AI's totally backwards Great Person priorities

[Alzadar]

Has anyone else noticed this? Pretty much without fail the AI will plant Great Engineers and Great Merchants (captured cities always have Manufactory and Customs Houses) and bulb Great Scientists, which is the complete opposite of what a human player will usually do.

This is especially tragic for Babylon, who inevitably bulbs the Great Scientist he gets on turn 10.

Seriously though, why did they make Manufactories and Customs Houses so weak compared to Academies? 8, which you can't get from a tile improvement otherwise, compared to 4 (with a level 1 tech providing an improvement of 2) and 4 (vs Trading Posts giving 1). What gives? Why are those two tiles improvements so bad, and why does the AI love them so much?

Bit of math for fun:
A Great Engineer used in a city of population 10 gives 600, even if you put him on grassland he still needs 150 turns just for the Manufactory to equal that, and it requires working a tile.

It's even worse for the Great Merchant, using him on a trade mission is worth like 1000 gold (if you count 30 CS influence as 500g), so even with a Market AND a Bank in the city he's planted, you need over 200 turns just to break even!

Compare this to a Renaissance Scientist being planted, he breaks even with what he would have bulbed in like 30 turns!

 

[TheBlackHole]

Wait a minute, there's no way that academies break even after 30 turns.

A very rough heuristic is to bulb everything after Scientific Method (note: opinions vary). No way that would be the heuristic if you could make back your bulb in 30 turns with an academy. You'd be placing academy nearly every single time.

I do the calculation every game: I estimate the rough yield of the academy per turn (factoring in modifiers and future academy bonuses based on tech). Then look at the bulb amount. Do some division to estimate how many turns it is. It's all very rough estimating, but usually the number comes out to be somewhere on the order of 100 to 150 turns -- which is about right as it then becomes an interesting decision of "now vs. later" (and note that this makes it on the same order of magnitude as your estimates for Great Engineer and Great Merchant). There's no way it could only be 30 turns, which would make it a no-brainer almost every time.

 

[Tachii]

AI do expend GEs on wonders sometimes. They probably won't if that wonder is low on their priority list.

As for Merchants, how do you even know they aren't spending it on CSes?

I do occasionally see an academy. Not often enough I guess. Would probably be a decent challenge for Immortals who are getting bored of Immortal, I guess.

 

[Alzadar]

Wait a minute, there's no way that academies break even after 30 turns.

A very rough heuristic is to bulb everything after Scientific Method (note: opinions vary). No way that would be the heuristic if you could make back your bulb in 30 turns with an academy. You'd be placing academy nearly every single time.

I do the calculation every game: I estimate the rough yield of the academy per turn (factoring in modifiers and future academy bonuses based on tech). Then look at the bulb amount. Do some division to estimate how many turns it is. It's all very rough estimating, but usually the number comes out to be somewhere on the order of 100 to 150 turns -- which is about right as it then becomes an interesting decision of "now vs. later" (and note that this makes it on the same order of magnitude as your estimates for Great Engineer and Great Merchant). There's no way it could only be 30 turns, which would make it a no-brainer almost every time.

I said Renaissance Scientist... Just a rough assumption of bulbing being ~600 (this means a tech like Printing Press is taking around 8 turns, which I think is right).

AI do expend GEs on wonders sometimes. They probably won't if that wonder is low on their priority list.

As for Merchants, how do you even know they aren't spending it on CSes?

I do occasionally see an academy. Not often enough I guess. Would probably be a decent challenge for Immortals who are getting bored of Immortal, I guess.

I've never seen an AI perform a trade mission, maybe they do but given the number of Customs Houses they usually have, it seems unlikely.

I'd rather see Great Improvements significantly buffed than have the AI swap its priorities though. Manufactory should honestly give like 8-10 and Customs House should be around 8. They should be good enough that it's actually a choice whether to plant or to use, not "ooh a Great Engineer, what wonder should I rush with it?".

 

[Strategist83]

Good question indeed...

I honestly don't think anybody agrees with Firaxis on the balancing on this, even if it's clear the yields were carefully decided upon with the large differences. is generally the most valuable resource, so having the Academy yield be so high (or rather, the yield of the others being so low) is incomprehensible, especially when it also gets the best % modifers.

My guess is they were afraid of putting 8-10 on the Manufactory because of what it could potentially do to settler production, wonder production or military production which all see some respectable modifiers. Still, having access to much more advanced units or unlocking the ability to construct wonders much earlier seems just as overpowering.

The Customs House appears particularly weak with generally being worth less than any other resource, but there is the new aspect that it impacts the value of trade routes by adjusting the output of the city. I haven't looked into the math of this enough to see if it's significant, but +4 and some extra trade gold isn't going to be earth-shaking either way.

 

[joncnunn]

I have a feeling the AI was coded for Vanilla's logic and not updated to the much higher academy yield in G&K (and BNW)

The AI does sometimes rush wonders with GE. But I suspect if at the time a GE is born that it didn't see the right wonder for it's flavors available it just plants a manufactory instead of waiting like a human would.
(It would be difficult to raise the hammer yield high enough to get me to plant one instead of rushing a wonder.) This is also a type that I normally only fill in poor hammer cities; and is still secondary to filling science slots.

Yup; I've never seen an AI conduct a trade mission with a GM. I don't fill the spots for Merchants at all unless playing Venice. They would have to change the rules to give Merchants their own counter for me to fill those slots, the last thing I want is a GM at the cost of either a GS or GE.

 

[Karmah]

I ve seen the AI conduct trade mission quite often (Unless they were just taking a walk into CS land but I doubt it)
Also I like to settle my GEs when I can (namely when I don t have a key wonder to rush soon) , especially early, it s a production bonus that will be multiplied by the workshop/factory and so forth , It can provide culture with the right congress resolution and will be quite ehanced if I go for freedom (even more if i play Korea)

 

[enki123]

1. Most of what the AI chooses to do is very bad. After learning the mechanics of the game ANYONE can beat deity.
2. That being said, the engineer and merchant bonuses need to be improved. As the OP says there is rarely a reason to settle them.
3. It doesn't ing matter because it isn't like the devs are gonna change anything based on what we discuss on these forums.

 

[dexters]

If the human is wonder hogger, it does make sense to plant GE than to keep then around eating GPT.

Industrial complexes is the single tile improvement Ive seen most of and its probably for this reading.

Titles from GS and GM are rarer in that order.

 

[Magma_Dragoon]

Of course you've never seen it! The great person is expended on a trade mission or wonder rush, so how do you expect to see it happen? And another thing, you are using human numbers to calculate ROI for planting vs special ability for the AI.

 

[Dylanhasanose]

welp today I learned how to use my great people efficiently. What about Great Arists? Landmarks or golden age? Is a religious site worth it for great prophets? Usually I just use them to spread.

 

[arand86]

welp today I learned how to use my great people efficiently. What about Great Arists? Landmarks or golden age? Is a religious site worth it for great prophets? Usually I just use them to spread.

All depends what you need at what time. Late game, of course, popping Golden ages and city conversion are best. Early game, the faith and culture stack up.

Another consideration for prophets are what you need faith for. I love the GP buying reformation belief, so if I have that, I'm much more inclined to slap a religious site so I can buy great scientists and engineers faster.

 

[rondiggity]

I like to build Holy Sites to help accrue faith for a late game win (usually a Cultural victory, allowing me to generate 3-4 Great Musicians).