HOI4 Dev Diary - 1.5.2 Update #3 and Telemetry!

[Dan1109]

Maybe PDS should focus on bugs only for 4-8 weeks. Content designers can help balance test. Try to get the state of the game as good as possible before moving on to new content in 1.6. And everyone at PDS is limited to speed 3 play in order to be more aware of the tiny problems often overlooked.

 

[chrnno]

I didn't read the whole thread. But people were asking for examples of a game where the AI got better (not just bonuses) at higher difficulties.

SC2. There are some levels with bonuses, but even without the AI is pretty amazing. I played WC3 at the professional level, so I am a very good RTS player, and I had to significantly try to beat the top SC2 AIs. And before the usual response of how HOI4 is soooo complex... a SC2 AI is harder and more complex to code than something like an HOI4 AI.

Not really, the major issue with Paradox games and especially HoI is being history based. The proper way to code an AI is to build the entire game around it, if there is anything you are struggling to have the AI deal with you simply change/remove until it can but being a WW2 game means there is tons of things you need to include, or at least give a nod to, and thus you can't do that so we get a bunch of workarounds and out of place limitations that introduce a bunch of problems and ensure the player can always run around the AI.

The AI of a game like HoI is simply never going to be as great as a game like SC*. If Paradox settled on precisely how they want the AI to play historically, and how they want it to play optimally it could be at least good; but often it feels like Paradox starts with the idea of what (they think) players want, develop it, try to fit the square peg in round hole in getting the AI to deal with it properly, inevitably fail and add arbitrary limits to help the AI a bit then move on to the next mechanic where they do the exact same thing, instead of figuring what they can make the AI do and then figure out how to present those options to players.
*Given the remotely close levels of time, money, effort and such invested anyway.

That said comparing Blizzard of all companies is kinda pointless because out of all strategy games I have ever played the AI of all fall short of theirs. Blizzard might horribly suck at the story part but damn their gameplay is great.

 

[myzael]

Maybe PDS should focus on bugs only for 4-8 weeks.

But they are focusing on bug ever since March 8. For a whole month they barely released new content, definitely nothing major and all patches are strictly bug-focused.

 

[Sir Garnet]

@podcat Any chance of an misclick undo or reverse-last-operation button for things like army organizational changes, decisions, pop ups, and fresh-created battle plans or changes. This is not just a problem for persons with minor disabilities or worse.

Anyone can misdraw a BP edit and on completion see it was a horrible mistake - the safest course is to pause and start over.

Or fail to shift a few divisions to a new mission correctly and instead merge two armies and derange the entire front or war. Losing the war from a bad decision is different from failing by being a bad mouse ninja. Having the troops stop, turn around, and march back to the initial dispositions may create its own exposures.

 

[puppet on a string]

i think that events should have a lock for a few seconds before the buttons work, so that players do not accidentally click an option when it appears

i believe some other games have this

 

[Primarh]

Hello! Add more detailed information about the captured equipment to your game. Not just a message of "captured 200 units of techniques", but at least a table after the completion of negotiations. How many captured units, what a technique. Thank you.

 

[Pellaken]

was there another update

 

[Stolen Rutters]

Yes, an update to the beta anyway.
Edit
It could also be no. Final release If the beta was delayed, so not today if you were expecting the patch update.

 

[Mastergrunt712]

If not in this update but the next, could submarines on convoy raiding retreat after all convoys are destroyed instead of trying to fight an escort fleet?

 

[Oskgamer]

I wonder how Paradox still doesnt understand how difficulty supposed to work in games.Higher difficulty supposed to mean better AI,not unfair and useless negative modifiers.

It's pretty typical for games to rely on modifiers rather than different levels of AI to do the job. We'll have to live with it until every game comes backed by a neural network that can share learning across every game played.

Or they could make it think "deeper" (as in chess difficulties) but then the game would run slooooow.

 

[nkibilko]

i think that events should have a lock for a few seconds before the buttons work, so that players do not accidentally click an option when it appears

i believe some other games have this

This needs to be a thing. I can't tell you how many times i've done something stupid like puppeting Slovakia instead of annexing it or giving it to Hungary, because the same second the event pops up i was already clicking the map to select a division... and then the event intercepts my click instantly and ruins my game.

 

[Hearth]

- Corrected British division namelist to use brackets instead of apostrophes.

Awesome! Fixed a big stickler for me!

Now to just fix the US namelist that is horrible.

 

[Vyacheslav Molotov]

If there is any reason why I play recruit or civilian, it is simply because it can get annoying when your playing Germany or Italy, to have the whole world against you, or trying to naval invade the U.S or U.K, when you cant create a competitive navy, and so you need the buffs that come from the easier difficulties. And with the new patch, I hate how you can only have a limited number of paratroopers. That's not how any wars, work its ridiculous to think the Wehrmacht could have 10 paratroopers in a full army, and when the U.S joins it becomes annoying as even if you manage to capitulate U.K after defeating the Soviet,(which in this patch you have to have half of the Soveit territory in your hands due to how war support works) you still need to defeat the U.S, there should be a decision to sign an armistice with the Allies, if both U.K and the France falls, like when France does so as well. There's no way the U.S would fight after if U.K historically was defeated.

 

[jye42]

I play on the easier difficulties when I play as a minor nation. I want to be able to impact the game more even as a small nation. If I am playing as an especially powerful nation I like to play with the difficulty turned up.

My Dad plays too, he always plays as the easiest difficulty, and nearly always as Germany "I can't win otherwise" he tells me. He enjoys the game but finds it too hard, based on conversations with him I have a few reasons for our difference in skill.

I read every dev diary and watch let's plays. He does not. (I, therefore, know new mechanics and understand them while he has to muddle his way through)
He seems to especially struggle with attrition (he does not reduce the number of divisions on the front line when having supply difficulties)
He never micros units and so only uses battleplans to execute invasions

I have watched him play and honestly, I can't be exactly sure what he does wrong but these things all contribute somewhat. I hope this was helpful.

 

[puppet on a string]

Awesome! Fixed a big stickler for me!

Now to just fix the US namelist that is horrible.

what needs fixing?

 

[mabus81]

"- Holding shift now creates army group fronts without creating sub army fronts"

Didnt see that at first. Nice!

 

[Zauberelefant]

If there is any reason why I play recruit or civilian, it is simply because it can get annoying when your playing Germany or Italy, to have the whole world against you, or trying to naval invade the U.S or U.K, when you cant create a competitive navy, and so you need the buffs that come from the easier difficulties. And with the new patch, I hate how you can only have a limited number of paratroopers. That's not how any wars, work its ridiculous to think the Wehrmacht could have 10 paratroopers in a full army, and when the U.S joins it becomes annoying as even if you manage to capitulate U.K after defeating the Soviet,(which in this patch you have to have half of the Soveit territory in your hands due to how war support works) you still need to defeat the U.S, there should be a decision to sign an armistice with the Allies, if both U.K and the France falls, like when France does so as well. There's no way the U.S would fight after if U.K historically was defeated.

The germans had a Maximum of 2 airlanding divisions. From 41 onwards, Fallschirmjäger were simply elite infantry. There was a Maximum of 13 active SS and Wehrmacht mountain divisions. (The SS had some foreign "Mountain" divisions on top of that.
So, historically, I would argue that limit works.

And I very much doubt that Roosevelt would have signed an armistice with the germans, barring some outrageous ahistorical premises, Like a victorious Japan and a nuclear armed Nazi Germany, both self-sufficient and Not at war with wach other or another major Power.

 

[Kaisha]

Not really, the major issue with Paradox games and especially HoI is being history based. The proper way to code an AI is to build the entire game around it, if there is anything you are struggling to have the AI deal with you simply change/remove until it can but being a WW2 game means there is tons of things you need to include, or at least give a nod to, and thus you can't do that so we get a bunch of workarounds and out of place limitations that introduce a bunch of problems and ensure the player can always run around the AI.

The AI of a game like HoI is simply never going to be as great as a game like SC*. If Paradox settled on precisely how they want the AI to play historically, and how they want it to play optimally it could be at least good; but often it feels like Paradox starts with the idea of what (they think) players want, develop it, try to fit the square peg in round hole in getting the AI to deal with it properly, inevitably fail and add arbitrary limits to help the AI a bit then move on to the next mechanic where they do the exact same thing, instead of figuring what they can make the AI do and then figure out how to present those options to players.
*Given the remotely close levels of time, money, effort and such invested anyway.

That said comparing Blizzard of all companies is kinda pointless because out of all strategy games I have ever played the AI of all fall short of theirs. Blizzard might horribly suck at the story part but damn their gameplay is great.

SC2 was designed from the bottom up for competitive e-sport multiplayer, not to make AI design easier. With HOI4 everything is quantized into discreet decisions. Planning, pathing, building positioning, front lines, all fall along simple and well defined regions. In SC2 everything has to be done in real-time at a much higher level of precision. There's also more difficult decisions to be made with building/unit production/upgrading, harassing, defending choke points, scouting. There are so many gameplay aspects and mechanics that don't even exist in HOI4 like a proper air component, stealth, formations, ability interactions (when to use fungal, psi storm, nukes, etc...). Its not even in the same league difficulty wise.

In fact the parts of HOI4 that aren't completely trivial, are the parts that the HOI4 completely fails at (like everything naval).

The HOI4 AI is a travesty, and for a strategy game is completely unacceptable. What they need to do is hire one proper AI programmer that knows his stuff, and not some intern/random gameplay programmer, and they could have a decent AI that could beat easily 90% of players without boosts. I've seen hobbyists write better AI code than the HOI4 AI for mods/bots. It certainly CAN be done, but since fanboi's will shovel money at them no matter how poorly it plays they have no incentive to actually bother.

 

[chrnno]

SC2 was designed from the bottom up for competitive e-sport multiplayer, not to make AI design easier. With HOI4 everything is quantized into discreet decisions. Planning, pathing, building positioning, front lines, all fall along simple and well defined regions. In SC2 everything has to be done in real-time at a much higher level of precision. There's also more difficult decisions to be made with building/unit production/upgrading, harassing, defending choke points, scouting. There are so many gameplay aspects and mechanics that don't even exist in HOI4 like a proper air component, stealth, formations, ability interactions (when to use fungal, psi storm, nukes, etc...). Its not even in the same league difficulty wise.

In fact the parts of HOI4 that aren't completely trivial, are the parts that the HOI4 completely fails at (like everything naval).

The HOI4 AI is a travesty, and for a strategy game is completely unacceptable. What they need to do is hire one proper AI programmer that knows his stuff, and not some intern/random gameplay programmer, and they could have a decent AI that could beat easily 90% of players without boosts. I've seen hobbyists write better AI code than the HOI4 AI for mods/bots. It certainly CAN be done, but since fanboi's will shovel money at them no matter how poorly it plays they have no incentive to actually bother.

The game design I am talking was inherited from Starcraft(Well strictly speaking quite a bit came from way before that with Warcraft, you can very clearly see Blizzard's improvement overtime by looking at their games) so just that alone leads me to believe you have no idea what you are talking about.

The AI could, and should, be a lot better but by being so ridiculously extreme you undermine any point you may have and, in fact, is the kind of behaviour that makes people believe it can't be that bad if people resort to such shitty criticism instead of actually valid ones. If your goal is a better game the way you argue is actively detrimental to it.

Not to mention that if you ever want to be taken seriously you should avoid both the words 'fanboy' and 'hater' since by this point it has been ingrained culturally that anyone who makes use of either does so because they have no point and just want to insult people.

 

[Stolen Rutters]

seriously you should avoid both the words 'fanboy' and 'hater'

So true. These are just insults. As accurately construed, both words define precisely zero humans. They always have.

Back to whether the game is good enough to buy, some people say no, some of us say yes. The computer player algorithms need to improve, but thinking back to how vastly different the game is from its earlier forms, I am not surprised.

A number of simple meta decisions have to be worked out and incorporated into the algorithm, except “Game Balance” shifts with every DLC/patch.

How many guns and planes is always good to build? What is the optimal unit template? What is the best time and place to attack? These factors have changed every time the revision was indexed.

I am still impressed how the computer can pinch my salient if I fail to cover my flanks now. I never had to worry about that before.