Hearts of Iron, Observations of Matchsticks

This entry is part 1 of 3 in the series Hearts of Iron

For the last decade, I’ve been a fan of Paradox Development Studio’s Hearts of Iron grand strategy games. HOI players control all aspects of their chosen nation during World War II: army, navy, and air force; diplomacy, espionage, scientific research, industrial output, and domestic politics. This can be as overwhelming as it sounds, and it’s interesting, and a little instructive, to compare the approaches taken by different games in the series.


The original Hearts of Iron (2002) was an unwieldy monstrosity, its vast scope at odds with its obsessive granularity. To build tanks, you had to separately research a tank chassis, and tank suspension, and tank propulsion, and choose a calibre for the gun, and repeat this for each model of tank… and yet, at the time, I loved it to bits. Hearts of Iron II (2005) was far more polished, with a far keener sense of what was genuine depth and what was just bloat, and I loved it too. Hearts of Iron III (2009) was poorly received at launch, but several expansions left it in much better shape. And lastly, Paradox eventually licensed HOI2 to several fan groups, which produced their own spin-offs; I tried two and enjoyed one, Darkest Hour (2011) (1).


(At this point, I should note: (a) I have played a good chunk of HOI3 and Darkest Hour, but I have not played them exhaustively – certainly not as exhaustively as I played 1 and 2; and (b) for that matter, I may be looking back at 1 and 2 through rose-coloured glasses.)


Comparing HOI2 to HOI3 is a study in contrasts. The former is a traditional strategy game – i.e. the player is supreme commander and supreme micromanager – and is about as elegant as a game of this size could be. Returning to my tank example, in HOI2 you would research a single technology – say, 1939 medium tanks. Presto. You would move the resulting armoured divisions around the map, order your air wings to bomb a specific region of the map to support them, etc. In contrast, Hearts of Iron III’s design philosophy could be likened to a more rudimentary Distant Worlds. Playing HOI2-style is possible, but I wouldn’t recommend it. Rather, HOI3 piles the complexity back on – but provides extensive AI automation to help players manage it. The effect is a game about playing Roosevelt rather than Eisenhower. I focused on designing and organising, rather than commanding, my army – i.e. I’d configure the home front so I could produce X motorised corps of 4 divisions each, Y mountain corps, and Z marine corps, then turn over each corps to AI control and set appropriate goals for the AI.


In practice, the success of HOI3’s approach depends on the quality of the AI, and while sometimes it worked impressively well for me, other times it went beyond “aggravating” all the way to “bizarre”. For example, as the US, I decided to send an expeditionary force to fight off a Nazi invasion of Norway. This was what happened:


1. From previous experience, I knew it would have been futile to expect the automated AI to transport troops overseas, so I manually loaded two corps of GIs onto transport ships, and also organised air support.

2. I manually sailed the transports and flew the planes to Norway. So far, so good.

3. I placed the troops and pilots under AI control, with two orders: (a) hold southern Norway, and (b) kick the Germans out of their beachhead in northern Norway.

4. The ground troops performed superbly, manoeuvring through Norway and stomping the Germans on their own, but…

5. … the AI flew the aircraft all the way back to Iceland and the continental United States. Huh?


Now, I did have fun in that HOI3 game (before it was cut short by a crash), but for a design to use AI automation to mitigate extreme complexity, the AI must be rock-solid; and as that example shows, HOI3’s AI wasn’t. Like Distant Worlds, HOI3’s design is an interesting experiment; military automation is something I’d actually like to see in other Paradox games, where fighting wars is often less interesting than preparing for them. But I am not yet convinced that the HOI3 experiment succeeds. As is, I can’t help but think that HOI 3, like Civ 3, introduced new ideas into its series but lost at least some of its predecessors’ magic. I would like to see any future HOI4 borrow automation (and the “manage your country, not small individual armies” concept) from 3, but fuse it with 2’s more elegant design philosophy and general level of polish. And hopefully, to continue my analogy, HOI4 will be the game in which the magic returns.


(1) The Darkest Hour base game feels extremely similar to HOI2, so similar that I can’t actually comment in detail – I was too burned out by the time I picked it up. However, I had a lot of fun with Kaiserreich, the included alt-history mod. Kaiserreich starts with a plausible premise (what if Germany won World War 1?) and spins off into a gloriously bonkers version of WW2, in which I led royalist Canada into a five-way Second American Civil War before eventually liberating Britain from communism. It took years for my aircraft carriers to establish mastery over the Atlantic, and years more to train up a decent marine corps; but once my Canadian armoured divisions made it ashore, they went through the communist militia like a hot knife through butter. Well pleased, I declared victory and wrapped up the game. In other words, Kaiserreich gave me the familiar HOI2 WW2 toys, but put them in an unfamiliar situation, and did that well enough to be worth my price of admission. (That said, prospective purchasers should be warned: I do not know whether that mod is compatible with the latest version of the game.)


My copy of the latest HOI3 expansion, Their Finest Hour, was supplied by Paradox. However, I bought the base HOI3 and its previous expansions out of my pocket.

Series NavigationHearts of Iron IV Q&A, with Dan Lind >>

6 thoughts on “Hearts of Iron, Observations of Matchsticks”

  1. I wholeheartedly agree with you on the problem of the Combat-AI. If it would work 95% of the time, and the AI would employ tactics that are at least on par with the ones players would actually use on their own, this would have been an amazing feature for a game of this scope. Providing your generals with the necessary resources, units and technologies and let them follow your orders to conquer/defend territory, while you don’t have to mico-manage individual divisons would have been absolutely awesome.

    Unfortunately, I ran into the same issues. The AI sometimes made dubious strategic choices and usually was only usable to defend a front while I set out to control another offense somewhere else. This left me to actually guide all units myself – resulting in a micro-management heavy game that missed the streamlined combat of the HoI2. Unit stacks in particular were really hard to guide and control, and my German advance on Moscow took way more time than necessary with the management mess I found myself in.

    Anyway, very valid points, and I can only hope we’ll hear something about a possible HoI4 soon! :)

    1. Thanks, Andreas! Glad you liked the piece. It’s interesting to hear your perspective on how the AI performs during Operation Barbarossa. I mostly play the US and UK in the Hearts of Iron games (though I’ve also tried other countries – Germany, Japan, Canada, etc), so my emphasis is a little different — I tend to pay more attention to how the AI handles navies and the Pacific War.

      If you are interested in a game where the automation works, I recommend giving Distant Worlds a look. For example, you can let the DW AI handle logistics, multiple fleets with their own objectives (garrisons, offensive forces, patrols, etc) and R&D, while you look after diplomacy, production, and a particularly important fleet. It’s very expensive, and its setting is obviously very different from HOI’s, but overall it’s one of the coolest and most unique strategy games I’ve played. :)

  2. I have played HOI 2 HOI 3 innumerable times, but always quit once the shooting starts because directing armies was so frustratingly difficult. I feel a little better upon reading that others have had the same problem.
    When I played HOI 2 as the Soviet player, it was easy to beat the Germans and go on to conquer the rest of the world. The difference was that the incometent AI wasn’t screwing up my plans. It reminds me of Napoleon’s famous saying: “Never interrupt your enemy when he is making a mistake.”
    It is perhaps the case that true AI-directed armies are still a bridge too far, and that, for HOI to get better, it must go backwards and give us back our micromanagement rights.

    1. Thanks for commenting! I have to disagree with you there — I like the idea of AI-directed armies; it’s just that my concerns relate to implementation. But it’s interesting to hear your perspective, too. :)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.