Come on devs, Ring Worlds are NOT the reason tech rush is so OP.

[Carl_Bar]

It's another bad case of the dev team thinking correlation = causation.

The real reason tech rush is so overpowered as to be mandatory is that the AI is *spectacularly* bad at managing their resources and buildings in a practical manner. The AI's priority isn't going to be tech, or if it is, it doesn't know how to go about it properly since for most empires it's a stage 3 manufactured resource, made from consumer goods that are made from minerals, It's bad at long-term investment for those, and balancing its own economy catering to technology, energy, and alloy production in that order.

The ideal fix would be for the community to do a demonstration about each of the development stages that player empires go through that results in tech booming, since it's a relatively consistent process. Programming the AI's priorities and behavior to better mimic players in the early game means they'll be in better shape to manage their economy around when players do.

Again this doesn't fix the problem that if you've got a super powerful strategy and the AI is programmed to keep up with a player doing that, any player not doing that strategy is going to seriously struggle. You want multiple good strategies that don't pack significant difficulty differences even with a competent AI to work against. And that means those several strategies have to be very broadly similar in power level. You can afford it not being balanced to the degree of a competitive multiplayer PVP game, but you need the majority of strategies to be within the same general ballpark of power.

 

[Bohaska]

Firstly, the AI doesn't bother trying to build Research Labs until the midgame, even Technocracies. So if you build a couple, you'll easily outpace them.

The AI's economic plans don't prioritize research enough. The early game plan doesn't ask the AI to build research labs, in fact it only aims for the AI to have more than 0 research. Like seriously. Below are the research goals for the AI for each phase of the game.

Spoiler: Exact requirements

Early game: No requirements
Medium game: 50 research for each branch of science
End game: 50 research for each branch of science
Ultimate plan (after AI finishes end-game plan): 50 research for each branch of science
And the Ultimate plan literally has comments that say "End game plan is fulfilled, let's invest what we can in pop growth and science"

This irritated me so much I just edited the economic plans.

 

[Tech Noir Synth]

I think Startech AI mod already does it and is able to produce far better research output

Yes Startech AI is basically build for tech rusing. I haven't tried it myself, however the description says something about there being almost no early wars.

But there is also another option. Original Starnet makes a lot of changes to opinion bonus and malus and cannot just be swayed with Envoys, unlike in Vanilla. This does create a lot of wars and a very opportunistic AI which often times gang up on each other. This can be a great thing against snowballing Driven Assimilators or other Genocidals. But it also means Starnet focuses A LOT on Alloys over Research.

You can use the Starnet AI Friendship patch, which keeps all the proper empire management of Starnet, but changes the diplomacy back to Vanilla.
This causes the AI to invest way less into alloys. There will be way less wars and Genocidal empires are expected to snowball a lot more. But it also means that the AI will build a lot of science labs. I am playing with the Friendship patch right now, I achieved roughly 1k research by year 2240. I conquered one empire and two of their Federation-Origin buddies and they had multiple planets with 4 Science labs out ot 7 total building slots.

 

[Ferrus Animus]

It's another bad case of the dev team thinking correlation = causation.

The real reason tech rush is so overpowered as to be mandatory is that the AI is *spectacularly* bad at managing their resources and buildings in a practical manner. The AI's priority isn't going to be tech, or if it is, it doesn't know how to go about it properly since for most empires it's a stage 3 manufactured resource, made from consumer goods that are made from minerals, It's bad at long-term investment for those, and balancing its own economy catering to technology, energy, and alloy production in that order.

The ideal fix would be for the community to do a demonstration about each of the development stages that player empires go through that results in tech booming, since it's a relatively consistent process. Programming the AI's priorities and behavior to better mimic players in the early game means they'll be in better shape to manage their economy around when players do.

Another reason is that the AI has big income and upkeep bonuses that warp AI economy to a stupid degree. Early game economy is simple and the phase where the AI has a comparable economic strength to even a good player. And to prevent rushes the AI is nowadays forced build up a proper fleet, adapting its economy for it.
However the result is that minerals go into alloy for ships instead of consumer goods for research. And once the AI has a proper fleet all the aggressive AIs (which usually are a majority), will now look for easy target, like any player who didn't manage a fleet of their own or defensive pacts in time.
And as proper economic management hasn't had the time to overtake the AI yet, players ahve to focus on research a bit to keep up, meaning they'll overtake the AI in that regard.

The main issue here isn't tech rushing or fleet rushing or such, the main issue is that the way the AI is made, the way it is scripted to behave and the way the game works all lead to particular behaviors that go down the same rabbit hole, that also restricts player options a lot.

And while I have seen the AI do rather well in specific circumstances, I don't think it is easily fixable like that. The AI mods all have their own pitfalls.
And in the end the AI needs to play the game too, and needs to be suitable to be a functional environment of a variety of skill-levels among players.
IMO for that it needs a complete revaluation first.

 

[FlyingPhoenix]

This is actually less important than it seems. You don't need to abuse DA or the energy deficit to reach times like this - these factors just seem to help to make it slightly more consistent on average. Otherwise, you are relying on things like a normal ME getting enough planets to produce enough pops to scale by peaceful expansion alone (under these conditions my best time was 2259, but I was able to colonize 9 extra planets peacefully), or another type of empire finding a megastructure to repair in space, which means your results will vary a lot. Stefan in this video does not have good luck, does not min-max as hard as he would off-stream, and yet still manages to hit a solid time. This is a testament not to tech rushing, DA, Shattered Ring, or the energy deficit 'exploit', but to the overall idea behind the build and his knowledge of the game.

So while I do wish that some of these things will be tuned down a bit, and the energy deficit thing fixed, the 'solution' to tech rushing as I said in an early post is not to focus on these things as much, but instead...



..focus on this. The AI being unable to either punish the player for tech rushing, somewhat keep up itself, or a combination of both is the real 'problem' with tech rushing.

As I said earlier in this thread:

While I agree with the general thrust of your argument, that a better AI would be better for everybody, I do think it is an exercise in futility and frustration to attempt to balance the game when such meta-problems and exploits exist.

One of the problems I identified was that machine empires have a much simpler economy compared to non-gestalts.
That was the key concept in the quote of mine, the word convoluted. The economy isn't actually complex and deep, but simple, shallow and convoluted.

 

[Alfray Stryke]

Just as a reminder folks:

• The balance changes take into account our vision for the game, not just what's currently "the meta".
• The AI is being worked on (I'm currently the primary QA looking at the AI, so hopefully you can trust me).

 

[Tech Noir Synth]

The balance changes take into account our vision for the game, not just what's currently "the meta".

This quote could be interpreted in many dangerous ways by some people. For example, one might accuse you guys of having the "vision" of Materialists completely overpowering Spiritualists. Because thats whats happening since around patch 2.2.7. I started playing shortly before Megacorp and I played pretty much every iteration and I saw the rise of Synthethic Evolution from an Ascension that would completely screw up your game due to being bugged to getting big buffs in the following patches.

I made this account around Megacorp release and as you might imagine my avatar picture and nickname, Synth Ascension was my favourite playstyle and I didn't even care that everything else was not even close in terms of power. However its been so long since Megacorp release and I have realized the game needs more diversity badly. You guys did alot of changes already which made empires more diverse, nowadays I am enjoying Hiveminds the most. However, you never updated Spiritualists to the point where they could compete with Synth Ascension. Or rather, you didn't nerf Synth Ascension enough to allow other empires to compete better with it. Considering that big parts of the game enforce the rivalry between Spiritualists and Materialists, especially in the Galactic Community, you guys need to work a lot more on bringing the power of empires closer together. Lots of people would like to play Spiritualists but can't enjoy them because we know we are playing a faulty empire, especially if we decide against using Robots for pop growth.

So if your vision really is to keep Spiritualist as the laughing stock, then so be it. If your vision is to have a collection of different kinds of empires with unique playstyles which are somewhat comparable in power, then please listen to the balance concerns of the community and see the solutions already presented in Stefan's perfectly balanced mod. Give this mod a try and see for yourself how fun and varied Stellaris could be if everything was more balanced. Interestingly, its not just about number changes. This mod also alters some of the laws in the galactic community to have a big impact on the game, for example reducing the upkeep of Biological and Lithoid pops or directly nerfing Machine pop assembly. This forces all empires to use their diplomatic weight to make sure their type of playstyle comes out on top, which is a fantastic use of the galactic community.

 

[FlyingPhoenix]

Just as a reminder folks:

• The balance changes take into account our vision for the game, not just what's currently "the meta".
• The AI is being worked on (I'm currently the primary QA looking at the AI, so hopefully you can trust me).

Stellaris's development of multiple game systems has always been reactionary to the meta that players are actually reporting.


What's your computer science/software engineering training and background?

 

[Enfield_PDX]

Stellaris's development of multiple game systems has always been reactionary to the meta that players are actually reporting.

lol nope

What's your computer science/software engineering training and background?

Can't think of even remotely how that would be relevant, no course out there is teaching Stellaris script. You don't get to vet members of the team either. That's pretty insulting.

 

[Meebleborp]

Nerfing the origin seems okay, albeit I'd argue that Void Dweller needs the same. It's almost in the same league.

See my earlier posts in this thread where I go into detail and show how VD is not actually anywhere near ME Shattered Ring in power. It's honestly not even the next best origin, as Necrophage is quite a bit stronger.

Again this doesn't fix the problem that if you've got a super powerful strategy and the AI is programmed to keep up with a player doing that, any player not doing that strategy is going to seriously struggle. You want multiple good strategies that don't pack significant difficulty differences even with a competent AI to work against. And that means those several strategies have to be very broadly similar in power level. You can afford it not being balanced to the degree of a competitive multiplayer PVP game, but you need the majority of strategies to be within the same general ballpark of power.

As there are only two 'real' strategies in Stellaris, based on the two resources that actually help you win the game - research and alloys - this outlook is a bit troublesome.

The reason why Starnet is able to provide a challenge is because it's actually programmed to care enough about those two resources to be able to somewhat compete on both levels. If the player focuses too heavily on alloys, Starnet can advance ahead in tech and eventually stop the player from winning brute-force wars. If the player focuses too heavily on tech, Starnet can take advantage and attack the player before they can build up a high-tech fleet.

Here is an example from a recent game where GA Starnet punished me hard while I was tech rushing with no fleet:

I had open borders with two of the members of the Hegemony nearby, and so I was attacked from all sides. Luckily, I was aware of the likelihood of this happening, and I have a decent Destroyer fleet queued up at the time of this screenshot.

I eventually survive a long war, but this is a significant setback. By the time I can go to war with them again, my economy and fleet power are quite a bit weaker than where they would normally be:

The war takes about a decade to win, and even though I'm now in a much stronger position around 2250, the other AI have kept up and the game is completely up in the air. If this had been the vanilla AI, there's a high chance I would have been able to avoid the war in the first place, a 100% chance the war I declared myself would have been a complete steamroll, and no chance that after consolidating the gains any AI on the map would ever be able to compete.

While I agree with the general thrust of your argument, that a better AI would be better for everybody, I do think it is an exercise in futility and frustration to attempt to balance the game when such meta-problems and exploits exist.

One of the problems I identified was that machine empires have a much simpler economy compared to non-gestalts.
That was the key concept in the quote of mine, the word convoluted. The economy isn't actually complex and deep, but simple, shallow and convoluted.

While I agree the economy is indeed more shallow and convoluted than it is complex and deep, I believe many of the recent changes in 3.0 were made to address the balance between the simple (machine) economies and the complex (consumer goods) ones while still retaining what makes them unique and fun to play. Despite some aspects like logistic growth being a bit awkward, the overall effect has actually been quite successful. The general balance between normal empires, machine empires, and hive minds is definitely in a better place than it was before.

The one major outlier is ME Shattered Ring. As another poster mentioned in this thread, Resource Consolidation is actually in a very good spot - a bit weaker than Necrophage and VD, but right behind it in the next group of origins. The almost infinite scaling that comes with Shattered Ring just takes everything that is good about current machine empires and turns it up to 11.

 

[Aeroclub]

Just as a reminder folks:

• The balance changes take into account our vision for the game, not just what's currently "the meta".
• The AI is being worked on (I'm currently the primary QA looking at the AI, so hopefully you can trust me).

Thank you again for stopping by and actually reading our discussion! Everyone participating in the thread also has a different opinion, and the whole reason to have discussions like this is to just give you some food for thought that you may or may not take into account when thinking about your game.

 

[Alfray Stryke]

Thank you again for stopping by and actually reading our discussion! Everyone participating in the thread also has a different opinion, and the whole reason to have discussions like this is to just give you some food for thought that you may or may not take into account when thinking about your game.

Just because PDS folks don't reply to threads doesn't mean we're not reading them. I just wanted to redirect the conversation back onto what people thought was busted without being dragged down by "the AI can't compete with X, Y, Z".

 

[Aeroclub]

Just because PDS folks don't reply to threads doesn't mean we're not reading them. I just wanted to redirect the conversation back onto what people thought was busted without being dragged down by "the AI can't compete with X, Y, Z".

Well, I don't think these questions are separable. Something can only be overpowered or "busted" if it gives an game-breaking advantage over others. This is what makes the whole campaign boring and shallow. It doesn't matter how much research you have if there are some (even if not all) other empires who can compete.

You can, however, separate the approaches to solve this problem into two: one is to boost the AI to make them better, and another one is to somehow limit the exponential growth of the min-maxed player.

 

[FlyingPhoenix]

lol nope

Personally I think you being contemptuous and dismissive with your lol nope is pretty insulting. Did you intend it that way?

That's the impression I have after 5 years and I think it is actually a pretty accurate and reasonable comment, particularly with regard to the warfare system.

E.g. why does the AI invest so much in an early fleet it then doesn't use?

Can't think of even remotely how that would be relevant, no course out there is teaching Stellaris script. You don't get to vet members of the team either. That's pretty insulting.

I'm not vetting members of the team, I'm expressing genuine curiosity on a public forum. No malice is intended whatsoever. This is part of what makes engaging on these forums worthwhile, the opportunity to roll-back the curtain and find out how other people live their lives.

On the stellaris script education - if you are operating at the frontier, no taught course is relevant. The point of education is to teach people how to advance human knowledge and technology, not to dictate to them how they should do things. There are a bunch of cool projects out there of people figuring out stuff we don't know how to do, yet.

@Alfray Stryke was the one who brought up the question of trust, and to me, it is competence that matters, so I do think the technical question there is relevant and fair.

 

[Alfray Stryke]

one is to boost the AI to make them better, and another one is to somehow limit the exponential growth of the min-maxed player.

Yup, that's our plan. Improve the AI, rework things so stuff that's far too good is reigned in somewhat, buff things that might need it. That doesn't mean a perfectly symmetrical balance, some things will always be weaker than others to allow folks to choose easier or more challenging starts (Stellaris lacks the equivalent of playing a count in the HRE in CK3 or playing an OPM neighbouring the Ottomans in EU4 after all).

I'm not vetting members of the team, I'm expressing genuine curiosity on a public forum. No malice is intended whatsoever. This is part of what makes engaging on these forums worthwhile, the opportunity to roll-back the curtain and find out how other people live their lives.

On the stellaris script education - if you are operating at the frontier, no taught course is relevant. The point of education is to teach people how to advance human knowledge and technology, not to di

You don't need CompSci/Software Engineering training to work in QA - I'm not writing or reviewing code. I'm looking at how the AI performs in game, doing analysis of save files, verifying bug fixes, logging bug reports and more.

 

[FlyingPhoenix]

You don't need CompSci/Software Engineering training to work in QA - I'm not writing or reviewing code. I'm looking at how the AI performs in game, doing analysis of save files, verifying bug fixes, logging bug reports and more.

Thank you for the civil answer

 

[HistoricalScore1]

Just as a reminder folks:

• The balance changes take into account our vision for the game, not just what's currently "the meta".
• The AI is being worked on (I'm currently the primary QA looking at the AI, so hopefully you can trust me).

I think you know the drill of what the colors mean. Good to hear from PDS folk some more. I'm glad to hear you're primary QA for AI. Our vision gives me night terrors of whoever at Mass Effect decided to force that ending for ME3. I just pretend the game doesn't exist for the last few minutes, because dues ex Machina is terrible writing, according to everyone who took Writing 099. Tangent done. Back to "our vision".

This quote could be interpreted in many dangerous ways by some people. For example, one might accuse you guys of having the "vision" of Materialists completely overpowering Spiritualists. Because thats whats happening since around patch 2.2.7.

Lots of people would like to play Spiritualists but can't enjoy them because we know we are playing a faulty empire, especially if we decide against using Robots for pop growth.

So if your vision really is to keep Spiritualist as the laughing stock, then...

rip that Band-Aid off so we know what kinda company we're putting our faith in. Super well said!

If your argument is we don't want everything to be balanced, that's a decision. There's been talk about transparency and trust. If you're not going to let us know what you're thinking or planning, it's kinda like leading someone on in a relationship. @grekulf talked about expectations. Well, here's your opportunity. I know there are plenty of folks on here who have been hopeful for many, many years for better Spiritualists and options of other playstyles without feeling like a handicap. You can tell them. Or ride that hope until it's all gone.

Since @Enfield_PDX referenced scripting, I'd highly encourage you to get in touch with @slv or the #starnet-ai channel in the Modding Den. A lot of useful folks there on their free time who A. Have a lot of know-how on scripting (likely would not mind helping out PDX staff with questions). B. Identify frequently limitations on AI because of things being hardcoded or lacking script values/definitions (i.e. how the AI cannot resettle pops or colonize during certain wars). For instance, how long did you all know or when did you learn that the AI will not resettle pops from the Doomsday planet? You know what some people who test StarNet do? We just make different empires and see how the AI stacks up with a fast_forward.

Just because PDS folks don't reply to threads doesn't mean we're not reading them. I just wanted to redirect the conversation back onto what people thought was busted without being dragged down by "the AI can't compete with X, Y, Z".

The conversation is about Ringworlds not being the problem with tech boom. The AI setup as it is now is a problem. I think when said problem is resolved, acknowledged, roadmapped to be resolved, then people may stop complaining about it.

And you can have multiple viable ways to play the game. It's work, but it's not THAT hard. There's something called Difficulty settings. Back when great games were around, difficulty meant, the Bots (they didn't call them AI, because, they aren't actually artificial intelligence, it's just input, output (am I wrong?)) would be more challenging on the higher difficulties. For instance, you could have the AI Empires have less optimized empires on the barely above Ensign difficulties. Cadet keep it what it is, so anyone can easily beat it. On Ensign, have it somewhat optimized. On Captain, now have it more optimized (colonization timings, pops/jobs/resources/tech paths/choices/etc). Then on the highest difficulty, that's when the AI can cheat and get more resources. That's what great games like StarCraft did. It's doable. It's just a matter of if it's a developer decision or a corporate decision.

Because tech rush should not be king. It just shouldn't. It should be good. But if they are tech rushing and have little-to-no alloys saved up, or a fleet? It would be really easy for another empire to punish them. 10 spearmen will overpower 2rifleman. Now 5 riflemen? Yeah, they probably beat back 10 spearmen and then have an advantage. Now if an origin is going to be more challenging, I can see that. But nearly every ethic/civic should have real opportunity cost. Back to AI. A Fanatic Authoritarian should have a very, very hard time becoming even neutral with Fanatic Egal/Xenophile, because the Authoritarian (who maybe have Slaver Guild Civic) want to enslave them.

If you want a great grand strategy game, it starts with dynamic AI and dynamic empires that are meaningful (see meaningful decision making Game Making 101) and not just different options, but one is far superior, because..."because." This isn't Thief of Time and we're not students of Ms. Cosmopolite, if you know the reference. Saw Discworld on the Book of the Month. Hint in purple.

 

[Kaoru Sen'nin]

You've got a base total research output of ~2k/month (according to the top bar) in the 2320s, that's somewhat reasonable. What we're talking about is having that much research in the 2220s.

Edit: Also I didn't realise we had fans in South Africa!

Are you also looking at the issue that we run into repeatable tech very quickly even when not really trying to be tech focus ?

On another notes it seems that the AI often don't build any science lab and just go through the game with the one lab they have on capital planet since the start.

 

[Alfray Stryke]

If your argument is we don't want everything to be balanced, that's a decision. There's been talk about transparency and trust. If you're not going to let us know what you're thinking or planning, it's kinda like leading someone on in a relationship. @grekulf talked about expectations. Well, here's your opportunity. I know there are plenty of folks on here who have been hopeful for many, many years for better Spiritualists and options of other playstyles without feeling like a handicap. You can tell them. Or ride that hope until it's all gone.

I'm QA and I've only been on the project for a year, I don't decide what the game vision should be.

Since @Enfield_PDX referenced scripting, I'd highly encourage you to get in touch with @slv or the #starnet-ai channel in the Modding Den. A lot of useful folks there on their free time who A. Have a lot of know-how on scripting (likely would not mind helping out PDX staff with questions). B. Identify frequently limitations on AI because of things being hardcoded or lacking script values/definitions (i.e. how the AI cannot resettle pops or colonize during certain wars). For instance, how long did you all know or when did you learn that the AI will not resettle pops from the Doomsday planet? You know what some people who test StarNet do? We just make different empires and see how the AI stacks up with a fast_forward.

We've got some good programmers and scripters on our team and I trust in my colleagues skills to do their jobs. Likewise, you don't need to explain to us how we should be testing the game.

The conversation is about Ringworlds not being the problem with tech boom. The AI setup as it is now is a problem. I think when said problem is resolved, acknowledged, roadmapped to be resolved, then people may stop complaining about it.

The Shattered Ring origin, as it currently stands, is capable of achieving a tech rush at a higher rate than we thought was appropriate, hence why it was mentioned at being looked at.

I have stated multiple times in this thread alone that we're working on AI improvements. Unfortunately, I can't give more information than that because I don't want to promise something that we don't meet if the testing discovers further issues with what we're doing.

 

[Apollo1784]

You've got a base total research output of ~2k/month (according to the top bar) in the 2320s, that's somewhat reasonable. What we're talking about is having that much research in the 2220s.

Edit: Also I didn't realise we had fans in South Africa!

Then how about we bring the AI up to par instead of nerfing ringowrlds? You know so it can actually compete with players