Thoughts on A/B/C bonuses

[Cokolwiek]

It's really amazing the difference between GL completion time and Terracotta time. Imagine how different the Classical game would feel if they were reversed, everyone would be messing around with GL optimal free tech timings.

There are more reasons to that than AI preferences or A/B/C bonuses but also the fact that it (along with great lighthouse which is the earliest) taken faster than terracotta. It's favored by tradition empires which currently have much stronger early games, and more science to research them, more culture to unlock them, and more production and population in the capital to build them. So the game mechanism itself tweaks it like that.

 

[Gidoza]

There are more reasons to that than AI preferences or A/B/C bonuses but also the fact that it (along with great lighthouse which is the earliest) taken faster than terracotta. It's favored by tradition empires which currently have much stronger early games, and more science to research them, more culture to unlock them, and more production and population in the capital to build them. So the game mechanism itself tweaks it like that.

I like this comment of yours, and I had to think about it for a bit.

I mean obviously the word *favoured* is important here. That is - for Terracotta Army for example, I can say that that's favoured by Authority because it synergizes with Authority's other bonuses.

My worry though is that Great Library is generally too inaccessible with its culture demand of 4 policies and becomes like a pseudo-policy-tree-completion wonder for Tradition. What do I mean by that? I mean basically all the other Wonders, except possibly Hanging Gardens but not even, can essentially be built by any policy tree as soon as you get the tech for it, because the culture from any tree will match up pretty well with the tech pace. Great Library is out of sync with that pattern. Moreover, I can think of perfectly good reasons for lots of different wonders to be built by different contenders. For example...

1. Terracotta Army
a. Authority - Synthesizes with bonuses.
b. Progress - Good for later expansion and deprives others of a bonus.
c. Tradition - That extra military count is important.

2. Temple of Artemis
a. Authority - It works if you're tending towards producing ranged units.
b. Progress - General growth in a large empire.
c. Tradition - Helps the Capital even more.

Statue of Zeus I might avoid as Tradition though, whereas I'd veer away from Parthenon as Authority other than to deprive other players...though I'd probably not be on that side of the tree with Authority anyways. The point is that realistically for nearly all of the Ancient/Classical Wonders, there's at least 2 policy trees, if not all three, that can meaningfully access the wonder and build it with useful, constructive consequences. Great Library seems to fall outside this realm: it's not practically accessible to all trees and feels like it sides towards Tradition - but this is what I expect of a policy finisher, not a normal wonder. So yeah...I don't like it. Dropping the Policies by 1 would make GL far more interesting.

 

[tothePAIN]

I actually got Great Library in my most recent game and was completely surprised to do so. The only reason I managed it was because I was Poland and got the free Social Policy. Failing Great Library never strikes me as a tech problem, it strikes me as a Culture problem - I can beeline plenty of techs but I can't create extra Culture out of thin air.

Deity?

 

[Gidoza]

Deity?

Immortal.

 

[Cokolwiek]

I mean obviously the word *favoured* is important here. That is - for Terracotta Army for example, I can say that that's favoured by Authority because it synergizes with Authority's other bonuses.

Yes, of course, I wanted to contrast that faster tradition with slower authority which favors things like zeus, terracotta or great wall but forgot. My mind is like Swiss cheese. Don't know about angkor wat because I always grab it now.
On deity I don't even touch upper tech tree wonders other than forbidden palace until architecture. Library, lighthouse, parthenon, sankore, just forget it. Theology ones are sometimes possible though. When tradition I focus on the middle, gardens, then angkor wat for the mandir. Also, expanding on your good points you mentioned: you will not always war in a statue of zeus or terracotta way as tradition or progress, while parthenon or hanging gardens and similar wonders is always great, no matter wide, tall, warmongering or peaceful, as science, culture, and food which can be converted into great persons and strong capital is always desirable and are active at all times, not only at war or only at siege.

 

[Cokolwiek]

My point in the post above being, some upper and middle tech tree wonders are universally great as they boost food, science, or culture, great people for any playstyle, wide warmonger or progress, or tradition. They are slightly more beneficial to tradition. But many lower tech tree wonders are beneficial only when at war, or taking cities, like zeus, terracotta, so it is sensible and good that tradition or progress AIs don't rush them over great library or parthenon. However I think we should reconsider why hanging gardens and terracotta are valued so low (in my games consistently being built after roman forum, even if they have four policies) by the AI, which creates a safe spot for a player, while significantly decreasing tradition AIs potential (gardens are just great, they should aim for them) or wide or warmongering play (terracotta).

 

[Stalker0]

However I think we should reconsider why hanging gardens and terracotta are valued so low (in my games consistently being built after roman forum, even if they have four policies)

Part of this for the HG at least may be because Roman Forum is theoretically in greater danger. As its prereqs or less, more civs have access to taking it, whereas the HG takes more culture, and so should go later in the game.

In reality of course, culture civs can often get the HG as soon as they get mathematics, so it goes very quick.

 

[a3kov]

I don't understand the main reason why do we need to reward already well performing AI to be even better ? All it does is creates more opportunity for crazy runaways, which are very hard to deal with (especially Tradition OP civs on the other side of the map). If you are good in something the reward is already there - if you are good in production, you produce more, if culture you have more policies, if science more techs, etc. I am totally lost here.
I am in favor of giving them %-based bonus or discount on whatever they do (that will also make AIs different depending on their natural advantages).
Also production bonus affecting World Projects is even visible on Emperor. I stopped proposing World Projects in my games, because I can't predict the outcome at all. The current state of things is very broken

 

[Amask]

I think it's the worst feature of this mod.
Among many other reasons that others already mentioned, it makes making predictions about the game that much harder.
Say I have a few neighbors and I scouted most of their land. It's still early, and I'm formulating my game plan. I'd like to figure out who's most likely to be a threat. Is it the civ with early bonuses? Is it the civ with the best land? Is it the civ with most land to expand into?
It's none of those. It's the civ who randomly lucks out to the most wonders. Or maybe it's the civ that randomly decides to enter classical era sooner than others, which in turn probably helps it boost out a wonder.

I've seen AIs with amazing capitals fail completely and utterly. My guess is they kept trying to build wonders but kept failing every time, purely to bad luck.
Now not only do they have nothing to show for it, but other AIs leave them in the dust due to these ABC bonuses.
The whole thing introduces too much randomness and makes things like an AIs starting land pretty much irrelevant.

Isn't there enough randomness as it is? Why not just randomly choose a few civs on turn 0 and shower them and only them with these bonuses? Wouldn't be much different in my opinion.

 

[Amask]

To add to what I just said, there have been several maps that I've played from turn 0 twice, experimenting with different approaches.
And almost every time civ A who was the clear leader in one playthrough was at the bottom in the other playthrough. They usually (or maybe even always, I don't remember for sure) took the same ancient policy tree.
Their land doesn't matter. Their neighbors don't matter. The CS they have access to don't matter. All that matters is if they can randomly get the snowball rolling.

 

[crdvis16]

Maybe the bonus yields can be flattened or averaged out some. One problem seems to be an AI that enters a new era getting a rush of production- maybe that spike should be divided up to a much smaller amount per tech unlocked instead?

 

[Ziad]

I think a good solution would be a scaling mechanics that scales based on how behind the AI is, rather than the same bonus for everyone everywhere independent of how well they are doing.

An example is if a civilization is at 75% of the global average, then it should receive half the bonuses of someone who is below 25% of the global average etc...

I think many systems like it are already in place... Scholars in Residence is one.

This could help solve the runaway problem.

 

[stii]

Deity?

It is possible on deity, I have got it before, but I think you just have to get a bit lucky as producing enough science AND culture before any AI could possibly get it is just impossible, so building it is a bit of a gamble.

 

[tothePAIN]

It is possible on deity, I have got it before, but I think you just have to get a bit lucky as producing enough science AND culture before any AI could possibly get it is just impossible, so building it is a bit of a gamble.

I'm impressed then. I've tried for it as Siam, America, and Arabia with all sorts of starts without getting close at all.

 

[stii]

I'm impressed then. I've tried for it as Siam, America, and Arabia with all sorts of starts without getting close at all.

I think culture is the big issue so you need a way of generating a lot very early. I was one turn off with peaceful Netherlands off a good start. Only missing because it was my 2nd wonder.

The time I got it was a very strange game as there is another thread discussing Celts and early tribute so I was trying that out. I founded a an early religion and took the AI's city for a bunch of science and culture. Even then the AI was certainly at writing before I could start building so if they had started it right away I'd have missed it. I built it on t75. It might be possible to do it a bit quicker if you can get the religion faster because you can take the science/production founder and rush it that way. Probably still a bit too risky as sometimes the AI builds it super early.

 

[tothePAIN]

I think culture is the big issue so you need a way of generating a lot very early. I was one turn off with peaceful Netherlands off a good start. Only missing because it was my 2nd wonder.

The time I got it was a very strange game as there is another thread discussing Celts and early tribute so I was trying that out. I founded a an early religion and took the AI's city for a bunch of science and culture. Even then the AI was certainly at writing before I could start building so if they had started it right away I'd have missed it. I built it on t75. It might be possible to do it a bit quicker if you can get the religion faster because you can take the science/production founder and rush it that way. Probably still a bit too risky as sometimes the AI builds it super early.

The AI usually takes it in the 60s in games that I play. I'll try again though.

 

[Recursive]

I think a good solution would be a scaling mechanics that scales based on how behind the AI is, rather than the same bonus for everyone everywhere independent of how well they are doing.

An example is if a civilization is at 75% of the global average, then it should receive half the bonuses of someone who is below 25% of the global average etc...

I think many systems like it are already in place... Scholars in Residence is one.

This could help solve the runaway problem.

I believe Gazebo is highly against rubberbanding as a solution.

 

[ElliotS]

Rubberbanding is bad. If you kick a civ into the dirt they shouldn't get rocket-boosted back to relevance just off of AI bonuses.

As I keep mentioning: ABC bonuses came about BECAUSE flat % modifiers suck. You know how late-game in VP matters? That's partially because of the ABC bonuses. % bonuses mean early game is hard, and late game is easy. The ABC bonuses should theoretically allow AI to keep up with the expected output of a human player on any given difficulty at any point in the game.

They're based on objectives because A) That means an AI not doing anything doesn't get a ton of bonuses, so they reflect the actual state of the AI instead of just being more-or-less flat bonuses. and B) That's msotly how human players work. You get a wonder, build a city, or win a war? You're stronger than you were before. The ABC bonuses are meant to close the gap between what the player can get out of a wonder or war and what an AI can get out of it, to allow them to keep pace.

If you've been reading what I said carefully you'll notice that a lot of my sentences use terms like "theoretically" or "meant" because we all know the values are not perfect. You can modify them all in the difficulty files and report back if you find better values. I also suggest you turn on logging to support your report with data showing yields by era and comparing the AI to the player.

But just wanting to throw ABC bonuses out would be a massive step backwards, because lategame would become trivially easy.

 

[Stalker0]

As I keep mentioning: ABC bonuses came about BECAUSE flat % modifiers suck. You know how late-game in VP matters? That's partially because of the ABC bonuses. % bonuses mean early game is hard, and late game is easy. The ABC bonuses should theoretically allow AI to keep up with the expected output of a human player on any given difficulty at any point in the game.

One does not beget the other. There is nothing special about ABC that allows them to scale, % modifiers can scale as well if you design them that way.

That said, everyone lets be realistic. Throwing out the ABC bonus in favor of an entirely new system is a massive time sink, both in coding and balancing. The issue is not so bad that its worth all that.

So I don't mind looking at the rough spots and making some adjustments, but an entire new system....please no.


With that in mind, one of the simplest adjustments you could make. Instead of ABC scaling on the AI's actual era, have it scale on the world average. This does two key things:

1) It reduces the imbalance of AIs who happen to get ahead in era (which is already a solid advantage) by not also rewarding them with even more ABC bonuses.

2) It gives the weakest civs a slight (and I do mean slight) rubberband, as even though they fall behind they can get some of the ABC bonuses from stronger eras. But again its not going to give them the power to roar back with the force of a 1000 suns....it would just prevent them from falling even further behind due to a lack of more advanced bonuses (so they wouldn't drop off a cliff quite as quickly).

 

[ElliotS]

One does not beget the other. There is nothing special about ABC that allows them to scale, % modifiers can scale as well if you design them that way.

The benefits of ABC over scaling % modifiers is that ABC should reflect that AIs doing well are doing well. At it's heart it helps snowballing, because that's kinda the core of the game. It also doesn't benefit AIs that are getting shut down in the same way. (Where scaling % bonuses give as much of a bonus to an AI getting shut down as an AI flourishing.

With that in mind, one of the simplest adjustments you could make. Instead of ABC scaling on the AI's actual era, have it scale on the world average. This does two key things:

1) It reduces the imbalance of AIs who happen to get ahead in era (which is already a solid advantage) by not also rewarding them with even more ABC bonuses.

2) It gives the weakest civs a slight (and I do mean slight) rubberband, as even though they fall behind they can get some of the ABC bonuses from stronger eras. But again its not going to give them the power to roar back with the force of a 1000 suns....it would just prevent them from falling even further behind due to a lack of more advanced bonuses (so they wouldn't drop off a cliff quite as quickly).

I do like this suggestion quite a bit.