Why is it so hard to generate science in the beta version?

[BaliorPal]

Maybe I’m doing something wrong, but it feels like one unit of science now costs as much to produce as one unit of alloys. I have a planet fully built for science that gives me around 700 science, but when I restructured it for alloy production, I was getting almost 1000 alloys per month.

 

[HFY]

3.0 researchers generated 12 science (4 of each color).

3.14 researchers generated 9 science (3 of each color).

Beta 4.0 researchers generate 4 science (of a single color each).

Also you get fewer research jobs per district (90 instead of 100 for other jobs).

So basically you need three times as many pops to get the same research output you had in previous versions, or a bit over two times to match the most recent current release.

 

[Newbee53]

The mechanics of it have been pointed out. But the reason is more mundane.

It's probably a bit too hard to produce science but given how important science is its way to easy to generate in 3.14. I should point out that the beta hadn't finished balancing so things might be balanced to slightly more science than the beta currently stands. I do think that it's been the intention to reduce the rate at which one sciences for a while now.

Balance in games seems to be a process of slightly less extreme over corrections until you get close to the target. Plus, every so often you go just way too far. So I expect researchers will get tuned upwards just a little in the future. But hopefully not enough to come back in line with 3.14 pacing.

 

[BaliorPal]

The mechanics of it have been pointed out. But the reason is more mundane.

It's probably a bit too hard to produce science but given how important science is its way to easy to generate in 3.14. I should point out that the beta hadn't finished balancing so things might be balanced to slightly more science than the beta currently stands. I do think that it's been the intention to reduce the rate at which one sciences for a while now.

Balance in games seems to be a process of slightly less extreme over corrections until you get close to the target. Plus, every so often you go just way too far. So I expect researchers will get tuned upwards just a little in the future. But hopefully not enough to come back in line with 3.14 pacing.

So the games are going to be even longer, even slower, and even more boring. Great.

 

[Newbee53]

So the games are going to be even longer, even slower, and even more boring. Great.

Don't know about more boring, but its kind of silly to be able to tech rush into endgame tech long before reaching the end game content. And I've done that purely by accident in some 3.x balance, which is even sillier. Like I said, I think they went too far with the beta's balance. But I fully expect it to be a little slower than 3.14. Which isn't a bad thing at all.

 

[BaliorPal]

Don't know about more boring, but its kind of silly to be able to tech rush into endgame tech long before reaching the end game content. And I've done that purely by accident in some 3.x balance, which is even sillier. Like I said, I think they went too far with the beta's balance. But I fully expect it to be a little slower than 3.14. Which isn't a bad thing at all.

What difficulty are you playing on? The game usually starts to slow down around the year 2300 — I set the endgame date to either 2300 or 2325.

 

[Newbee53]

What difficulty are you playing on? The game usually starts to slow down around the year 2300 — I set the endgame date to either 2300 or 2325.

default settings. Because you balance the game first and foremost on default and then try if you have time and energy to balance other common settings. mid game at 2300 and end game at 2400 and the score victory at 2500.

 

[BaliorPal]

default settings. Because you balance the game first and foremost on default and then try if you have time and energy to balance other common settings. mid game at 2300 and end game at 2400 and the score victory at 2500.

Then first of all, the developers need to improve the performance of this crumbling code. I have a Ryzen 5600H, and the game starts slowing down after the year 2300. On top of that, multiplayer lobbies start disconnecting around 2280–2290, and by 2330 we get disconnected every 5 years. Galaxy size 400

 

[BaliorPal]

Another problem is that on Grand Admiral difficulty, I usually catch up with the AI by around 2280, and by 2320 they don’t stand a chance against a near-meta setup with missiles and strike craft. Slowing down science production will stretch out the game, but it won’t really change the point at which the AI turns into punching bags.


So for the remaining 100 years before the crisis arrives, I’m either stuck with slowly conquering the entire galaxy, or just continuing what I was already doing for the past 100 years — developing planets and spamming orbital habitats. But paradoxically, at that stage there’s actually less micro involved due to the scale of the economy. The resource surplus lets you build out an entire planet for the next five years in one go and not touch it again. Resource management and priority-setting degrade significantly — and that’s exactly why building stuff becomes boring, even though the mechanics are technically the same. The context of your empire's economy has shifted, and what used to be careful decision-making and tight deficit balancing has turned into a massive surplus of absolutely everything.

 

[Newbee53]

Then first of all, the developers need to improve the performance of this crumbling code. I have a Ryzen 5600H, and the game starts slowing down after the year 2300. On top of that, multiplayer lobbies start disconnecting around 2280–2290, and by 2330 we get disconnected every 5 years. Galaxy size 400

One of the big reasons for the changes to how pops work is to improve end game performance. No word yet on how it's done, but from what they've said it should make it much easier to Parrell pop calculations. I tend to play on a large galaxy, but that might just be the large ram talking. I've never really cared about multiplayer, and in my experience I've never enjoyed competitive multiplayer in games even remotely similar to this. So I can't speak on any of that.

Another problem is that on Grand Admiral difficulty, I usually catch up with the AI by around 2280, and by 2320 they don’t stand a chance against a near-meta setup with missiles and strike craft. Slowing down science production will stretch out the game, but it won’t really change the point at which the AI turns into punching bags.

Another difference. I've never built a meta fleet intentionally and don't plan to. So, I can't talk about that balance. But one of the advantages of slowing research compared to the current live version is to control the player's snowball stretching out the time before the AI fails to compete. I don't know how much it will help.

The stream today had a throw away comment where they said that the balance on researchers was rather terrible in the beta and they'd fixed that internally now. they said it was 'closer' to 3.14 which suggests that they are continuing to adjust the strength of researchers to better balance the pacing of the game.

 

[Falrinn]

The stream today had a throw away comment where they said that the balance on researchers was rather terrible in the beta and they'd fixed that internally now. they said it was 'closer' to 3.14 which suggests that they are continuing to adjust the strength of researchers to better balance the pacing of the game.

Yeah, as I recall the baseline for researchers in 4.0 will probably still be lower than it is in 3.14, but not nearly as low as it is in the open beta.

My guess is that the baseline rate is lower because there are more tools to specialize and improve research. It's going to be a lot easier to cram a whole bunch of physicists on a world with a physics research bonus.

 

[cerberus189]

Yeah, as I recall the baseline for researchers in 4.0 will probably still be lower than it is in 3.14, but not nearly as low as it is in the open beta.

My guess is that the baseline rate is lower because there are more tools to specialize and improve research. It's going to be a lot easier to cram a whole bunch of physicists on a world with a physics research bonus.

The tech is actually a lot better now. I was the one that said in the stream about the research being so bad that you needed a size 30 world to match a size 10 in the current release. when they went in to the jobs tab later, they showed that 120 researchers was making 8 engineering research, double what is in the current beta patch and more then the current release, which has a base of 3 tech per category

 

[Lorenerd11]

If tech costs are a problem, you can easily reduce them in the settings.

 

[Newbee53]

The tech is actually a lot better now. I was the one that said in the stream about the research being so bad that you needed a size 30 world to match a size 10 in the current release. when they went in to the jobs tab later, they showed that 120 researchers was making 8 engineering research, double what is in the current beta patch and more then the current release, which has a base of 3 tech per category

I'm wondering if that was before or after they cheated all the tech. I don't remember when that happened.

I'm a little disappointed if it was before. It feels far too easy to tech rush in 3.14. Though I guess there might be other constraints that make tech progress more of a reasonable pace.

 

[Abdulijubjub]

The tech is actually a lot better now. I was the one that said in the stream about the research being so bad that you needed a size 30 world to match a size 10 in the current release. when they went in to the jobs tab later, they showed that 120 researchers was making 8 engineering research, double what is in the current beta patch and more then the current release, which has a base of 3 tech per category

Basic math time.

120 researchers making 8 engineering research means 6.6 research per 100 (which the equivalent of 1 old pop). Yes, an old researcher made 3 of each type, and 6.6 > 3. But an old researcher made 3 of each type, which means they made a total of 9. Just with those numbers, a new researcher makes 27% less than an old researcher, and that's before accounting for the fact that the new one's 6.6 has output modifiers baked in, while 3/3/3 is the base numbers, before modifiers, for the old one.

The stream definitely didn't show them being more than the current release.

---

That said: They may give a hefty bump between now and then. And even what they've shown may add up to being about the same:

• They may have already bumped the base from 4; I didn't see any direct indication of a higher base, but the bonus at the start should be around ~10-20%, which would make 4 base a bit implausible.
• The stream also showed a bunch of buildings that added +1 to each type (and the beta already has a bunch that add more, like the institute).
• The new multipliers are higher by default: the new efficiency and output modifiers are multiplicative, therefore are more powerful when stacked than they would have been before. e.g. +100% efficiency and +100% output gives 4x normal output, while two +100% output effects give only 3x.

So it may go back up to the current baseline by the time they've gotten everything implemented (or beyond it). We won't know till it comes out. But I'm just saying: 120 making 8 is not more than the current.

 

[Critical Ethics]

Component A: One of the issues with the Stellaris tech system is that you can't directly deep dive to get to a specific end game tech. If I want to do a megastructure build I can't deliberately research all the way up to megastructures while ignoring side-chains, I have to come at it sideways using various weighting systems that still rely on random chance.

With a more traditional tree it's much easier to get the sweet zone where a slow researching empire can still get the specific end-game content they want while a fast-researching empire just gets more end-game techs and a bit faster. Since launch Stellaris had needed every empire to get every tech to make sure they get the techs they actually want or need. This is a big part of why empires start feeling pretty samey, because one of the standard 4x playthrough differentiators just can't be used for that.

Over the past bit though they've been adding in more ways to get keystone techs, especially with the new focuses and the much more frequent techs from core events. If you're able to get all the techs you want or need as a low research empire without researching all the techs then you no longer need every empire to be able to research all the techs so it makes sense to slow down the tech to get the playthrough differentiation I mentioned earlier.

Component B: In the buildings and districts model an early tech planet was limited to 11 research buildings of 2 scientists each and those science buildings were competing with amenity buildings, civic buildings, and so on. With the new pop growth model it's harder to get new pops to fill new jobs but with the new district/zone/building system it's much easier to make some of those jobs be new scientists - even your starting homeworld has 2.4 scientists compared to the 3.14 starting 2, and just upgrading your city district 2 or 3 times will get you as many scientists as a whole second building would have before. So scientist output needs to go down to compensate or everyone will be working off the old broken void dweller science economy.

E: Component C: And as @Abdulijubjub said, there's a bunch of previously additive multipliers that are now fully multiplicative.

Combining all these you're looking at a pretty hefty drop in base per-scientist output. They probably went too hard with a full drop to 4 from 9(12) and a lower pop per district count, especially with all the booster buildings locked behind t2+ tech, but I'd say that's the basic reasons.

 

[Edorath]

So the games are going to be even longer, even slower, and even more boring. Great.

People like you is why the beta should have had a giant NUMBERS NOT FINAL event popup appear every five seconds.

 

[Shadeseraph]

I've said it before, but tech in the beta is non linear. The research institute and the advanced laboratory give a huge boost to science and you should rush it asap. That said, 3.99.7 actually worked quite well and felt rather balanced last time I tested it, even early game.

 

[SolSys]

No word yet on how it's done

They talked about it here

Stellaris Dev Diary #370 - 4.0 Changes Part 4

Hello everyone! This week we’re going to look at the upcoming changes to Pops in the Stellaris 4.0 ‘Phoenix’ update. Last week I said we might also look at the Planet UI, but I’m going to save that until next week since there’s quite a bit to...

forum.paradoxplaza.com

Example: Then vs. Now​

Before (3.14):

• Take a planet with 100 Pops working Metallurgist Jobs, where 20 of them have a +10% Production Bonus from a Species Trait.
• These 100 Pops produce 612 Alloys per month.
• Every Pop is individually checked - 80 produce the standard amount, while 20 get a 10% Alloy production bonus from their species trait.

Now (4.0):

• Instead of tracking individual Pops, we track Workforce filling Jobs.
• The Jobs are now filled by 10,000 Workforce (since Pops are scaled up by 100).
• 8,000 Workforce comes from regular Pops, while 2,000 Workforce comes from the bonus-earning Pops.
  • The species bonus is now “10% bonus Workforce when working Alloy jobs” - those Pops contribute an extra 200 Workforce, making the total 10,200 Workforce. Bonus Workforce is allowed to go over the required Workforce for a job, yielding extra production.
• If 100 Workforce still produces 6 Alloys, the planet still produces 612 Alloys - same output, different system.

Why This Matters:​

The key benefit is efficiency. Instead of iterating through and calculating production for every individual Pop, the game now only checks once per planet. This makes the system more scalable and improves performance, while still allowing for species based bonuses and modifiers.

 

[HFY]

The key benefit is efficiency. Instead of iterating through and calculating production for every individual Pop, the game now only checks once per planet. This makes the system more scalable and improves performance, while still allowing for species based bonuses and modifiers.

Once per pop-group per planet, not once per planet.

The assumption is that there will be fewer 4.0 pop-groups than 3.x pops, but that's an untested assumption.

Given how 3.x AIs routinely generate 10+ versions of each species, in a galaxy with 20 species there could easily be 200 sub-species, times 4 ascensions, times 8 ethics, times 3 stratum (or 4 stratum). That sums up to more possible combos on one colony in 4.0 than I have individual pops in my whole empire most 3.x games (19,200).

And 20 starting species is not a lot.