I want to latch on to this thread because it has some great discussions on the balance of the game.
Disclaimer: I'm seeing this from the outside looking in, as I haven't bough the game yet - I invested a lot of time into GC3 so I'm watching some playthroughs like Quill and DasTactic to make up my mind.
That being what it is, there seem to be some fairly obvious balance issues with the system the game uses to deal with bonusses, because they are all (save 1) additive. Because almost everything is implemented as a bonus, almost everything is affected: difficulty settings, population, buildings, relics, (some) tech, leaders.
* Tech speed: Lets say you set your tech speed low (-75%). What does that mean? Well, techs will cost 4x as much, so research will go at 1/4 speed right? Wrong! It gives you an additive -75% modifier to your research. But what does that mean?
If your species is +2 intelligence (+10%) and you have a smart governor (lets say +15%) those bonusses just get added on top of the -75%. It's 100%-75%+15%+10%=50% - your planet is working towards new techs at 1/2 base rate.
But if your species is -2 int (-30%) then your base science rate is -5%. You produce negative science! I'm actually presuming this is capped, and won't actually happen, but who knows?
Further in the game, when you have a tech world with +200% research, that -75% modifier slows your research by 75%/300% = 1/4.
Anyway, the point is that these sliders have wildly inconsistent effects because they are *added* to the pile of bonusses, instead of multiplying.
* Population vs Buildings: Both pops and facilities give a +x% boost to the output of your worlds. Population gives a bigger bonus than buildings (until high upgrade levels), but its also far more limited because at higher population levels growth basically stops so you can't replace all production buildings with houses. On the other hand... you can totally neglect population.
You don't need any population to run your homeworld's 30 buildings anyway. You can have 30 science regions and no scientists. You can also have 30 scientist and nowhere for them to actually work. It doesn't matter, you get roughly the same bonus either way, but buildings are easier to manage, and as we'll see in the next part, lower pop is way better!
* Approval: The outlier! Happiness doesn't stack additively, it multiplies. But it doesn't stack as some kind of bonus or malus at very high or very low levels. No, it's a direct multiplication of % approval to resource output. A world with 250% manufacturing bonus at 50% happiness only produces half as much as it could. Same for science.
That means: anything that improves your happiness is *more important* than any other improvement/policy. The science ideology has a +10 happiness bonus to scientists option. If you turn all your workers into scientists, they'll be happier - scientists are +10 happier anyway. Your *manufacturing* will be much higher with 5 happy scientists than 5 ho-hum laborers, because the happiness bonus is multiplicative over everything.
5 resources +25% bonus manufacturing from labour x 0.5 happiness = 3.125 manufacturing
5 resources +0 bonus manufacturing from scientists x 0.7 happiness = 3.5 manufacturing. Plus you get extra science for free. Actually, it's more like tripple-dipping - the improved happiness also boosts the science output.
This can also be easily gamed. Keep your population minimal, and imprison everyone but 1 low-expectation pop - so you can easily max the average approval (prisoners simply don't count) of that one dude.
* Raw input is king: Beyond even happiness, raw input is what matters. No amount of additive buffing will let a world with 3 raw resources equal one with 12. Even with a fully geared up +300% bonus size 30 paradise is going to be inferior to a size 5 world with just a single manufacturing improvement and citizen. Buildings become weaker (relatively) the more you add. This also means that tile abilities are almost totally pointless. I don't care that this tile makes research go from +3% to +4% on a world with 3 base science and 21 base manufacturing. It's gonna be a factory.
An important consequence: avoid making core worlds. Say you have 2 worlds with 12 manufacturing each. If you turn them both into core worlds, and build up +200% manufacturing, they'll produce less than if you'd made 1 of them a core world, and the other a colony feeding into that world. First, concentrating 2x the base resources can build up those early factories faster, getting the ball rolling. You can concentrate your one-per-empire buff buildings, you don't need to spend strategic resources on improving both worlds (removing pollution, special buildings) etc. You don't need to feed populations on both worlds. You don't even need to micromanage them.
In essence: nothing you do in the game has as much influence as the RNG of base resources of worlds you happen to come across. So better rush those colonies and hope that the factions in other sectors didn't get way better ones...
* Conclusion: The way stacking of bonusses is handled makes everything... wonky. A 10% bonus doesn't mean what you intuitively think it means: a +10% science building could double the research output of a dumb drengin on slow research settings but is near pointless to build for a smart navigator on fast science settings - In both cases, approval buildings probably help more, and you'd better make sure to have as few citizens and as few core worlds as possible or you're really gimping yourself.
I really enjoyed analyzing these systems from what I could see of them. Some of these are easy to fix (like happiness), some less so. It's not really enthusing me about the game though, unfortunately - obviously opinions on the importance of this kind of thing may differ, but I think this should really have been recognized as broken a lot earlier in development.