Home Trending Latest threads New posts Developer diaries

HoI4 Dev Teasers (previously Podcat's Twitter Teasers)

Jump to latest Follow Reply
  • 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.
Hearts of Iron IV 45536 Bug Reports 15231 After Action Reports (AAR)
Show only dev responses
Reader mode
It's worth noting that the AI HAS substantially improved with Waking the Tiger. Before, it would often leave entire fronts abandoned at random. I haven't seen this happen in nearly a year now. Sadly, the bar for improvement is pretty darn low.

I'm currently play a multiplayer game with Alex Brunius, Secret Master, Dalwin, and some others and in our current game, I've taken control of Poland when Germany begins "Danzig or War", and France following the capitulation of Poland.

In 1939, the Polish AI was producing ONLY infantry equipment and support equipment. Not only no AT, but not even ANY artillery. It was also still on volunteer only and civilian economy. It had not even trained one single division. I was stuck with the starting OOB plus 19 divisions I managed to just barely squeeze out in the 70 days I had.

When I switched to France, they were also on civilian economy and limited conscription(I believe they start with it).

I wish that they would indeed take the development of the AI a bit more seriously. I don't think it necessarily needs a huge amount of man hours dedicated to it. Some semi-scripted moves RE production and mobilization laws, and country-specific strategies would go a long way.
 
Iskulya said:
It's worth noting that the AI HAS substantially improved with Waking the Tiger. Before, it would often leave entire fronts abandoned at random. I haven't seen this happen in nearly a year now. Sadly, the bar for improvement is pretty darn low.

I'm currently play a multiplayer game with Alex Brunius, Secret Master, Dalwin, and some others and in our current game, I've taken control of Poland when Germany begins "Danzig or War", and France following the capitulation of Poland.

In 1939, the Polish AI was producing ONLY infantry equipment and support equipment. Not only no AT, but not even ANY artillery. It was also still on volunteer only and civilian economy. It had not even trained one single division. I was stuck with the starting OOB plus 19 divisions I managed to just barely squeeze out in the 70 days I had.

When I switched to France, they were also on civilian economy and limited conscription(I believe they start with it).

I wish that they would indeed take the development of the AI a bit more seriously. I don't think it necessarily needs a huge amount of man hours dedicated to it. Some semi-scripted moves RE production and mobilization laws, and country-specific strategies would go a long way.
Click to expand...
This is very sad if this is what the AI does. No wonder I can crush Poland in like 8-12 days. Historically it took about 37 days and with the USSR also helping out.
 
Iskulya said:
It's worth noting that the AI HAS substantially improved with Waking the Tiger. Before, it would often leave entire fronts abandoned at random. I haven't seen this happen in nearly a year now. Sadly, the bar for improvement is pretty darn low.

I'm currently play a multiplayer game with Alex Brunius, Secret Master, Dalwin, and some others and in our current game, I've taken control of Poland when Germany begins "Danzig or War", and France following the capitulation of Poland.

In 1939, the Polish AI was producing ONLY infantry equipment and support equipment. Not only no AT, but not even ANY artillery. It was also still on volunteer only and civilian economy. It had not even trained one single division. I was stuck with the starting OOB plus 19 divisions I managed to just barely squeeze out in the 70 days I had.

When I switched to France, they were also on civilian economy and limited conscription(I believe they start with it).

I wish that they would indeed take the development of the AI a bit more seriously. I don't think it necessarily needs a huge amount of man hours dedicated to it. Some semi-scripted moves RE production and mobilization laws, and country-specific strategies would go a long way.
Click to expand...
It is impressive what modders did with Expert AI, it goes a long way to solve these issues.
I'm trying MtG with it, hopefully it won't crash.

But that only gets you so far.
Things that can't be modded, such as frontline behavior, are still lackluster.
That's really where the major effort should go into imo, or at least to make it moddable as well so that the community can have a hand at it.
 
Vohen said:
It is impressive what modders did with Expert AI, it goes a long way to solve these issues.
I'm trying MtG with it, hopefully it won't crash.

But that only gets you so far.
Things that can't be modded, such as frontline behavior, are still lackluster.
That's really where the major effort should go into imo, or at least to make it moddable as well so that the community can have a hand at it.
Click to expand...
If they made the AI fully moddable we might have solved all of the AI issues already.
 
Vohen said:
It is impressive what modders did with Expert AI, it goes a long way to solve these issues.
I'm trying MtG with it, hopefully it won't crash.

But that only gets you so far.
Things that can't be modded, such as frontline behavior, are still lackluster.
That's really where the major effort should go into imo, or at least to make it moddable as well so that the community can have a hand at it.
Click to expand...

Doesn't Expert AI give the AI countries a bunch of cheats?

I don't want the AI to conjure equipment out of thin air. I want it to stop being stupid. There's a big difference here.
 
Iskulya said:
Doesn't Expert AI give the AI countries a bunch of cheats?

I don't want the AI to conjure equipment out of thin air. I want it to stop being stupid. There's a big difference here.
Click to expand...

No, Expert AI does not give any cheats on the default settings. No free equipment or construction or anything like that. There is one exception: the AI takes 50% less occupation resistance. This is because the AI is horrible at handling this and it can't be modded in a more direct manner. The AI is so bad at managing everything else that this one free thing doesn't even begin to quality as a significant cheat, it's just another band-aid on top of everything else the mod does try to nudge the AI in the right direction.

There is an extensive options menu where you can enable or disable additional options globally or for specific countries. All Expert AI does is what you were saying in an earlier post, better scripting of political power spending and focus choices and research, and doctrines. It doesn't even change the battle plan AI at all. There are dozens of pre-made research +production+focuses AI templates he's created both for generic countries and for specific countries and there are a few reasonable ones he's chosen as defaults for each important country.

IMO many of the detractors of this mod don't understand what it does, or are disappointed when they tried it once and didn't see a perfect replication of a competitive multiplayer game being played out. It *is* better than vanilla, and it is still vanilla other than how the AI plays. (Unless you enable some of the other optional settings maybe) However, it will still not really provide much of a challenge if you are playing a big major like Germany or Soviets unless you enable some of those additional options for the countries you are fighting against, and past a certain point it will feel like the AI just winning because of crazy buffs. (Using the default AI sliders provided in vanilla on top of Expert AI is not recommended by the mod's creator, as these are just OP buffs that don't really feel fun to play against)

It is objectively not difficult for them to script in some basic gameplay decisions into the AI that have long been established as good practice and these are largely the way things are intended to be, I know Paradox doesn't like 40 width divisions but it's not like taking Silent Workhorse first or escalating your Economy Law as soon as possible are metagamey things that paradox didn't intend for people to do. So why does the AI seem to lack basic strategy in a grand strategy game?
 
Last edited:
Vohen said:
I understand that the AI has to cope with a lot of new stuff, but if every time we get new mechanics the AI is neglected, things aren't going to look good in the long run.
Click to expand...

I agree with you, but one thing is adding new mechanics like the Governments in Exile, License Production or similar stuff and another completely different is adding something like fuel while at the same time you are redoing from scratch the naval system.

The changes in Man the Guns are of similar size to the ones introduced in Stellaris since version 2.0 So I will happy as long we don't get a broken toy such as the Stellaris fans did with Megacorp.

I do certainly hope that is not the case since Podcat and the team has taken their time to do things right.
 
Ethereld said:
I agree with you, but one thing is adding new mechanics like the Governments in Exile, License Production or similar stuff and another completely different is adding something like fuel while at the same time you are redoing from scratch the naval system.

The changes in Man the Guns are of similar size to the ones introduced in Stellaris since version 2.0 So I will happy as long we don't get a broken toy such as the Stellaris fans did with Megacorp.

I do certainly hope that is not the case since Podcat and the team has taken their time to do things right.
Click to expand...
Well, I feel like it'll set a bad precedent if that's the case.
This will certainly not be the last big mechanic to be added/changed in the game.
Espionage, peace conferences, a more involved logistics system and occupation (all that just from the roadmap for the near future) might perhaps be just as demanding, and if that precedent is set, the future will look quite bleak for the AI.

The gradual, "constant struggle" approach they like to talk about requires discipline or it might just lead to stagnation.
I definitely love big expansions like MtG and WtT much more than smaller ones like DoD even when they take almost double the time to make, but if the AI is going to be the one getting the short straw, I'd much rather have it be big/small/big/small releases.
 
Watched the feedbackgaming series on MtG just now and I have to say, some stuff regarding the AI did please me.
Germany successfully navally invaded Norway, Japan and China were stalemated up until 1944 and South America immediately joined the allies all at once as soon as the US attacked one of them, opening a very large theater of WW2.

Couldn't see anything on frontline behavior yet, and probably won't until more in depth videos come out and I get the patch in my own hands, but seeing those things already made me quite happy, so praise where praise is due.
 
FYI from the rddt ama:

"PDX_taytay
•
Feb 21, 2019, 7:11 PM
At the moment we don't have an exclusive AI programmer but all our programmers do AI works. Generally in the form of implementing AI for the features they implement. We are doing improvements and fixes in existing AI code as well, and improving the overall front line & unit assignment AI is one of our long term goals."

I don't know how you would make serious AI improvements when you only have a group of people working on it part time or just to implement the features they are assigned. I expect a general decrease in the effectiveness of the AI since it now has to actually do something with naval construction, research, and recruitment. The land and air forces will have less resources (as they historically should), but the excessive land and air resources that were masking some bad AI behavior will be removed.
 
Vohen said:
Watched the feedbackgaming series on MtG just now and I have to say, some stuff regarding the AI did please me.
Germany successfully navally invaded Norway, Japan and China were stalemated up until 1944 and South America immediately joined the allies all at once as soon as the US attacked one of them, opening a very large theater of WW2.
Click to expand...
And he noted that performance didn't take as much of a hit as earlier when Barbarossa starts, so apparently the game will run more smoothly too.
 
Vohen said:
Watched the feedbackgaming series on MtG just now and I have to say, some stuff regarding the AI did please me.
Germany successfully navally invaded Norway, Japan and China were stalemated up until 1944 and South America immediately joined the allies all at once as soon as the US attacked one of them, opening a very large theater of WW2.

Couldn't see anything on frontline behavior yet, and probably won't until more in depth videos come out and I get the patch in my own hands, but seeing those things already made me quite happy, so praise where praise is due.
Click to expand...
His stream is hilarious: "who is that" "whats that" like the most hilarious and least knowledgeable streamer there is.
 
Reader mode
You must log in or register to reply here.
Top Bottom
Back