Owb would require licensing out the fallout brand from Microsoft now. Seems unlikely.
- 1
Does the DLC Model Mean HOI4 Will Get Worse Over Time?
- Thread starter Faeelin
- Start date
-
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.
Honestly, I don't think the AI is as bad as it once was (yes, looking at you, USSR. I know you want to move all your divisions to Manchuria. Thx for not doing that anymore).
Naval Invasions + Coast defense is pretty good, for AI.
The only real 3 issues I see, and I hope the Devs will at least improve some of it this DLC (I dunno if you read this podcat or can answer * wink wink *) are the following:
Tanks. The AI should create dedicated tank armies, in order to at least in theory be able to concentrate assaults.
Supply. Currently, the AI has no clue on where to deploy (especially when not fighting on core territory like greece, UK and US etc.). They sometimes yeet around 30-50 divisions into norway or pacific islands. They should realize that it wont work. Nobody benefits of 5 more divisions in this theatre, so fuck off Belgium. Or maybe even give the player the option to say the AI where they should deploy (like in EU4). Example: do NOT stack up on me on Borneo, but DO stack up with my naval invasion army in Normandy.
Commitment. Sometimes they have a really good push or invasion, but they stop bc the numbers don't add up (seemingly?!). But for a player it's quite clear that they could conquer all of france would they only stay and reinforce. The AI should do less offensives and invasions, but try to go "all in", to minimize pointless casualties and replace them with casualties that matter.
I don't really want to personally attack anyone with this, AI is ridiculously difficult to program, but that's what I personally find annoying.
Naval Invasions + Coast defense is pretty good, for AI.
The only real 3 issues I see, and I hope the Devs will at least improve some of it this DLC (I dunno if you read this podcat or can answer * wink wink *) are the following:
Tanks. The AI should create dedicated tank armies, in order to at least in theory be able to concentrate assaults.
Supply. Currently, the AI has no clue on where to deploy (especially when not fighting on core territory like greece, UK and US etc.). They sometimes yeet around 30-50 divisions into norway or pacific islands. They should realize that it wont work. Nobody benefits of 5 more divisions in this theatre, so fuck off Belgium. Or maybe even give the player the option to say the AI where they should deploy (like in EU4). Example: do NOT stack up on me on Borneo, but DO stack up with my naval invasion army in Normandy.
Commitment. Sometimes they have a really good push or invasion, but they stop bc the numbers don't add up (seemingly?!). But for a player it's quite clear that they could conquer all of france would they only stay and reinforce. The AI should do less offensives and invasions, but try to go "all in", to minimize pointless casualties and replace them with casualties that matter.
I don't really want to personally attack anyone with this, AI is ridiculously difficult to program, but that's what I personally find annoying.
- 6
Actually, an official Kaiserreich or some other TC in the form of a standalone game could potentially be really, really sweet.
I know people have suggested giving access to the engine to mod devs to let them make their own Clausewitz game, but as far as I understood things, the devs are skeptical about licensing the engine to third parties, out of fear of it being leaked IIRC.
Maybe this is just an AI problem then, Skylines has a similar model but doesn't really have the issues because its not Vs and you don't need to worry about the AI being any more daft than normal.
At the same time I can understand the AI being Very tricky I mean do you want Paradox to create an AI that is a Military Genius I feel that's a bad road to go down
At the same time I can understand the AI being Very tricky I mean do you want Paradox to create an AI that is a Military Genius I feel that's a bad road to go down
- 1
Great War DLC I think would be great and could be simplified from the mods available online that I think go a bit OTT
- 1
I think it would be good to repeat the idea that they did in the second part of the HOI with the Darkest hour of creating a spin off of the game in a semi-official way that could make the great war and that had some official support as was that project
First off, thank you for replying to this.
Secondly, your quote is basically my point, but more artfully phrased?
The reason the AI is "good" or appears such to players is because the game is confusing. Players don't know what a proper template meta is and there's no tutorial that clearly says what you want. But once you google this the AI doesn't know how to respond and doesn't.
Your posts *sounds* like it's saying we don't want the AI to be that good because players will struggle with a good AI, but as you also note more than half your players play on settings other than easy or very easy.
This is honestly a good example of what players want in the AI though? Human players know not to overflow supply and to build tank units. Consider the game has been out for what, five years? And we don't have those issues fixed. But instead we get yet more mechanics for the AI to deal with (or ot deal with). Thatis the source of frustration here and why I think EU4 is the fate of all Paradox games.
- 6
- 1
“Fix” sounds like there is an easy straight forward solution. There isn’t.
You have to invest a lot of work in inventing a new system and it’s not nearly sure that it will be more enjoyable for the player community.
Why would you be frustrated if a game developer cannot solve ALL difficult tasks, but only some?
- 5
- 1
Because, currently, as a Player France you do not want AI troops in your country. That has been the case for years. You want to see German tank armies, since this is a WW2 game. That has been the case for years.
But instead, we get new gadgets and gizmos that the AI can't use super well, which are then dropped for the next DLC. The fact that we can't get the AI to conduct a Warsaw Uprising is kind of silly, isn't it? That means that Resistance is just a player mechanic to toy with.
Obviously, you can say it's fine to have a WW2 strategy game where the AI doesn't use supply or build tank divisions; I think that is bad, and adding new systems the AI can't use well will make it worse. But I understand if you don't agree.
- 2
- 1
Personally I think the best far future route (I say far future, because I know this isn't easy to do in short order), is to open up the ability to write advanced scripts for the AI and let the game load these. Create a giant function library and put effort into continuously expanding it to provide options. There are legions of modders and hobbyist programmers out there who would gladly try to write bits and pieces. No doubt some big compilation by the community would be developed for each game tailored to many tastes (new, veteran, and niche players alike).
I'd have to bust out the old college textbook on eigenvalues and eigenvectors to create something that didn't run at potato speeds, but I myself would personally dabble in such an endeavor if it were possible.
I'd have to bust out the old college textbook on eigenvalues and eigenvectors to create something that didn't run at potato speeds, but I myself would personally dabble in such an endeavor if it were possible.
- 1
Coding AI isn't easy at all unfortunately.
1) The AI has to be believable. Sure you could make something that can solve every problem thrown at it, but then it's impossible to beat and not fun to play against anymore.
2) It has to be adaptable. Programming the AI would be 100x easier if HoI4 was set up like HoI3 where WWII happened in the exact same way every time. In HoI4 the AI has to account for any possible alternative-history outcome, while also being able to handle completely different game setups with mods.
3) It has to be performance friendly. This is the biggest one, we aren't running Dual Xeon server processors with 64 threads at home. Every AI computation uses CPU power and making a smarter AI involves making more and more calculations.
Using Chess as an example, there are 69,352,859,712,417 outcomes possible. Chess only uses 16 pieces per side with very specific movement and capture rules for each and it's played inside of an 8x8 board. HoI4 is far, FAR more complex and chess engines already take a lot of computational power, and time, to process. Imagine if each AI country took 10-30 seconds to decide what to do after each game day.
edit, forgot to add: So yeah, that's why they just use modifiers between different levels. It's hard enough making an AI that's "good enough" let alone 3+ different ones that all play at different levels of quality.
- 3
- 2
And your definition of "performance"
They used to have external AI scripts in earlier products I think but the speed improvements for going to fully compiled code rather than interpreted was just too useful.
I mentioned this a few times here, I think, if not the reddit, but yes. Take my South Africa game a few months ago. For some dumb reason, the allies thought putting over two dozen units on me, South Africa, in an area where they only have one, low level port. Like....why? Africa is important, but....they were literally throwing away all those units. I stopped that game because Germany started to die and I was kind of hoping they would hold long enough for me to get all of the necessary colonies liberated. Nonetheless, the fact that this happened is sad.
This really captures the problem quite well, in my opinion. And i'll add a fourth condition that lies at the heart of making a good AI for HOI -- it has to produce a "narrative" for the game that looks something like World War II. Here the contrast with EU is instructive. For its historical flavor, EU sets the stage in a certain way and drives toward certain types of developments. But for the most part, the game doesn't crucially depend on any given country always behaving in a certain way. England or Spain or Venice, for that matter, aren't "supposed" to form alliances, always colonize certain places, etc. The time scale is one reason for this difference, obviously. EU takes place over 400 years, HOI over about 10. For that reason the constraints on the nations in HOI are far more severe, because it is far more dependent on particular historical events and not just the characteristics of an historical era.
By the way, in answer to the original question, I have to believe -- on no authority whatsoever, mind you -- that the game is nearing the end of its development cycle. At some point, they have to make the call that it's better to address the game's remaining issues by tearing it down to basics and starting over with a new version.
EDIT: I just reread the posting I am replying to and note that it mentioned the ability of the AI to respond to alt-history paths in various countries' NF trees. I couldn't comment on how difficult that is for creating the AI, but I would note that perhaps in the next version of HOI some thought could be given to making those alt-history paths more of a separate game than they are now. In other words, perhaps HOI5 could be built on a single platform, but then there would be parallel development of historical paths, one meant to have countries behave reasonably close to their historically "scripted" roles and another offering a full range of alt-history possibilities. This could well require two separate content teams, and I have no idea at all whether this is feasible. Probably not, I fear.
Last edited:
- 1
Existing AI performance is lacking. But only from a viewpoint of what I want. I have no idea what PDX goals and restrictions are.And your definition of "performance"
One day, I do plan to work the AI strategy and try to make it "better". I've looked at what is there and it is involved. But one day.
(This is going to sound harsh, but not sure of a better way to word it)
Grand Strategy is a Niche genre that encourages players to have comically high play-times, so for a large portion to not understand how to play the game and needing to play on Very Easy or Easy makes me think something is very very wrong with the learning experience.
- 1
Sorry, one more comment about dlcs in this game. The early dlcs were kind of ok -- better NF trees for the Commonwealth, Axis minors, and so on were ok, but not terribly exciting. But both the MtG and LaR dlc's actually made the game better - a lot better. I've been playing Hearts of Iron since the original, and in all that time naval warfare had never been very good. Ditto espionage and occupation mechanics. So I for one want to give credit where credit is due. I am keeping my fingers crossed that this next one fits into the latter category!
- 4
Yeah. New players losing to the AI doesn't really mean that the AI is super good. It means that the players are bad. The question shifts into why new players are bad, and that is mostly because the learning 'curve' is practically a line going straight up, because there are so many complicated systems, and the tutorial is... lacking.
- 5
It may not be a question of "needing" to play the game on those settings, but instead of wanting to play it with certain outcomes in mind, like conquering the world, for example. I play HOI to study and re-experience the problems faced by particular countries. So i don't care all that much about "winning." That's clearly not everyone's ambition, which is fine.