I did a quick skim through the dev diaries and didn’t see mention of new achievements. I can’t think of an expansion that released without them, though, so I would be surprised if new ones aren’t included. Maybe next week’s dev diary will cover them?
The expansion looks like a heck of a thing. Actual politics (sort of?), encircling (oooh). Actual combat logs (gasp). Way bigger focus trees (yay).
Also the Dominion of Canada.
Yeah I’m pretty excited! HOI4 was a terrific game out the gate, with AI blunders being the only major black mark I had against it, so I’m really looking forward to these additions. A great game is getting even better. I know they plan on reworking the whole air war in an expansion soon as well, after that all my big gripes will be gone.
Oh, speaking of AI blunders, the AI dev (SteelVolt) is doing a series of AI dev diaries from now until release. He shares some war stories of HOI4’s AI development that highlight how difficult and frustrating AI can be in a game like this.
I said I’d pay $10 for better AI and UI. so the extra $5 for all the commonwealth improvements is really a fair deal.
I only played one short game as the UK so I’m anxious to play a real game as the UK. India was probably the most interesting game, I played and with a better-defined path toward independence, it should be interesting. Plus I’ll probably play Australia once.
I do hope the AI improvement are pretty substantial though.
If the expansion lives up to expectation this will make HO IV my game of the year.
I wish they said if naval battles and island hopping were better (AI). Has anyone seen comments on that yet?
SteelVolt has been doing mini dev diaries on the AI for the past few days. He said in today’s diary that he is going to talk about naval AI stuff on Wednesday.
Here you go, @jpinard, if these leaked patch notes can be believed (should be officially released within a couple days). From a very quick skim:
Awesome - now just need some big-ass naval battles ala Midway :)
I know Japan and its AI/scripting is exceptionally challenging due to being stuck in 4 major operational theatres (South Asia, China, South Pacific, US). But that is exactly why it’s soooo much fun when it works.
Here’s the developer’s post on changes to invasion AI. Interesting quote:
[quote]This weeks tale from the trenches will actually be about one of the points in the changelog. Sorry about cheating like this for the last diary, but it was too good to leave out:
- Scary fix makes AI actually consider all target areas of enemy, in stead of one at a time. Should set pacific area ablaze when the time comes, and keep the invasions rolling.
Wow…so…way back in development before the invasion craft tech limited the number of divisions you could use for invasions, the AI sometimes had a tendency to set up A LOT of them. As a fix for this, that did do the trick in the general case, the AI would consider ONE area per enemy for invasion at a time, before doing another one. When I say area here, it is any enemy territory with land connection to all parts of itself…shitty description, so a couple of examples: Ostpreussen is a separate area from main Germany, which is its own area, so Germany initially consists of two areas. When/if they annex Austria, it becomes part of the main German area. Similarly, every single island is its own area.
At this point, I am sure you can see at least a few problems this could cause in the pacific. My hopes were that Japan and USA would invade each other, and the invasions would be small enough to happen often enough that this would not become a problem. However…when the invasion is to an island that has had its port taken over by someone else, the AI would not be able to plan an invasion to it, and thus not make any more invasions until that area was fully conquered…which was not a guarantee.
This restriction was lifted, and the scary part of this was letting the AI be potentially a lot more active. It has turned out that the invasion craft limits and tweaks to the front balancing balances this out really well.
This is a good example of how fixes to earlier problems can cause new ones further down the line as development progresses. This happens a lot in AI, and not rarely will the same symptom pop up over and over but due to new reasons under the hood.[/quote]
One thing that makes me hopeful on this front is the new Paradox DLC model. If they can keep the same AI programmer crunching away on these issues for another year or more, the game will be immeasurably improved. Playing the latest Strategic Command game has convinced me that good AI needs a huge amount of time and a lot of iterations. It’s also reminded me how important good AI is.
It’s an interesting question why the AI in the Strategic Command series seems to accumulate over the course of years, but progress in AI for Hearts of Iron has been more haphazard. Is it because they have a new AI developer each time, or too many rules changes between iterations? Or is it the design decisions (like detailed control over unit composition) that creates more work for the AI developer and more places where the AI can go wrong?
I think it’s like all of the above, but especially the part I bolded. From a computer POV, I don’t know how much you can carry over from HOI3 to 4. Not only do you have to teach the AI new combat mechanics, but they also have to be taught a whole new strategic game again.
I’ve never coded AI, but from what I’ve heard developers in the field (including Paradox) say is that the limiting factor is rarely CPU horsepower or anything like that, it’s development time. The quality of a video game AI is mostly about a developer being able to iterate over and over and over and over to get the numbers just right. And since AI needs to be developed towards the end of the project when the rules/systems are mostly solidified… it often times goes out the gates with some issues.
I agree with you in that I think the Paradox support model can work out pretty well for this. They can keep working on the same core for years instead of moving on to a new project, and they’re also able to keep an AI developer on the title full-time after release. That’s huge, because normally that’s going to be limited to a patch or two and then it’s off to another project. It’s largely why I wasn’t overly concerned (annoyed at times, yes) about some of the AI blunders/ineptitude at release, I’m pretty confident in Paradox continuing to work on it post-release.
That is the most exciting thing I’ve read about HOI IV since it released. With the state of the game, fixing the pacific is all I really cared about because othing else mattered without a fix to the AI. I just really hope it works out like he said and we don’t have Japan sending undefended cannon fodder all over the place.
I’m quite excited about this also. I do like how they are podcat on this game. It almost seems like the AI for a game pretty much has to be done by one person and it is really just a pure iterative process.
Still, the one thing I would hope the AI should be able to better than a human is plan production.
After a playing the game a while, and being pretty good with doing math in my head. I can keep a pretty good feel of how many truck factories I need to have vs tank factories. Of course with efficiency changes and running out resource it is always a rough approximation.
Still it was exactly those type of challenges that lead to the of operational research and the American army’s superior logistical process. I heard from a WWII expert that the steel for the engine for the LST for the invasion of IWO was allocated 18 months before the invasion. Which involved whole team using early computers to make sure all this happened. So I’d hope that rather than being chronically short of something critical important like infantry weapons the AI player would be able to figure this out more exactly.
Look at all those negative Steam reviews over the price.
DLC isn’t free, you know.
At this point, I can pretty much copy and paste this exact same top ranked review from every DLC/expansion from every game I follow, regardless of developer
That’s the top ranked review every. fucking. time.
I actually quoted that same line in a Steam chat today, as my most hated ‘review’ of anything.
I just don’t understand why they buy it in the first place to then complain. Paradox are at least transparent about what you get for your money. Honestly I would not have bought this particular DLC at full price (getting it for free anyway because of the field marshal thing), but these things are subjective.
The HoI 4 team have a great sense of humour. I love the debug dog with the helmet you saw in all the alpha screenshots, and I just noticed this one:
Paradox DLC isn’t so bad if you get on their payment plan (buying it when it comes out) but it is almost impossible to catch up. Even in sales there is just too much.
I do think every DLC should be released with its companion sprite packs though (packaged with them that is).
While I think the lend-lease changes and the blitz command should have been part of a free patch, I do join the complains against the complainers.
The best (worst) one is a review comparing HOI4’s model against Stellaris’s and saying how the latter is way better and fairer (without, of course, acknowledging the reception Stellaris got was less than stellar -eh- at release and that maybe that’s why the different approach)…
I guarantee if the blitz was part of the free patch 95% of the negative reviews on Steam would have been positive (or just not posted). They made their bed on this one. Sales would have been EXACTLY the same regardless if this was a free feature or not. They were warned, they knew how touchy it was, and they wouldn’t budge. I’ll give it a positive review, and I too am tired of side-swipe reviews on Steam. But in this case, I think Paradox didn’t handle this right. I have a different take on things however in that I am supporting stuff to back their continued AI development.
I only saw a single negative review mention blitz. Almost everyone is whining about $15.