Well think of it like chess against an AI. Virtually any chess AI these days will win every game against almost any opponent, except the very best. To counter it we reduce it's ability to predict and maybe introduce a little variability and randomness to it's actions. It's still playing the game though, so as a human I can observe all the moves it makes, understand how and perhaps why it makes them, and maneuver my assets strategically. What we don't do is turn down it's difficulty and then have new pieces spawn for the AI to compensate. That would take all the strategy out of my game, because I wouldn't be able to put together and execute a strategy based on the board state. Instead, my strategy would change to respond to the spawning mechanism that actually poses a threat, turning the complex game of chess into a one dimensional game of checkers. Yes, it's still a game, but it's not the game I want to be playing.
Is game supereasy after 2.2? (i play on grand admiral diff.)
- Thread starter soylukan
- Start date
-
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.
Well think of it like chess against an AI. Virtually any chess AI these days will win every game against almost any opponent, except the very best. To counter it we reduce it's ability to predict and maybe introduce a little variability and randomness to it's actions. It's still playing the game though, so as a human I can observe all the moves it makes, understand how and perhaps why it makes them, and maneuver my assets strategically. What we don't do is turn down it's difficulty and then have new pieces spawn for the AI to compensate. That would take all the strategy out of my game, because I wouldn't be able to put together and execute a strategy based on the board state. Instead, my strategy would change to respond to the spawning mechanism that actually poses a threat, turning the complex game of chess into a one dimensional game of checkers. Yes, it's still a game, but it's not the game I want to be playing.
Is this a serious question? Because if the AI plays with its own rules we're not playing the same game. Strategy is out of the window, depth is out of the window.
The question "Why" is important I think, and it's important because the game as it currently stands already abstracts away many of the core concepts that would be relevant to a space-faring empire. The mechanics put in place strain credulity, and from a game-play perspective seem to stray pretty far from providing a fun, compelling experience. If you add on top of all that an AI that doesn't actually play the game, but just appears to, why exactly are we playing at all? The game doesn't simulate anything close to reality, the game-play mechanisms in place are a mess of 12th century feudal rules that don't translate well to space empires, most of the actual things you do in Stellaris aren't particularly fun, and the one bit that is fun, managing an interstellar empire, is only fun in a vacuum since your opponents don't really play by the same rules. To point, most of the fun I have in the game is in the early expansion period where I'm basically playing the game solo.
What? You are making no sense whatsoever. The strategy comes in your reactions to the situations the game world presents you. Your AI opponents are... not real. By definition they are not "playing" the same game you are. Their job is to present various game conditions for you to try to figure out and/or deal with. Whether they build a 10k fleet x years into the game and conquer your homeworld because they are super smart and built great planets, or they build a 10k fleet x years into the game and conquer your homeworld because their mediocre planets received production buffs, the "strategy" comes into your reaction to a 10k fleet x years into the game heading your way.
It doesn't matter how the AI gets there.
Even with Glavius AI, by 2250 all AI empires were so far behind me that I couldn't declare rivalry against any of them. Even when the fallen empires awakened around 2350, I was able to walk over them with no effort. I didn't even bother to upgrade my ships. The only thing that was slowing me down was the fact that planetary bombardment takes ages.
That kind of makes my point. You "reduce its ability to predict and introduce artificial randomness" to get an outcome you desire, that being a game which "appears" to be evenly matched even though one side had to have handicaps given to it in order for it to maintain that illusion.
Nerfing an opponent to create the illusion of competitiveness is not really different than buffing one to create the same illusion.
In my last game I got the first league precursor chain and hence got a size 25 ecumenopolis in like 2240 or something. AI economy was a bad joke compared to mine, even with grand admiral buffs. I suggest a rather strict set of house-rules to remedy this. Personally I always set limits on what areas / planets I am "allowed" to take, and have been doing that for quite a few versions now.
After the first league debacle (the early size 25 ecumenopolis sucked all the fun out of the game), I will now add a special rule that planets may only be turned into ecumenopolis if they do not exceed size 13. Also if I get the first league precursors again, I will restart the game.
After the first league debacle (the early size 25 ecumenopolis sucked all the fun out of the game), I will now add a special rule that planets may only be turned into ecumenopolis if they do not exceed size 13. Also if I get the first league precursors again, I will restart the game.
Don't Bombard planets, build more armies, its a lot faster. A 1k army stack will take out all but the hardest planets. 4-5k will take a FE homeworld, and you can build these while not at war.
I think it is different depending on whether the opponent follows the same rules or not. Reducing an AI's ability to branch predict and causing it to make some less than optimal decisions doesn't change the state of the board or the rules being followed. In your explanation, you want to change the rules for the AI so that it is a challenge to the player. I don't disagree that this is still a game, and some games intentionally pit the player against an unbalanced opponent. However even against an unbalanced opponent, maintaining enjoyable play (which is a very complicated topic) involves a few factors, one of which is the ability to learn and improve, employ a variety of strategies, and be challenged by different scenarios to react in different ways. Basically, the game has to have a large enough learning curve that expertise can be developed, and then needs to challenge that expertise by presenting different scenarios within the framework of the rules of the game.
Let's take your hypothesis a little further, their worlds are always fully undeveloped, and the game spawns fleets based on your relative power. Maybe the AI doesn't even colonize other worlds, because it doesn't need to in order to give you a competitive experience. Now you're reduced a fairly complex game of diplomacy, population management, and so on to a (bad) fleet combat simulator. Maybe that's the game you want, and that's fine. It is not, however, the game Stellaris claims to be.
There are also issues like this that are still unexplored, where the AI is just flat out cheating the game (is it a bug?). https://forumcontent.paradoxplaza.com/public/423498/War exhaustion issue no mods.png. It just isn't *fun* anymore when you can't strategize, plan, and execute, fail, iterate, and try again.
Of course it does. How can you counter a strategy when the AI simply doesn't have one. What you are suggesting is effectively reducing the AI to a bunch of standard events that helps them cheat. I have one better, lets just remove all of this altogether and put a big bad you lose event in place. Now that would be tough to beat. So no, sorry, not a "game" I'd like to play.
I also want to point out I'm not saying the AI can't fudge the numbers a little. There's a reality that AI programming is hard, and giving the AI a bonus or malus to production can go a long way to adjusting the game difficulty. However this has to be done within the existing rules of the game. Increasing mineral production by 15% is ok, but giving the AI 150 minerals a month is not. The former requires that the AI still play by the same rules, build the same buildings, etc, they just get a bit of leniency. The latter means the AI no longer has to build mines. However if you give too large of a % bonus, then the game falls into the latter territory. A fundamentally sound AI is the foundation of a good 4x game.
If you avoid min/maxing I think the game in the beta patch can be somewhat enjoyable for the first 50 years if you start with GA and all AIs set to advanced start.
That you rush non-advanced GA AIs by focusing on alloys early is no fun although advanced start AIs seem to start with enough ships to defend themselves if they bother to move their fleets.
As far as I know that is also MP meta too though.
Didn't they buff the original starports at some point to prevent homeworld rushing? Maybe home-systems starbases should get an unique buff as long as the original owner controls it?
There are so many ways to break the game now especially if you cook the map.
Some ways that used to make the game a bit more difficult actually make the game easier now.
Maxing the number of empires and setting them all to advanced start and setting the midgame date as early as possible and poking the L-Gate was fun in a way.
Now it means that you can hit 3000 pops by 2260 because of all the refugees.
This version sucked the joy out of min/maxing even worse than 2.0 already did(well plain 2.0 was fun in a way because of the max. fleet cap of 1000).
That you rush non-advanced GA AIs by focusing on alloys early is no fun although advanced start AIs seem to start with enough ships to defend themselves if they bother to move their fleets.
As far as I know that is also MP meta too though.
Didn't they buff the original starports at some point to prevent homeworld rushing? Maybe home-systems starbases should get an unique buff as long as the original owner controls it?
There are so many ways to break the game now especially if you cook the map.
Some ways that used to make the game a bit more difficult actually make the game easier now.
Maxing the number of empires and setting them all to advanced start and setting the midgame date as early as possible and poking the L-Gate was fun in a way.
Now it means that you can hit 3000 pops by 2260 because of all the refugees.
This version sucked the joy out of min/maxing even worse than 2.0 already did(well plain 2.0 was fun in a way because of the max. fleet cap of 1000).
Last edited:
Well to your last point I would agree just spawning fleets based on your own fleet power would start having tangible effects on strategy that simply buffing AI production does not. In that scenario you couldn't conquer a neighbor to reduce their war economy because their fleet capacity is not tied to their economy whatsoever. With buffs you still can, as I said earlier whether the AI has 3 great alloy planets or 3 mediocre alloy planets buffed to simulate great planets, taking one in either case removes 1/3rd of their alloy production. So your strategy is essentially untouched. And therefore what is behind the curtain remains something that does not really concern you because you don't notice it.
As to your larger point I understand where you are coming from and it's not an illegitimate position, I just mostly disagree. For instance I would argue that even with buffs they are still "playing by same rules" as the player. They have to spend alloys to produce ships, minerals to make districts etc. Their production is ultimately tied to planets/stations and the size of their empire. They just have a baked in bonus to the resources they produce to remain competitive with your human advantage of being able to be more efficient on when/how you specialize your economy. I guess where I draw the line is where something would have to be done that would alter your strategy to where your interactions with them would be different than if the AI was simply better.
Yeah, for me that line was crossed long ago. I absolutely play the game differently because of the poor AI than I would (and do) against human opponents. Not that I want to play the way I do against human opponents either, but they use the game's rules efficiently. I want the game rules to be different, and I want the AI to abide by them. Also I'm not here to bash on the game, it's really so close to being a great game and there isn't much else like it out there right now. But I want it to be better, so, so much better.
I don't know. I think a puristical point of view is based on somewhat unrealistic expectations here, and is bound to make you disappointed. The super-human state of chess AIs somewhat obfuscates what can reasonably be expected from the Stellaris AI, because Stellaris' rules are much more complex than those of chess. Chess rules and victory conditions can be written on a single piece of paper with no effort at all. "Teaching" that to an AI is easy.
Teaching Stellaris to an AI is in a completely different league, because it is not even really well-defined what the actual goals are for any given space empire: Think role-playing here... If your friendly xenophile pacifist neighbors basically starts acting like a complete sociopath because that is the most efficient way to "win", I wouldn't call that a great AI either (incidentally it is also why I don't like the multiplayer mode). Plus, there is obviously the manpower issue, because Paradox certainly does not have hundreds of developers assigned to the Stellaris AI.
For me personally buffs for the AI are perfectly fine, even if it are large buffs. I draw the line at the point where I feel the AI is playing an entirely different game. I can't say for sure when exactly I start to feel that way. But AI having to pay no maintenance on troops in EU2 was crossing the line for sure. I would have been fine with 1/2 or even 1/5 maintenance, but not with NO maintenance. Because at that point it felt like fighting magical unicorn troops. Another example would be the deity setting in CIV games. If the AI starts spamming cities and swordsmen by the point where it legally could not even have produced a single warrior, that doesn't feel like a buffed AI. That feels like an AI that is playing an entirely different game.
I've found the most recent AI build fairly competitive, especially when using full starlanes. The AI definitely doesn't do great with chokepoints.
On the other hand, Glavius is actually broken on my install. It can't assign pops to jobs, so whenever it's enabled every planet just sits with full unemployment.
But idk... the AI feels okay to me right now.
On the other hand, Glavius is actually broken on my install. It can't assign pops to jobs, so whenever it's enabled every planet just sits with full unemployment.
But idk... the AI feels okay to me right now.
There will not be an AI that isn't trivial for a somewhat competent human to out-strategise in the lifetime of this game. The AI as it stands is a long way below where it should be, but even if it's fixed to a decent level, and half-decent player is going to have to give the AI bonuses in order to have a challenging game.
Bottom line - if you want the fun of out-strategising an opponent, it has to be a human opponent or you're just fooling yourself.
after 2.2 the AIs just seem like they love to commit suicide. I don't have to do a thing. They automatically embroil themselves in social chaos and crisis.
Yeah, I have noticed this too.
I was curious to see exactly how bad the AI is and I created a galaxy where all 6 AIs were Fanatic Pacifist and Inwards Perfection. No crisis, no marauders, so there was absolutely no threat to any AI whatsoever with no possibility of anybody declaring war. Grand Admiral, no scaling, 25% tech cost. I left the computer play the game overnight in observer mode. When I woke up it was around year 2400 and literally all AIs had unempoyment, not enough housing on all planets. But they had enough resources, all planets had free district slots and free building slots, so the solution was obvious. Yet the AI just refused to construct more buildings / districts for some reason and was obviously entering a death spiral. However the lag was getting extremely bad (tiny galaxy, 25% habitability), so I put an end to this misery and stopped the game.
To the AI's credit, it figured out how to build A LOT of habitats to continue playing tall once expansion was impossible. However, all habitats were also badly developed, just like the planets.