Game is too hard with StarnetAI

[BakedPotato]

Starnet AI also have a sub mod (Same author) that make the AI follow their ethics (Instead of going for powerplay and treat the game as a MP).
That way diplomacy do work better and outside of fanatic purifiers you will likely be fine as long you spend a envoy + Some influence keeping empire X happy.

In fact that option makes the game harder lategame because AI´s will not join together quickly to defeat a Crisis.

Also its fun to make them use Mixed fleet. (Again, same author). It does "nerf" them a little but i also avoid using power fleets with only battleships and/or corvettes.
Mixed fleet submod -> https://steamcommunity.com/sharedfiles/filedetails/?id=2473773085

"Friendship" submod patch that keeps the diplomacy like vanilla.

Steam Workshop::~ StarNet Friendship Patch

steamcommunity.com

This is what I'm currently using, StarNet + Friendship and Mixed Fleets patches. Seems to be working pretty well. The AI knows how to efficiently expand, knows how to use different trade policies, how to build some economy and colony infrastructure, and will engage in diplomacy like one would expect. Honestly it plays more or less how I would expect vanilla to.

Haven't played enough to observe any repeatable anomalies, malfunctions, or "bugs" with the mod yet, but I'm kind of prepared for there to be some, just as long as they're not game-breaking.

 

[theBigTurnip385]

My problem with starnet is it breaks mega-corps since mega corp branch offices are built around a stupid AI that employs clerks.
Starnet doesn't do that so the AI planets have no trade on them.

Starnet should replace the clerk jobs on the mega corp buildings with +Trade just like the criminal mega corp buildings do.

 

[Troyen]

So a more "intelligent" AI being locked behind a higher difficulty level totally works for what you are describing.

This sounds like a good idea for players, but for developers it's a headache. Now you have to juggle multiple (how many levels are there? six?) different AI scripts and make sure they all can appropriately handle new features, balance tweaks, don't have bugs, etc. It also makes designing new things more difficult because you have to consider how it interacts with each of the different AIs. The scaling bonuses, while a bit of a lame way to add difficulty, makes things easier to develop and maintain because all difficulties share the same scripts.

Not to say they can't do it, or maybe add advanced features to their single AI (like different economy plans) locked behind difficulties, just noting there's a non-trivial cost to going that route. And it's an ongoing cost that has to be dealt with for every future patch, not a one-time-and-done cost.

 

[MathyM]

This sounds like a good idea for players, but for developers it's a headache. Now you have to juggle multiple (how many levels are there? six?) different AI scripts and make sure they all can appropriately handle new features, balance tweaks, don't have bugs, etc. It also makes designing new things more difficult because you have to consider how it interacts with each of the different AIs. The scaling bonuses, while a bit of a lame way to add difficulty, makes things easier to develop and maintain because all difficulties share the same scripts.

Not to say they can't do it, or maybe add advanced features to their single AI (like different economy plans) locked behind difficulties, just noting there's a non-trivial cost to going that route. And it's an ongoing cost that has to be dealt with for every future patch, not a one-time-and-done cost.

That's assuming they are using entirely different scripts. It could be done so they are all capable of being as smart, but lower difficulty levels make them opt for less optimal choices. It they want to keep the current script, it can be the script running for Cadet and Ensign. which require a lot less maintenance, since no one expects any challenge from those two.

 

[Incompetent]

Just play with standard difficulty without giving the AI any bonuses. I found that to be the most chill playthrough. Also playing on a huge galaxy and instead of filling it with 24 AI, fill it with around 18 so you have more space and borders meet later on. We all know the vanilla AI is trash, so simply setting it to normal difficutly setting is enough for the AI to be able to properly manage the planets. Also fixes the issue of vassals being completely useless in the vanilla game, because they don't get any economic bonuses regardless of difficulty -> leading to complete collapse as soon as an AI empire with difficulty cheats gets vassalized.

The problem is getting the balance right between the first 30-40 years and the rest of the game. If you set the AI bonus low enough to have a "chill" early game (depending on your skill level), you'll absolutely crush the AI later, even Starnet AI; sure, the AI won't collapse, but the scaling up of the economy just isn't comparable to what the player will do if they are left alone and allowed to invest everything in growth and tech. If you set the bonuses high enough that the AI can compete mid-game, you have an extremely tense early game as the AI pumps out masses of corvettes and tries to bully you with them, and then the AI still doesn't get that rich later because all those corvettes have taken an economic toll. Scaling difficulty theoretically addresses this, but the implementation is inflexible, and you still have the problem that early investment in military puts a huge dent in the AI's economic development curve.

I'll have to give StarTech a try, sounds more like the kind of galaxy I want to play in. I understand the "prisoner's dilemma" logic of Starnet (which also applies in competitive MP, presumably), it's just that I don't find Stellaris all that fun as an RTS-style game where it's all about "timing attacks" on your opponent.

 

[Tech Noir Synth]

The problem is getting the balance right between the first 30-40 years and the rest of the game. If you set the AI bonus low enough to have a "chill" early game (depending on your skill level), you'll absolutely crush the AI later, even Starnet AI; sure, the AI won't collapse, but the scaling up of the economy just isn't comparable to what the player will do if they are left alone and allowed to invest everything in growth and tech. If you set the bonuses high enough that the AI can compete mid-game, you have an extremely tense early game as the AI pumps out masses of corvettes and tries to bully you with them, and then the AI still doesn't get that rich later because all those corvettes have taken an economic toll. Scaling difficulty theoretically addresses this, but the implementation is inflexible, and you still have the problem that early investment in military puts a huge dent in the AI's economic development curve.

I'll have to give StarTech a try, sounds more like the kind of galaxy I want to play in. I understand the "prisoner's dilemma" logic of Starnet (which also applies in competitive MP, presumably), it's just that I don't find Stellaris all that fun as an RTS-style game where it's all about "timing attacks" on your opponent.

Yes I agree, you described it very well. I didn't even know Startech was a thing. The "problem" is, the Starnet AI goes pretty much all-in on Alloy production over tech, which does limit their potential somewhat. I think in a perfect world you would have the AI pursue both Tech and full alloy production depending on empire type, but that would probably be a ton of work to mod in. Although you could simply decide to go full alloy route yourself and not invest a lot into tech. I wouldn't advise it of course since rushing tech is the easiest win condition. But I would imagine this will make the game a lot more even between you and Starnet AI.

 

[ChiefBigFeather]

Ah yes. Supercharging the aggression of all empires and making the game steamroll you if you don't do a perfect meta game is the kind of AI that is going to be fine for everyone and not just the 1% of hard-core competitive players.

Starnet IS a very good AI if you know all the systems of the game almost perfectly and want to play competitive. Which is a niche case and definerly would be a TERRIBLE replacement for the most part of players.

And again, all these "boohoo the devs are lazy" kind of responses are incredibly tiresome and mostly show ignorance on how the development process of any complex software piece works, really.

Purposefully making the AI play stupidly is not the right solution though. PDX should start to actually balance the game instead. Intelligent players want fun too, the systems in this game are way too easy to break.

 

[Had a dad]

The OP asked for help playing a mod, and then others wanted to hijack the thread... stay ontopic

multiple posts deleted / infracted

moved to the modding forum

 

[Incompetent]

Yes I agree, you described it very well. I didn't even know Startech was a thing. The "problem" is, the Starnet AI goes pretty much all-in on Alloy production over tech, which does limit their potential somewhat. I think in a perfect world you would have the AI pursue both Tech and full alloy production depending on empire type, but that would probably be a ton of work to mod in. Although you could simply decide to go full alloy route yourself and not invest a lot into tech. I wouldn't advise it of course since rushing tech is the easiest win condition. But I would imagine this will make the game a lot more even between you and Starnet AI.

As I understand it, default Starnet is actually set up with a "war economy" and a "peace economy", it's just that the "peace economy" is not used very often in practice because there's bound to be some antagonistic empire nearby, and it's set up to respond to a military buildup by building up its own military (plus there are empires that will build up fleet early anyway because it's in their character). The reason it has to be this way is that if the AI tech-rushed despite seeing a nearby empire build up fleet, it will probably get conquered by the fleet builder, and then once that empire has blobbed, it starts winning on every metric, including tech. Players have the tactical awareness to know when they can get away with having a lot less fleet than their neighbours, and how to put off fleet building until just before that fleet is needed, but the AI can't be taught how to do this reliably, because the economic decisions are made based on relatively crude statistics.

If you compare to the Paradox AI, Starnet actually invests a bit less of its economy in fleet. It's just that it has more of an economy to start with because of better job usage, and because it spent at least few years on peaceful development, rather than choking its economy with corvettes from year 0.