own forum.
As for ETW itself, game has potential to be great. However ETW AI is , even now, roughly,still in stone age when comes to comparing it to most of the Paradox games, let alone to HOI series.From what we know, CA had chosen to focus on "candy" looks rather then on competitive AI , because that is what is selling games for them(after ETW not to me anymore though).
Having played games from both companies (and i own every single one TW or PI game), i have to fully agree with your previous statements about non-existing AI in ETW , simply because there is no such a thing as AI in that game right now.
To be fair, HoI1 AI pre-patch was horrid. As I mentioned before elsewhere I was able to steam roll the Soviet Union as Turkey with a handful of cav divisions because the AI was too dumb to react to a Cacaus invasion while dealing with Germany.
But to be really fair, you could not do this after a patch or two.
Still... There was the problem with the HOI2 AI creating D-Day invasion (including EU3 pre-nappy).
And Rome's barbarians are just that... Braindead barbarians who ping pong provinces
Well... You wanted some criticism. Its not like Pdox programs are always perfect.
That's nothing.
I expect that, even after all this time, my original HoI1 AAR still sets the bottom for AI behaviour in any Paradox game ever: Chopsticks in the Mist. Bah, it looks like my screenshots have disappeared?? I'll have to check my ISP about that.Upon release, that game just wasn't programmed to deal with people in unaligned countries running their own little wars instead of fighting WW2. (Not that I can really complain about the focus of a WW2 game being on actually fighting WW2, as that is what the majority of the players will be trying to do, but this was taking it to extremes). Let's just say that HoI2s AI most definitely did take such silly actions into account..
I've given a lot of thought to the differences between Total War games and Paradox games.
I think that the fundamental difference in AI is because of the underlying game. CA has in my experience bitten off far more than they can chew with the last three games. Their design focused on the idea of adding new features each time. There is a big problem with this. If your computer game is going to primarily focus on player vs AI gameplay, ever bit of complexity you add, every choice that the player is enabled to make, must be compensated by additional work on the AI. If the player is allowed to make X decision, for the AI to compete, it also needs an effective algorithm to make X decision in the best way. With the advent of Rome Total War, the decisions that can me made grew exponentially. The map became an enormous grid of an insanely large number of possible battlefield combinations, and an insanely large number of possible paths. This was put forward as a new and exciting feature, and brought in a lot of sales for CA. However, CA could not exponentially increase their investment in AI to compensate.
In Medieval Total War (the first one) the map is a risk-style board of regions that connect together. Each of these connections in each direction has a terrain setup. It was much, much easier to create an AI that was effective in its decision-making in such a scenario. Thus, the AI in MTW can be made to be much more challenging and seem much more intelligent than that of RTW, even if RTW had 10 times as much effort put into the AI.
This is in my opinion a big part of the reason that Paradox Games have better AI -- the fundamentals of the game design are well thought out with the AI in mind. Instead of creating new features to advertise to people who have not yet played the game, Paradox has wisely expanded their complexity in ways that the AI can still be effective. Without access to the source code, I naturally can't know how their actual AI coding compares, though I have a strong hunch (influenced by knowledge of C++ and use of data files, and extensive examination of the data files for both TW and Pdox games) that the Paradox programmers are much more efficient at programming their AI and making maintain-able code.
So, the argument is often made that CA didn't do a worse job, their game was so much harder to make, &c &c. The fact is, CA did not plan their games well with this in mind, either knowingly, considering the money they make off of new features, or unknowingly. Either way, I conclude that they did a worse job at planning their game from a gamer's perspective (again, from a business perspective, they make a lot more money, and this is very much because of their constant expansion of the features, since the players don't know how the AI will handle them until they have actually bought the game). Paradox on the other hand from my observation gives a lot of intelligent thought to the fundamental design from a gamer's perspective. Which is why we don't have a lot of ambitious cash-cow products from Paradox that disappoint everyone.
And it shows, consider the vast difference between the spirit of the Total War forums, totalwar.com for instance where criticism is rampant but censored, and twcenter.net where the criticism is also rampant but there is much more freedom to discuss it. It is clear that a large percentage of forum-goers who purchased ETW were very disappointed. Look at the Paradox forum, and unless 90% of rants are instantly deleted without anyone knowing, it is clear that the forum fanbase is very pleased with the job the developers have done, and are not disappointed -- with the exception of EU:Rome which I think most of us agree was not representative of Paradox typical work. Rome is a negative for PI, no way around it.
I think that the fundamental difference in AI is because of the underlying game. CA has in my experience bitten off far more than they can chew with the last three games. Their design focused on the idea of adding new features each time. There is a big problem with this. If your computer game is going to primarily focus on player vs AI gameplay, ever bit of complexity you add, every choice that the player is enabled to make, must be compensated by additional work on the AI. If the player is allowed to make X decision, for the AI to compete, it also needs an effective algorithm to make X decision in the best way. With the advent of Rome Total War, the decisions that can me made grew exponentially. The map became an enormous grid of an insanely large number of possible battlefield combinations, and an insanely large number of possible paths. This was put forward as a new and exciting feature, and brought in a lot of sales for CA. However, CA could not exponentially increase their investment in AI to compensate.
In Medieval Total War (the first one) the map is a risk-style board of regions that connect together. Each of these connections in each direction has a terrain setup. It was much, much easier to create an AI that was effective in its decision-making in such a scenario. Thus, the AI in MTW can be made to be much more challenging and seem much more intelligent than that of RTW, even if RTW had 10 times as much effort put into the AI.
This is in my opinion a big part of the reason that Paradox Games have better AI -- the fundamentals of the game design are well thought out with the AI in mind. Instead of creating new features to advertise to people who have not yet played the game, Paradox has wisely expanded their complexity in ways that the AI can still be effective. Without access to the source code, I naturally can't know how their actual AI coding compares, though I have a strong hunch (influenced by knowledge of C++ and use of data files, and extensive examination of the data files for both TW and Pdox games) that the Paradox programmers are much more efficient at programming their AI and making maintain-able code.
So, the argument is often made that CA didn't do a worse job, their game was so much harder to make, &c &c. The fact is, CA did not plan their games well with this in mind, either knowingly, considering the money they make off of new features, or unknowingly. Either way, I conclude that they did a worse job at planning their game from a gamer's perspective (again, from a business perspective, they make a lot more money, and this is very much because of their constant expansion of the features, since the players don't know how the AI will handle them until they have actually bought the game). Paradox on the other hand from my observation gives a lot of intelligent thought to the fundamental design from a gamer's perspective. Which is why we don't have a lot of ambitious cash-cow products from Paradox that disappoint everyone.
And it shows, consider the vast difference between the spirit of the Total War forums, totalwar.com for instance where criticism is rampant but censored, and twcenter.net where the criticism is also rampant but there is much more freedom to discuss it. It is clear that a large percentage of forum-goers who purchased ETW were very disappointed. Look at the Paradox forum, and unless 90% of rants are instantly deleted without anyone knowing, it is clear that the forum fanbase is very pleased with the job the developers have done, and are not disappointed -- with the exception of EU:Rome which I think most of us agree was not representative of Paradox typical work. Rome is a negative for PI, no way around it.
Not with Vae Victis it isn't
Although the original game was a bit too much like the original EU to suit anyone's expectations for complexity.
Yes I agree with you about this.
Small example the new regional system in Empires has one perhaps two glaring flaws, if you take the main settlement of a region the ownership of that regions resources also changes, so if the British take Paris then Marseilles also becomes British.
Now explain that to the French AI, one minute its guarding its own Mediterranean port and the next minute its occupying a British port. I suppose as far as the AI is concerned its doing very well but it must be as confused as hell.
I think this is just crazy and an example of not thinking things out , it creates huge problems for the AI so much so that it may as well not turn up for the fight.
Which may in fact be the proof that there is an AI but its not going to play with these rules because they're daft.
@JOHAN. Please sign up to TWcenter.net. We love you on that site. I love you, just look at this thread your awesome CA ownage has generated:
http://www.twcenter.net/forums/showthread.php?t=268065
The majority of us agree with you and we are CA fanboys.
You sir are the reason i buy paradox games.
http://www.twcenter.net/forums/showthread.php?t=268065
The majority of us agree with you and we are CA fanboys.
You sir are the reason i buy paradox games.
@JOHAN. Please sign up to TWcenter.net. We love you on that site. I love you, just look at this thread your awesome CA ownage has generated:
http://www.twcenter.net/forums/showthread.php?t=268065
The majority of us agree with you and we are CA fanboys.
You sir are the reason i buy paradox games.
Just for the record: I don't love you Johan, you're a fine chap but I'm not into men :rofl:
By the way: Will you (or someone else for that matter) have the time anytime to write about AI in HOI3? I'd be interested
I think that a more expansive DD concerning HoI3's AI (we had stuff on it before, but it wasn't specifically dedicated to STRATEGIC AI, i.e., the other "players") would be in order after this interesting incident You know, as a coup de grace
You know, as a coup de grace 
Cars on the contraflow system.
Read through the TW thread and clear message is that CA have alienated their customers to a quite scary degree. As Paradox continues to power along, all power to their collective elbows, as CA once did, be worth their while to keep an eye on the wing mirror at the CA car crash behind.
I can remember being excited in anticipation to new TW games/add-ons, much (through, truthfully less) as I'm looking forward to HOI3. Now, I see the TW game as trash (have a vista 64 box so the game simply didn't work for over a month after getting it). Once it did work, I had one game then left it. The empty boast of "see my mighty works and tremble", comes to mind, I sincerely hope HOI3 is the wonderful game it promises to be and makes PI a shedload of money, I also hope that their culture can be retained, even if they have a 1m buget for Vicky II
Read through the TW thread and clear message is that CA have alienated their customers to a quite scary degree. As Paradox continues to power along, all power to their collective elbows, as CA once did, be worth their while to keep an eye on the wing mirror at the CA car crash behind.
I can remember being excited in anticipation to new TW games/add-ons, much (through, truthfully less) as I'm looking forward to HOI3. Now, I see the TW game as trash (have a vista 64 box so the game simply didn't work for over a month after getting it). Once it did work, I had one game then left it. The empty boast of "see my mighty works and tremble", comes to mind, I sincerely hope HOI3 is the wonderful game it promises to be and makes PI a shedload of money, I also hope that their culture can be retained, even if they have a 1m buget for Vicky II
Not even in HoI the IA was permanently suicidal.
In Empire, there are incredibly funny moments, like having a port uncovered and a little Italian faction declares war just to block it while you have a big, seasoned army right in front of their border. IA is completely unable to "think" in long terms. And even worse, it never sues for peace even when there´s no way for the IA to win not just a war, but even a single battle against you.
On the other hand, I have played both HoI´s and a country NEVER declares war on you unless it has some chances to win. It would be funny to see a CA IWWII game and see how Haiti declares war on the USA just because Florida is not garrisoned.
In Empire, there are incredibly funny moments, like having a port uncovered and a little Italian faction declares war just to block it while you have a big, seasoned army right in front of their border. IA is completely unable to "think" in long terms. And even worse, it never sues for peace even when there´s no way for the IA to win not just a war, but even a single battle against you.
On the other hand, I have played both HoI´s and a country NEVER declares war on you unless it has some chances to win. It would be funny to see a CA IWWII game and see how Haiti declares war on the USA just because Florida is not garrisoned.
Last edited:
funny thing about RTS games, in my experience the smartest AI is actually the easiest to win while the best or better said most challenging "AI" is a scripted campaign "AI" or a product of good game design.
And while a good AI its nice to have, for me a good RTS game was never about good AI and was all about rich tactical multiplayer (ppl are the best AI i know)
i still dont get why ppl keep comparing ETW AI to HOI AI, a comparison to EU3 will be a way better fit (or if its still AI wise so do WOW or any other game)
Last edited:
OK; EU3 then since I have it, played it and now is resting somewhere (I do not have In Nomine or Napoleon´s Ambition, so my oppinion is closer to how the game was in the first stages just as ETW is now).
I played with all the big playes of the board: England/GB, France, Spain, China... Well, Tuscany NEVER declared a war on me, even when relations with them were at -200 unless they had a plethora of allies and even in that case, phaving a strong army prevented them from declaring wars that would obliterate them in less than a year.
In ETW I´m fed up of rampaging in Italy just because all those minors declare wars on me because I have an uncovered comercial port.
Another example? Let´s say that a German state declares war (just because again) and I place a huge, veteran army that pts their only capital under siege. How the hell can the AI think that their militias and armed citizens can compete with a 20 unit army in which the worst one is a line infantry unit with two levels of experience?
@JOHAN. Please sign up to TWcenter.net. We love you on that site. I love you, just look at this thread your awesome CA ownage has generated:
http://www.twcenter.net/forums/showthread.php?t=268065
The majority of us agree with you and we are CA fanboys.
You sir are the reason i buy paradox games.
Man that thread is 18 pages at the moment. There is some built up frustration over there it seems
I have played M:TW campaign mostly as that is the things I like most in such games. It was a bit sad that the campaign part was the worst with M:TW. Nice graphics though, the AI... well the most is already said about that.
- Status