I’ve played Stellaris before but I know so much as changed that it’s almost a new game to me… but I hate dropping time to learn a game and then not enjoying it (which is likely why I have a decent Steam backlog because I can’t commit!).
I think SotS1 was pretty decent about moving fleets about the star map and choosing what designs to build. It helps that the planetary economy was dead simple in that game (SotS2 AIs suffered complete economic collapse from just a slightly more complex economic model). I think the AI was also pretty good at smacking around players who didn’t realise that 6 destroyers and a few satellites wasn’t much of a defence. It also made good use of stealth fleets.
For the tactical battles, I don’t recall it doing anything smart. Basically it was wait for wave to warp in, charge with auto-targeting. Maybe I am forgetting something. It speaks to the game design that the tactical battles were so much fun, even if it was sometimes like shooting fish in a barrel.
Yeah sorry, I meant the AI was pretty good at the operational movement. Having distinct movement modes was a core and very awesome feature. And Stellaris shows that it’s not that easy to pull off.
Also great was how the racial ship design and weapons probabilities influenced everything.
SOTS 1 is good fun, its simple, like an early Total war game, Shogun 1 Total war had actually a pretty good campaign AI :)
That said, right now, Stellaris is the best 4x space opera game out there, ES 2 is cool, but an odd bird, and space combat while pretty, is just weird and tedious to get into, ES 1 was actually better…
The rest are just Moo clones that just time has run away from.¨
Quick edit : Forgot Galciv 3, its a snorefest of a game.
Honestly I only even have these qualms with the AI on games I’ve got hundreds of hours in because at that point I’m invested in the game enough to really understand the nuts and bolts behind what is going on and you realize that the AI is pretty unfortunate at playing the game. I also understand in something like Stellaris most players don’t actually want an AI to defeat them 15 hours into a campaign, which is why the Crisis is such a pitiful thing on default settings.
As far as examples of actually good AI in strategy games (beyond the obvious ones that AI has mastered and can’t really be defeated anymore like Chess and Go), I don’t know that there really are any that I could point out as being a worthy opponent without massive cheats or flatly playing a different game than the player. I just like to complain about stuff that bugs me. Obviously I still play the games but there is always room for improvement.
I’m told by people in the game industry that the main reasons AI remain kinda hot garbage are vaguely 3-fold.
-
People on average do not enjoy losing to an AI that isn’t cheating or playing by somewhat different rules than they are. Of course programmers could just input generically strong build orders for the AI to follow, but then new players would lose on normal every time and games are many hours long.
-
Marketing says “Fantastic AI” doesn’t sell units. I find this one harder to swallow because there really hasn’t been a game with fantastic AI, how the hell do they know? But I understand that you can show a video of beautiful graphics and that sells games a lot more easily than you can with vague promises of a good AI.
-
Good AI is hard to program, and most of the people who are any good at it are working at places like Google and not on video games.
And I get it. I don’t really want the AI in Stellaris to just run exact build orders or anything. But I’d be happy if they could get it together a bit on the war front. Stuff like adding in a check before they declare war to make sure their fleets are any where near their war target. They clearly don’t, I’ve seen the AI declare war on smaller enemies, but oops their fleet is on the other side of the galaxy and proceed to lose what should be un-loseable wars because their homeworld gets devoured or whatever in the literal 3 years it takes to get their fleets into position. That stuff makes me face-palm pretty hard.
EU4 still feels somewhat better here because the static world with asymmetrical starts doesn’t really lend itself to having the player micro-analyzing what decisions the AI is making outside of combat.
It wouldn’t surprise me if Stellaris 2 is set in a static, asymmetrical galaxy with pre-made races, and an option to have custom ones like in EU4. I think it would address a lot of the issues people have with Stellaris in general.
And despite all my nit-picking I still agree with @Janster that if you want a 4X space game, Stellaris is probably as good as you are going to get unless you want to play something with 90’s eras graphics or a MoO reskin.
The first reason is bullshit because of the third reason. If you somehow made a fantastic AI, you could always gimp it for “normal” levels. IE: Just make it collect half the resources of the player or something.
Also the third reason is somewhat bullshit (but not totally). For anyone who says the AI is super hard or impossible, I would just point to Age of Wonders 3. The strategic AI is not brilliant, but vastly better than the strategic AI in nearly every other game. The tactical AI is actually quite good.
I’m sure there are a number of examples of strategy titles with good AI and I’m simply unqualified to speak about them. I’ve played Age of Wonders 3, but didn’t find it compelling enough to consider myself good (let alone expert) enough to tell how well or poorly the AI is performing relative to a human that is making min-maxing decisions.
And you can call bullshit all you want on point one, but I’ve spoken to numerous people in the industry about this topic because I’m at least somewhat passionate about it, and they all agree that setting your “Normal” difficulty to be “Beatable by anything with a pulse” has become the industry standard because of the sheer amount of negative comments and posts they get when people lose on Normal. Older people would consider that “easy” or “beginner”, but that’s the new Normal. No one wants to feel like they can’t beat a game on Normal.
I was not calling reason 1 bullshit because people do not want to lose on a “normal” setting. I was calling it bullshit because it was framed as a reason the AI sucks. Not sucks on normal, but in general, just is terrible.
Ah I see. I include it because some of the rhetoric I’ve had thrown my way by actual developers is that since people don’t really want to lose strategy games anyways according to their internal metrics, investing hard into AI feels extra pointless when it is already a pretty non-glamorous selling point from a marketing perspective.
So their internal reasoning is “Well, players get pissed when they lose to AI, and AI doesn’t sell games, therefore why bother”.
Obviously this answer frustrates the hell out of me.
A quick browse on reddit and they’re saying it’s just battleships all the way.
I gave up on Stellaris a long time ago.
It’s not much worse than its competition - Endless Space 2 was hamstrung by some awful expansion content that actually introduced worse behaviour into the main game, and a combat system they never really finished or balanced; SotS1 is great but showing its age graphically (and that voice-acting, gah!); Stars in Shadow is actually great but low a much simpler game (though there’s lots to like in there if you aren’t expecting a ton of features). I forget the rest.
But Stellaris seems to fail on every front. The combat system isn’t fun or even pretty compared to other games; ship designs aren’t impactful (nor are the ship art design worth any time to look at - necessarily they are generic designs); strategic military positioning doesn’t matter (contrast with SotS human-play, where node-travel and tanker refuelling stations are critical!); the AI is poor even by 4X standards; the population management annoying not interesting; the game hits end-game tedium by the mid-game…
Tried Stars in a Shadow, its okay, but its just another moo clone, its nothing special, a few quality of life improvements…I’m not going back in time.
As for combat system and stellaris, its not as simple as you think, also it’s also not more than it should be, combat often clutters a 4x, there is a LOT more happening thank just simple pew pew, so I don’t mind that combat is less detailed. Its about the same style as EU 4 really.
Also strategic military positioning? Have you tried fighting at a neutron star?
If you find population management annoying, I am thinking this is a 4x not for you anyway.
Too many 4x today focus too much on fighting, and too little on building, Stellaris is far from perfect here, civ 6 actually got closer, but their combat system is something the AI doesn’t even begin to understand.
I actually think the combat graphics look fine for the genre. It’s a whole hell of a lot better than Distant Worlds. The explosions are the worst culprit, they look like 2d and as if they have about as much impact as a feather.
And ship design is pointless because it’s badly balanced and yea, it’s battleships all the way once you get AI combat computers. Throw in a Titan aura and you never miss except a bit on corvettes who blow up in one hit and will still get hit half the time, before they even get in range. God help the AI if you get the enigmatic targeting accessory. There’s maybe some argument to be made for a handful of picket destroyers just so you don’t lose random battleships to stray missiles but it isn’t impactful enough either way to matter. Strike craft still may as well not exist in game, even after patching them.
I kind of wish they’d get their act together on making the basic stuff actually good instead of constantly adding new features that essentially add up to +% modifier to X, but this is the oldest Paradox complaint in my long book of grudges against them. 😅
It would be nice if the combat graphics were more comparable to something like Battlefleet Gothic, though apparently metrics say that a large chunk of grand strategy and 4x gamers have potatoes for PCs or prefer to play them on a laptop, which is unfortunate. But I dunno, Total War Warhammer is a 4X that actually looks pretty good and sold a lot of copies, so with the precedent there maybe we’ll see some better looking efforts in the future.
Sorry, another PDX tangent - this reminds me of one beef I have about HOI4: the division designer. I am Der Fuhrer, the leader of the reborn German Reich leading my trueblood to victory in Europe… aaaaand I have to decide how many artillery/AA/AT battalions each land division has. Unlike some games, you can’t delegate that task to the AI.
That one doesn’t strike me as too egregious, because it isn’t a task you need to perform dozens of times and the composition of your regiments does actually have relevance to your production needs which outside of maneuvering your troops is probably the screen you need to mess around in the most.
Try terraforming all 40 of your inhabited worlds into Hive Worlds in Stellaris if you want to see what going through actual UI hell feels like. All the information you need isn’t there, you need to go through two more clicks than necessary for each world, and there is no visual indicator on the planet screen about whether a planet is currently undergoing terraforming, you need to examine each world individually for that information. It’s an absolute clusterwhoops of the highest magnitude and the game developer has a UI background.
My confidence in Stellaris as a project is basically nil at this point.
Edited to add: Which is sad, because I think the game was actually fun back when they took out non-hyperlane travel. The tile system was bad but it didn’t feel so endlessly tedious like the new pop system does. The diplomacy expansion not actually making diplomacy any good is not great. In a game that runs on the same engine as EU4 which has very good diplomacy. Ugh.
Yikes. I’ll make a beeline back to Distant Worlds before this. I enjoyed any micro I had with DW.
Sounds like you guys are describing the Chick Parabola. Summoning my thread on this topic. One notable mention there is Pandora, which I have bought but not yet tried. Apparently post-release, they allowed a modder to rebuild their AI, and it’s awesome. Also, Deity Empires is supposed to have good AI – the programmer seems to know his AI programming stuff. And I think Alpha Centauri has fixes by modders that make it pretty decent.
The sad fact (restating the Chick Parabola) is that most strategy games simply don’t have sufficient AI to handle their mechanics, which makes investing time learning those mechanics in-depth seem wasteful (master the game only for it to disappoint you). This is doubly true for Paradox games, which keep evolving their mechanics over time. Actually it’s triply true for Paradox games – the realtime thing makes it impossible to come up with good AI, especially since the AI has to be light enough that it can work for a gazillion entities in 5x speed. It’s pretty much the very worst case you can have for a poor AI programmer to deal with.
(Also note that I don’t mind if an AI cheats, so long as it’s done behind the scenes. If the AI cheats but still can’t function decently in the portions that are visible to the player, then that’s not helpful.)
Through a number of marginal bonuses!
There’s very little marriage of theme with mechanics in Stellaris. It hinges hugely on the player’s imagination turning a description and a small gameplay effect into a narrative.
This. I approach Stellaris as an RPG. It’s an open-world story engine. It’s great at that! I never finish games, however, because they always reach a point where the tedium overwhelms the narrative.
okay i’ll bite.
I’ve played just about all space 4x’s I’ve gotten my hands on so far, and its been a rough ride, in the end, I find Stellaris to be the best of the bunch, cause mainly it keeps receiving stuff and adding onto an interesting universe.
Secondly, I’ve played Distant Worlds a lot…and even with ALL micro turned on, I found precious little to do in that game…planetary management …nothing…space…well you move some ships, but they mostly move themselves…combat is weird and not really good…
It does give some sense of a living universe, but its not really that good, a very barebones game compared to Stellaris…there is simple very little there to be honest, even though it looks like it.
Stellaris has gotten more complicated, and its no longer just pick up and play, management however can get hard work in a certain cases and when you do very large universes…but you shouldn’t really be too bogged down.
It REALLY matters how you build your planets, the AI however is so and so, but you can adjust it decently, they however do 72 man MP games online now…if you have the time…
Then you’ll really get to test your mettle.
Some thoughts here. Strategy games are interesting beasts… on the one hand, you have the actual strategy mechanics, and the strategy mechanics should provide for interesting decisions, which should have varying and non-obvious results in terms of your success in the game. If an optimal path is obvious through a strategy game, that’s a problem. Additionally, strategy mechanics imply variety – if the same choices are given to you every time, you’ll quickly learn which ones are optimal.
On the other hand, you have the immersion of the game subject. Board games generally provide minimal immersion – they’re mostly about the mechanics and about interacting with other people. The mechanics tend to be very abstractly related to the subject matter since you have a very low complexity budget (mortals are expected to run the game, after all). In order to be able to run the game within the limited complexity budget and still provide non-obvious choices, you have to have heavy abstraction, with the subject matter often serving only as inspiration.
Computer games can go much heavier on the complexity budget, since now the game is managed by a machine. This, together with improved presentation, can give a much better feeling of immersion than anything supplied by a board game. However, this leads to a different problem: because you can have so many mechanics interacting, it’s much harder to make sure that the experience generated for the player actually has meaningful choices (i.e. non-obviously optimal ones) to make, and that the experience is sufficiently varied. Additionally, the computerized game experience may include entire artificial enemies (AI), which can lead to serious problems, as a malfunctioning AI hurts both immersion and strategy, while a non-malfunctioning AI is extremely difficult to program.
Paradox games sit very heavily on the side of immersion. They are the ultimate computerized strategy games, running as they do in pausable realtime. They feel like simulations of entire virtual worlds, with their hundreds of AIs, and this is one of the reasons people love them so much. But they suffer from being so far down on the immersion scale: it’s very hard to make sure that real strategic decisions are provided to the player, rather than ‘busywork’ (obviously optimal actions) to make the game hum along. There are so many mechanics, in fact, and so many custom ones per race/nation, that they provide multiple unique immersive experiences, depending on which part of the gameplay you wish to sample. But those experiences are highly unlikely to be strategically challenging. In this way, they really do cross over to story generators - procedural RPGs where you’re just trying to play out a story - rather than fully functioning strategy games. All computerized strategy games possess this element of story generation, but the best ones have it balanced out by the strategy elements.