Research and science vs. historical accuracy

[Infixo]

[It’s a really long one]

What's this about?
In my first Civ6 game I got so surprised that I've finished entire Tech Tree somewhere around 19th century. And I wasn't even trying. In my 2nd game I progressed even faster, getting units waaaay too early. For me it totally breaks immersion and takes a lot fun from the game. I also watched many Let’s Plays on YT and it seems that everyone has the same experience. I know that there are mods that extend Eras (8 Ages of Pace, Eras Times X, etc.) but I don’t think that just extending Eras blindly solves the problem. I did some analysis and I would like to share results and thoughts.

Goals
I wanted to tweak/ adjust/ change the Tech Tree and science progress in general to achieve two simple goals:

1. To progress through Eras with historical accuracy as much as possible – concerning timing and development of previous techs required to research e.g. more advanced weaponry.
2. To be able to beeline 1 Era ahead but with only 1-2 selected techs; beelines should be balanced, so no matter how you enter en Era, it should always take the same amount of research.

Historical accuracy – timing
One would ask what this actually means in Civ6. Well, the designers did actually pretty good job here and took the following assumptions:

Era           Time          Turns* Sci/trn
Ancient       4000BC– 1000BC   75     7,5
Classical     1000BC–  500AD   60     21
Medieval       500  – 1350     60     40
Renaissance   1350  – 1725     60     89
Industrial    1725  – 1890     60    121
Modern        1890  – 1945     50    168
Atomic        1945  – 1995     50    236
Information   1995  – 2030     45    390
Future        2030  – 2050     40    n/a

(*) On standard speed

As you can see there some minor changes in timing:

• Medieval era lasted till 1500 AD
• Industrial is usually associated with industrial revolution in England which started in 1750
• Modern with start of WWI (1914)
• Information – various sources place it somewhere around 70’s

I don’t think these are important – it is clear that the goal here was to make last Eras approximately the same number of turns. I.e. changing Modern to 1915-45 would make it only last 30 turns, same with Atomic, etc. I say we can stick to the timeline designed in the game. It’s not bad considering turns.

Historical accuracy – pace
The last column in the table shows how much Research you need to progress though Eras as designed in the game. This doesn’t take into consideration Eurekas – just plane Tech cost. These values are not very demanding. Usually 2-3 Campuses are enough to go past them. And here lies the problem – everyone usually builds more of them. Especially in the late game, especially when going for science victory (obvious) or domination (you simply have too many Campuses – your own and from other civs). I call this effect “science crazy” – Research output usually sky-rockets after Modern Era. Add to this Eurekas, which are huuuge and Science gets broken.

Ok, there are a few ways to deal with it:

1. Increase Tech cost so it will match the players behaviors.
2. Decrease Eurekas – it will make progress more stable.
3. Decrease Research output through yields decrease.
4. Increase Tech cost to match player’s increased Research output.

I decided to go with no. 1 and 2 approach, mostly. As for now people are used to new science-related yields and those are in line with other yields in the game, so I don’t think no. 3 is a good way.

Ad. 2. Eurekas are the easiest to fix. I believe setting them at 35% would give much better effects. This value is still worth going for it and it also presents itself nicely on the screen (it’s approx.. 1/3 of the circle).

No 1. is the most interesting one. As I mentioned at the beginning – there are mods that increase cost of Technologies. But the real question is: how much?

I have analyzed several games (mine and from YT) and generally the Science output follows the formula: 0,001*x^2,1+0,02*x+3. Of course the parameters vary for each game but I needed to make some assumptions and the closer to reality they are – the better. These games were played on King/Emperor difficulty and players usually went for a mixture of domination and science, building a total of 4-5 campuses. I’d say these are good guidelines.

I also assumed that 50% of Eurekas will be triggered in first 2 Eras, then 40% in the next 2 and 30% for the rest of the game. I know that one can get much more – but it’s ok – there needs to be some incentives. More important is that the model assumes that many Eurekas will be triggered.

The results are:

Era         Inc% Eurs Science Simulated
Ancient      10%  50%     513     490
Classical    30%  50%    1886    1882
Medieval     55%  40%    5045    5015
Renaissance  25%  40%   10785   10721
Industrial   40%  30%   19882   19855
Modern       45%  30%   30809   30720
Atomic       35%  30%   45054   45091
Information   5%  30%   61561   61457
Future       60%   0%   65561   79059

The second column shows for how much each Era’s tech cost should increase to achieve historical accuracy.

Historical accuracy – beelining
I have added new connections to the Tech Tree to patch many omissions and to make it more logical. This was done also to prevent beeling 2 Eras ahead or generally research things without proper historically sound background.

Examples:

• Need Masonry to build a Great Lighthous
• Need some kind of furnace to produce Iron
• Workshops need rather Iron and various Constructions to operate than Horses
• Medieval walls and castles were quite complex engineering works and those Knights have to park somewhere
• Machinery needs calculations
• Bombards too fast to get, cannot produce them without machinery
• Need more knowledge to handle gunpowder
• No advanced science without printing
• Coal is crucial for steel production
• Computers require electronic parts which were developed along with i.e. radio
• Ballistics is crucial for launching rockets
• Satellites more need Ballistics than airports and planes
• Must complete entire tree to proceed with Future Tech

There are many more. The original Tree has only 89 dependencies. This also explains why beelining is so broken now. I changed the tree and there’s now 148 dependencies. All techs in one Era require more or less similar number of prerequisite techs so no matter how you enter en Era – it always take similar number of techs to develop. You can beeline to 1 Era ahead, no more.

Last but not least
Unfortunately – the above doesn’t fully solve the problem. You can still build 10 or 20 Campuses. And get like 1000 beakers per turn. And that’s why after all this deliberation I genuinely think that the mechanics from Civ5 where the Tech cost increased when new cities were founded should be brought back. This time however it should be connected to Campuses however, so the main goals of Civ6 (make wide a more viable gameplay and cities specialization) will still be fulfilled.

There can be a rationale for this – increased complexity of research network in the civilization causes ineffectiveness of research. Let’s say – each new connection between Campuses is 8% overhead on communication, ideas exchange, bureaucracy, etc. This is enough to make 6 Campuses operate like 5 and 9 like 6, etc. This would make inefficient to make them more than 6-7 in general. This, combined with decrease in Science yield from Population (0,5 instead of 0,7) should totally balance the game in this area.

How to implement the above
It’s very simple actually. I’ve created a mod that changes Tech Tree and decreases Eurekas (called it Real Tech Tree).

Edit, added 1.03.2017.
I managed to create a mod that does the most important thing - links empire size to research speed. In Civ6 you cannot change the tech costs in the game, but you can decrease yields output. So, the below mod introduces this mechanism: each Campus (except the first) you built will decrease the science output from scientific buildings: Real Science Pace.

Well, that’s all. Eagerly awaiting for comments and discussion.

 

[Victoria]

I think the answer to this and a few other issues is refuce the science per citizen from 0.8 to 0.5... Or even 0.3
I agree beelining is a bit broken too

If you spent as long on your mod as this post its bound to be good. Sadly I'm vanilla

 

[Infixo]

I updated the tech tree and it's much better. But the main change i.e. increasing tech cost with Campuses is more difficult. As for now I can't find any API function to change the cost of the technology for one player. There's a general tree, same for all players and an object that governs the research. But the object only holds techs that has been ever reasearched, not the future ones.
Changing science per citizen is easy and also agree that should be lower. But as I discussed in my post - will not solve the problem. You can build 10 or 20 or 30 campuses and the tech cost is still the same.

 

[Jonathan]

I never finish the tech tree because I set my games to end after 220 turns. Problem solved.

And that wasn't even the problem I was trying to solve. I just find the later stages of the game tedious.

 

[God of Kings]

The same could be done with civics as well.

Did Johann Sebastian Bach uploaded his music on Soundcloud?

Was Shakespeare one of the most prolific Twitter users?

Did Michelangelo spent time Photoshopping?

Was Leonardo da Vinci busy editing Wikipedia articles?

Was Girolamo Savonarola concerned about Pokémon?

 

[plus]

i use the slow eras mod, i can than slow down time whenever i want...best thing about civ6 is that it's not limited like civ5 was regarding beakers cost..imo that results in great game,where actually u get to long wars and use ALL the untis the game has to offer..i know it's a bit weird, but for me the pace is about 80 turns for each tech..add to that the more buildings mod and u never run out of things to do.also i reduced the eurekas bonus to nothing really to avoid imbalance .

 

[Infixo]

Well, I was looking for a solution that would actually fit to game's grand design. In general, standard 500 turns is not good for war-style game. So, if you want to have more time to play with units, you should play Epic or even Marathon. And within this 500/750 turns you are supposed to discover all techs, in a pace that is somehow aligned with human history. That's a grand design for me. So, it shoudn't be possible in game to have planes in 17c, or computers in 18c. Or the other way around - playing with swords in 18c also is out-of-balance.
Extending eras, no matter by what factor, solves the problem only partially, sometimes introducing new issues, because it doesn't address the root cause - that the science cost is constant for a given tech throughout a game.

 

[plus]

Well, I was looking for a solution that would actually fit to game's grand design. In general, standard 500 turns is not good for war-style game. So, if you want to have more time to play with units, you should play Epic or even Marathon. And within this 500/750 turns you are supposed to discover all techs, in a pace that is somehow aligned with human history. That's a grand design for me. So, it shoudn't be possible in game to have planes in 17c, or computers in 18c. Or the other way around - playing with swords in 18c also is out-of-balance.
Extending eras, no matter by what factor, solves the problem only partially, sometimes introducing new issues, because it doesn't address the root cause - that the science cost is constant for a given tech throughout a game.

well, i dont really care about what century it is at all.imo game has its own time, so if we still fight with swordsman in the 20 century that's fine by me
man, if u could make a mod that just lowers the sciense output per citizen + disabling any eureka bonus than it will be great.

 

[Infixo]

man, if u could make a mod that just lowers the sciense output per citizen + disabling any eureka bonus than it will be great.

Well, it's actually 2 lines of SQL:

UPDATE Boosts SET Boost = 0;
UPDATE GlobalParameters SET Value = '30' WHERE Name = 'SCIENCE_PERCENTAGE_YIELD_PER_POP';

Sometimes you don't need much to make people happy

 

[plus]

Well, it's actually 2 lines of SQL:

UPDATE Boosts SET Boost = 0;
UPDATE GlobalParameters SET Value = '30' WHERE Name = 'SCIENCE_PERCENTAGE_YIELD_PER_POP';

Sometimes you don't need much to make people happy

that's great man

 

[plus]

that's great man

i need one more favor
i want to make it that each district i build ,from each type, it's hammers cost will increase .
remember in civ5 there was a mechanism that increase cost the more cities u have ? something like that.
so than,if i build my first science district ,it will cost lets say 20 hammers, the next one in second city will be 5% more, the third one 10 % more etc, for each type of district.
any idea?

 

[Gehennas]

I actually found the "Endless Legend" approach interesting: there's no tech tree. Instead you have 10+ techs for each era and you can't research techs from the N-th era until you get at least (N-1)x8 techs from the previous ones. Doing something similar in Civ6 may restrict beelining a little bit. Tech tree can be left as it is. Instead there'll be tech requirements before advancing to the next era.

 

[dunkleosteus]

I am really happy that you're looking into this. However, the Dev's timeline is actually more accurate than the notes you made below it

1350 is listed as the end of the medieval era because of the black plague which occurred (although the exact dates are disputed) 1348-1352. About 1/3 of Europe's population died in this short period. Prior to this time, peasants were treated like a commodity: tied to the land, they lived, worked and died for the owner of the land they worked on. After the plague, the vastly reduced population made peasants much more valuable, which shifted leverage towards the peasats: it gave them more social and political power. Ultimately, and with the introduction of gunpowder around 1400, the medieval age as it is recognized as the culture that existed in Europe since the fall of the Western Roman empire c. 500 ended. The feudal economy and social hierarchy was immobile with military power for kings supplied by Lords that contracted Knights that owned land. Knights were often the second sons of Lords without an inheritance. They spent at least 14 years of training starting at age 7 before they became qualified as knights. They purchased their own horses, equipment and armour.

Guns completely changed the dynamic- a peasant can be trained to operate a gun reasonably well in about 3 months. Guns suddenly enabled an uneducated, untrained peasant to destroy a knight instantly, and this upset the social order heavily.

The plague heavily disturbed the social institutions and this was exacerbated by the introduction of gunpowder. Scholars agree that 1350 is really the transitional phase between medieval and renaissance because the social and economic changes were so rapid and wide spread.

While the industrial revolution may be cited as 1750, we can agree that without an enlightenment period between renaissance and industrial, it is more reasonable to start the industrial era earlier than to have the renaissance later.

The modern era definitetly started in the late 1800s, not 1914. With lightbulbs, electricity, telephones and the telegraph, all the hallmarks of modernity were present: rather than taking days or weeks to communicate with other cities or countries, it could be done instantly. This is a stark contrast to the industrial era which featured rapid economic and social growth but still lacked the ability to communicate quickly: it could take weeks to learn about foreign events.

The above is nearly direct from professors of history courses I have taken. For the information era, I think I can reference my own experiences as a computer scientist.

The information era largely represents the adoption of the internet. Sure, universities and research institutions began incorporating small networks as early as the 1970s, but this had little social, economic or cultural impact before the mid 90s. Additionally, I feel like the devs heavily associated the atomic era with the cold war, which ended with the fall of the USSR in 1991.

 

[dunkleosteus]

You have an interesting point that each campus should increase tech cost but I disagree slightly: does this harm partial research towards techs? (people often half research a tech and wait for the eurekas).

I think tech scaling should be handled like this: campuses should produce more science each as the game progresses.

Instead of increasing tech cost by the number of campuses, I think each additional campus should be penalized. So if you have a campus that generates 10 science per turn and another that generates 8 science. It is reasonable that at least some of it would be redundant: we assume that science is always unique in a single campus. So the 2 extra science that the first campus has compared to the second is definitely unique (otherwise it would not be there). However the second campus's science may be the same science as the first. (this is difficult to put into words).

I guess imagine you have 50 people split into groups of 10. Each group tries to brainstorm ideas to solve a problem. The first group comes up with 10 ideas. The second group produces 8, two come up with 7 and one comes up with 5.

How many DIFFERENT ideas do you have? 37? Almost certainly not. It's way more likely that a lot of the ideas overlapped with at least one other group. Lets say that we've come up with 20 unique ideas after talking to the first 4 groups. The last group has 5 ideas. What are the chances they have new ideas? Obviously, the chances are lower than if we only had 15 total ideas. We would also be more likely to have a new unique idea if the last group had come up with more ideas.

In this analogy, each group is a campus and the science they generate is the number of ideas they come up with. It would be better to make each group come up with more ideas than to just add more groups that come up with the same ideas.

Does that make sense?

Basically, the more science you have in total, the less science each university gives you.

Imagine that you calculate science for each university in order of how much science they would produce. The adjusted science would be (for example) science * science/ total science.

From the above brainstorming example, the first univeristy would generate 10. The next would generate 8 * 8/10 = 6.4 science. Total: 16.4

The next was 7, so 7 * 7/16.4 = 2.99. Total: 19.4

The next was also 7, so 7* 7/19.4 = 2.52. Total: 21.92.

Final: 5, so 5 * 5 / 21.92 = 1.14. Total: 23.06.

So the normal science output would have been 37 but if we use this example formula, we'd only have 23.06 per turn.

As you can see, it is way better to try to improve the universities you have than just spam more of them. Quality over quantity.

Also, because there are a lot of great scientists that give a flat amount of science, its important that science costs dont change.

 

[dunkleosteus]

As a response to my previous post, a better calculation for science might be

(where x = net adjust science for this campus, y = base science for this campus, and z = total science)

x = y^2 / (z + y)

I think this way it doesn't matter what order you look at the campuses in and it works correctly in all situations. Your first campus always gives 100% of its science and the next campuses give a fraction. Its better if you dont have to calculate in order though. (the end result is different with this formula though)

we'll do a simple example with a campus that has +10 and one that has + 6 per turn.

If we start with the 10, we have a total of 10 science before we look at the other campus.

x = 6^2 / (6+10) = 36 / 16 = 9 / 4 = 2.25

Our total science would be 12.25 from both campuses.

If we start with 6 science, our total is 6 first.

x = 10^2 / (6 + 10) = 100 / 16 = 25/4 = 6.25

our total is still 12.25.