[NFP] AI crazy about Science

[MrRadar]

What difficulty setting are those victories on?

Deity AI, if you're asking about my tests, I've posted the settings.

OK, last test.
I have jacked up the science value back to 150, as per April patch (<Row ListType="ClassicalYields" Item="YIELD_SCIENCE" Value="150"/>), added back both types of laser projects to the list of AI favoured items, and ran the game with the same civs and settings.

All the AI were visibly poorer and had less troops. Tech leaders had somewhere double the science in comparison to the previous games with the value of 40.
on t266 three civs were on or close to the Moon:

Spoiler :

Look at their science yields: Teddy and Mvemba clear leaders, though Gilgabro got to the Moon before Mvemba with considerably less of their science:

Spoiler :

And then AI took a break from space race for about 30 turns. Teddy did everything in his power not to research Nanotechnology, he even went for Smart Materials and further techs down that line first:

Spoiler :

There are three unresearched techs before Composites to the left down here

By the way, here is Kongo's policies on t280 (Kongo reached Moon by this time):

Spoiler :

Btw, there was zero chance for anybody to get a CV at this point

20-30 turns later, Teddy, done launching Exoplanet Expedition and killing off Scotland on the side, finally applied himself to building lasers, four at a time, exclusively Lagrange ones, as he had a nice pile of aluminium, and one after another:

Spoiler :

And he reached a record speed of 10 ly/t:

Spoiler :

In the end, Teddy who was showing >700 science yields at some intervals, arrived at an exoplanet on t336, which is 1 turn later than the previous game, where it was "ClassicalYields" Item="YIELD_SCIENCE" Value="40" instead of "ClassicalYields" Item="YIELD_SCIENCE" Value="150" here.

So in conclusion it looks like that jacking up those science yields for the AI does nothing. Human player just does not get many great scientists from the earlier eras, as they get grabbed by the pumped up but otherwise completely clueless AI which isn't even capable of supplying a credible story building environment for a roleplaying human player, like in any previous iteration of Civ. And both the player and the AI have to face more advanced barbs.

 

[Kotofei]

Deity AI, if you're asking about my tests, I've posted the settings.
By the way, here is Kongo's policies on t280 (Kongo reached Moon by this time):.

Nevermind the policies, if the AI is running a Tier 1 govt on turn 280 somebody at Firaxis needs to be held accountable. Where was Barbarian Carl during playtesting?
Moderator Action: Edited to remove problematic language. ~ LK

Edit: Is that a Kongo thing? I just loaded up a recent game and saw that they were running Oligarchy on T270.

 

[DizzKneeLand33]

@Infixo after reviewing the old Victories file from Oct, 2020 (I hadn't updated the game on one of my other computers), here are two lines that were added between then and the final patch:

<Row ListType="ClassicalDistricts" Item="DISTRICT_CAMPUS" Favored="true"/>
<Row ListType="ClassicalDistricts" Item="DISTRICT_THEATER" Favored="true"/>

So the AI favored status changed for Theater Districts as well as for Campuses (or was created, I don't see this for any other district). The science value for Classical at 150 is bizarre, especially when it is changed in Medieval to 15. @MrRadar maybe that is why you experienced the Science rush in the beginning but then the game ended up being the same in the end (whether it was set to 40 or 150)?

Classical Faith is 20, Food is 15, Gold is 20, and Science is 150.
Medieval Faith is -25 (MINUS! -- I mean there are uses for faith besides spamming missionaries iirc), Food is 25, Gold is 15, Production is 20 (that wasn't listed for classical) and Science is 15 (That's right, FIFTEEN: Did they accidentally add a zero to the Classical figure??).

You know, the more I think about it, the more I think the 150 is a complete typo, making it ten times what it was meant to be. And, this is why the games that I've been playing recently on Immortal have opponents who don't like me but don't even bother attacking me (perhaps). They used to rush you, I haven't seen that in almost a year, probably because they are all busy building campus and theater things.

So, I'm going to remove both favored entries, change Classical Science to 15, and see if we get at least one war in the ancient/classical/medieval era. I may adjust the faith to something more than zero as well (even 5). It goes to -40 in the Industrial Era. I'm going to test it first without the change to faith.

 

[aieeegrunt]

Bwahahahah a typo. My sides. I’m dying. A single typo was what made the Alien:Colonial Marines game have dysfuntional AI.

That is the level of professionalism we are at.

This glorious fail brought to you by the Department of Zero Playtesting.

 

[Infixo]

@DizzKneeLand33 A typo is a very probable explanation.
Those values are percentages to the base and they stay. I.e. Faith 15 means that faith yield is valued 15% more. -25 means that is valued 25% less. Time strategies are based on a condition „later or equal to”, so what is put in the classical yields - stays throughout the game. Later other yileds are also pumped up so the effect may be less visible.
Plus bear in mind that scoring functions that evaluate what to build in the city are much more complex, yield or what is favored are one of many factors taken into consideration. At least I hope so

 

[Infixo]

My theory why they made districts as favored is that when evaluating the item to build, the game doesn’t „look into the future”. The value is what is atm. I observed and exolained when working on Real Balanced Pantheons. So here we have districts. when I was working on RST I observed that pseudo yields associated with Great People are the ones that actually influence the building of discrricts. And that is consistent with „ai doesnt see the future”. The only value you get from a discrict which is for now and easily calculated from tables is adjacency and Great People Points.
I think Firaxis really wanted to pump up science but the person who did so maybe didn’t know how to that and just made campus a favored district, which is the easiest.

 

[Linklite]

I wonder what would happen if the preference for Science were to be set at 150 throughout the game? Or if gold were set to a similar level as well? Would that make the AI a major competitor for an SV? As I said before, the AI have me stiff competition for science generation in the early part, if they could continue into the rest of the game? Would be nice if that were for high difficulties, though.

I'm confused about TDs being favoured, I thought the AI didn't seem to be building them? We has a thread about it here and its rare I come across them. Is it because science has such a high bias?

Regardless, can we tag one of the FXS people here? They seemed to be planning mire updates (even if just bug fixes etc) and this seems a very easy fix that they could do in seconds (before testing, etc at least). Not Andrew though, iirc he doesn't have anything to do wirh coming and doesn't appreciate being tagged for things outside of his focus, the history side of things.

 

[aieeegrunt]

I wonder what would happen if the preference for Science were to be set at 150 throughout the game? Or if gold were set to a similar level as well? Would that make the AI a major competitor for an SV? As I said before, the AI have me stiff competition for science generation in the early part, if they could continue into the rest of the game? Would be nice if that were for high difficulties, though.

I'm confused about TDs being favoured, I thought the AI didn't seem to be building them? We has a thread about it here and its rare I come across them. Is it because science has such a high bias?

Regardless, can we tag one of the FXS people here? They seemed to be planning mire updates (even if just bug fixes etc) and this seems a very easy fix that they could do in seconds (before testing, etc at least). Not Andrew though, iirc he doesn't have anything to do wirh coming and doesn't appreciate being tagged for things outside of his focus, the history side of things.

The problem with campus/science spam is then the AI neglects the most important district in the game; Commercial/Harbour. This probably leads to things like the AI going bankrupt and disbanding their military which seems to be a thing.

 

[MrRadar]

Well, the previous one wasn't my last test, I ran another one

I've removed campus and TD preferences that @DizzKneeLand33 quoted above:
<Row ListType="ClassicalDistricts" Item="DISTRICT_CAMPUS" Favored="true"/>
<Row ListType="ClassicalDistricts" Item="DISTRICT_THEATER" Favored="true"/>

Then reduced classical yields from 150 to 15, and added increasing yields to Renaissance, Industrial and Modern, like this:

Spoiler :

And I've kept both types of laser projects as favourited.

The resulting AI only game was plenty of action including nukes:

Spoiler :

and a very participatory space race:

Spoiler :

Sadly, there were only two lasers built: Trajan and Mvemba one apiece, and both Lagrange. I can't understand why they woudn't build terrestrials even if they are favourited.

And would you just look at those AI yields on the penultimate turn (Mvemba won SV on t342):

Spoiler :

AI have plenty of all kinds of yields and impressive military, even with planes zooming around.

So jacking up science in classical and making campus and TD a priority does nothing else but cripples the AI and lowers the quality of the gameplay significantly. Finishing times are almost the same.
By the way, making TD a priority also arguably takes away the slim chance AI had to achieve a CV. In previous game domestic tourists were 1200 and nobody had any hope. In this game Germany was somewhere at 250/540, and ~840 for the others.

I think Firaxis really wanted to pump up science but the person who did so maybe didn’t know how to that and just made campus a favored district

It looks like very much the case.

Zero Playtesting.

And this as well.

Either they did not test, or tested insufficiently or could not understand what those changes or test results mean.

Tweaking yields is one thing, but also the decisions of the AI must be looked into. What I saw was that AI goes to the Moon and then has quite a few techs remaining to Nanotechnology, however AI does not go straight for it, but meanders around going for Computers, Robotic and Smart Materials first. Hence the observed pause in the space race. As soon as it gets the required tech, AI immediately starts respective projects.

By the way, the detour for Computers maybe is triggered by the Global Warming that comes just around the time when the AI has reached the Moon, so the AI gets Computers and starts Flood Barriers. But there's a bug which makes Deforestation level jump from 0% to 50% in one turn, speeding up the Climate Change phases considerably, so the world sinks fast and AIs costal cities may get stuck with unfinishable and terribly expensive Flood Barriers for a while.

 

[DizzKneeLand33]

@MrRadar wow that is impressive! I'm changing my numbers to what you used above and will see what happens. I play currently with 129 civs in total (randomly selected) 13/22 Immortal on a large map with 12 religions (which should encourage religious wars) using Civs Expanded, Extended Eras, Dramatic Ages, More Movement, and No Spies (the spy system is atrocious and only aids the human in the long run imho -- I eliminate the leaders with a mostly spy focus to compensate).

Nukes *and* space wars *and* planes. Nice. On which level did you run that test?

I'm hoping this makes the AI good enough so that I can back the difficulty back to Emperor and have a fun game.

 

[MrRadar]

Nukes *and* space wars *and* planes. Nice. On which level did you run that test?

All tests on Deity AI, so that it wouldn't be mostly score victories
Nuking pattern is the same old one: multiple nukes on the same city and no other follow up.

 

[DizzKneeLand33]

Well I'm on turn 60 now (standard speed with Extended Eras) and started as Ethiopia. Pope Pius IX started next to me (ironically) and declared war on turn 48 and sent a *ton* of units at me. First declaration of war I have seen in my testing since the update (through about turn 120 in each game). Turn 60 I saw a missionary from Gaul show up at my territory (I had not seen a missionary by turn 120 in any game since the update). This is with adjusting all of the numbers as you have them, so the game is starting more dynamically if nothing else. Religious wars can be so unpleasant.

I'm sure that nuking pattern is inside the code, sadly -- if they could fix a few things like that maybe we can fix the rest.

EDIT: I didn't realize that there was an absolutely game-breaking bug in using DA mode, so I will restart and turn all modes off (since almost all have their bug issues) -- also CUI has trade screen issues that CQUI seems to now fix, so I'm going back to that. I'll start with Ethiopia again and hope that I get similar early game results!

 

[MrRadar]

Ok, another one
Minimal changes this time, classical science yield loses the zero and is reduced from 150 to 15, also campus and TD removed from favourites for all:

Spoiler :

Lasers stay favourited.

Here are some not too bad midgame AI yields and an example of how painfully Deity AI can be shafted by a continuously triggered barb camp:

Spoiler :

Ouch, I feel for ya, Freddy, mate, show 'em some gewalt!

Endgame AI yields and all kind of assets also look very respectable:

Spoiler :

The main SV race were between Korea and Kongo. Korea was on the Moon around t220 - quite early, but then twiddled her tumbs for a while until Kongo almost caught up. So they're both en route to some exoplanet, Korea travels some 15 turns ahead, they both go at 1 ly/t and both refuse to build any kind of laser project despite me leaving lasers among favourite items:

Spoiler :

They have no aluminium for Lagrange, because they've overbuilt planes and depleted all stocks. And they just won't build terrestrial lasers. At this point Seondeok was on her way to win a SV somewhere around t339.
But as none of the science leaders were in a visible hurry, Freddy went and kicked their ass by pulling off a DV on t327

Spoiler :

He must've gotten 5 points in the congress because I had just checked victory tabs at that point and he had 15/20. In the spectator mode the game pauses during the Congress votes and you must return to the map and Sift+Enter to continue. I checked during the pause and he won as soon as I pressed shift+enter.

So there you go - minimal changes and fastest AI victory time No need to sweat with that +150 classical science and prioritised campuses and TDs, plus the gameplay is back to more varied and engaging one, with AI possessing lots of units.

Looks like, @Infixo found the root of several issues that have been damaging the gameplay for some months now. FXS has made quite a sizeable oopsie in their 'final' patch. Now a few 'ultimate' ones are very much desirable to sort out the mess.

Some more observations: no matter what the patch notes said, AI massively overbuilds resource consuming units and depletes all their stocks and then runs low on them. In this particular game there were no nukes because everybody who could built a gazillion GDRs and no uranium was left for the A bombs. Lots of infantry and tanks etc. starving for oil.
AI hardly builds any ships until the most modern units come online, so almost all game, until they launch Earth satellite, they stay ignorant of the wider world, they only explore their own landmass. In this civ selection only Harald consistently had some sort of navy containing his unique units and did some ocean exploration.

 

[aieeegrunt]

It’s absolutly unacceptable that a couple of randos had to find and fix this problem for them.

Even more so because everyone on console is essentially screwed.

I mean Bethesda at least whole heartedly embraced their laziness and embraced and fostered modding

 

[uhu]

So there you go - minimal changes and fastest AI victory time No need to sweat with that +150 classical science and prioritised campuses and TDs, plus the gameplay is back to more varied and engaging one, with AI possessing lots of units.

Congratulations for your FANTASTIC work! I bet it was not only exhausting, but also quite exciting and rewarding to find out all that stuff.

We see again: Civ6 has great potential and is still a good game (and needs only some polishing and fine tuning).

I think Firaxis really wanted to pump up science but the person who did so maybe didn’t know how to that and just made campus a favored district, which is the easiest.

Zero Playtesting.

And this as well.

Either they did not test, or tested insufficiently or could not understand what those changes or test results mean.

Timeout. Budget exhausted. Manpower transferred to other projects. replacing Developers modifying huge piles of code of colleagues under extreme time pressure ...

I'd say some beancounters pulled the plug a bit too early. You know, they seem quite neurotic based on short term profit maximization. None of them plays games or even love games. Why should they make a great or just sufficiently matured game?

When I started developing software in the area of production automation (some time ago ), one of the first sayings I heard from the business people (on the other floor): "At one point you have to shoot the developers in order to get the product out of the door duly ..."
So nothing really new, but the development of mankind is accelerating more and more ... (and the acceleration is accelerating)

Still I had always the impression, Sid Meier, Brian Reynolds and Soren Johnson had an economical look onto their games, but also wanted to be proud of them because the games were great (kind of the games they wanted play themselves) - talking about reputation and long term customer satisfaction & loyalty.

Looks like, @Infixo found the root of several issues that have been damaging the gameplay for some months now. FXS has made quite a sizeable oopsie in their 'final' patch. Now a few 'ultimate' ones are very much desirable to sort out the mess.

I'm sure that nuking pattern is inside the code, sadly -- if they could fix a few things like that maybe we can fix the rest.

An while ago I was sure, that we will receive at least one more patch.

But the naturalness of absolute silence concerning gameplay or development makes me feel like on a ghost ship now ... some lonely historian setting straight the deckchairs on the luxury floor while down under the first hatches are bursting ...

I'd be not surprised, if Kevin's final jokes are true now: We are on our own.

 

[Kotofei]

@MrRadar awesome work. I'm definitely using some of those changes.

 

[aieeegrunt]

Congratulations for your FANTASTIC work! I bet it was not only exhausting, but also quite exciting and rewarding to find out all that stuff.

We see again: Civ6 has great potential and is still a good game (and needs only some polishing and fine tuning).


Timeout. Budget exhausted. Manpower transferred to other projects. replacing Developers modifying huge piles of code of colleagues under extreme time pressure ...

I'd say some beancounters pulled the plug a bit too early. You know, they seem quite neurotic based on short term profit maximization. None of them plays games or even love games. Why should they make a great or just sufficiently matured game?

When I started developing software in the area of production automation (some time ago ), one of the first sayings I heard from the business people (on the other floor): "At one point you have to shoot the developers in order to get the product out of the door duly ..."
So nothing really new, but the development of mankind is accelerating more and more ... (and the acceleration is accelerating)

Still I had always the impression, Sid Meier, Brian Reynolds and Soren Johnson had an economical look onto their games, but also wanted to be proud of them because the games were great (kind of the games they wanted play themselves) - talking about reputation and long term customer satisfaction & loyalty.

An while ago I was sure, that we will receive at least one more patch.

But the naturalness of absolute silence concerning gameplay or development makes me feel like on a ghost ship now ... some lonely historian setting straight the deckchairs on the luxury floor while down under the first hatches are bursting ...

I'd be not surprised, if Kevin's final jokes are true now: We are on our own.

If you only care about this quarter’s bonus you make Civ6

If you want people to auto buy your purchase on launch day because they know they’ll still be getting fixes and content a decade later you make Fallout 4 and Elder Scrolls

 

[Socrates99]

If you only care about this quarter’s bonus you make Civ6

If you want people to auto buy your purchase on launch day because they know they’ll still be getting fixes and content a decade later you make Fallout 4 and Elder Scrolls

So much of this is voted on or decided by a board that may or may not still own shares when Civ VII launches. A privately owned company or worker co-op where individuals depend on the brand might make the type of decision you describe but a nonliving entity like a corporation doesn't. Immediate short term profit matters too much.

 

[Aristos]

So much of this is voted on or decided by a board that may or may not still own shares when Civ VII launches. A privately owned company or worker co-op where individuals depend on the brand might make the type of decision you describe but a nonliving entity like a corporation doesn't. Immediate short term profit matters too much.

I don't know... I hear what you say, and it may be true, but it is just too comfortable for the coders at Firaxis... I mean, if ONLY the suits are to blame, and we do not even consider potential plain and simple ineptitude as another factor, then how can that change in the future?

As I said before, I would not dismiss plain and simple ineptitude. Just look at the general demeanor and attitude of some of them in their videos... do they really look like professionals?

@MrRadar : you should definitively release these "simple" changes as a mod.

 

[aieeegrunt]

I don't know... I hear what you say, and it may be true, but it is just too comfortable for the coders at Firaxis... I mean, if ONLY the suits are to blame, and we do not even consider potential plain and simple ineptitude as another factor, then how can that change in the future?

As I said before, I would not dismiss plain and simple ineptitude. Just look at the general demeanor and attitude of some of them in their videos... do they really look like professionals?

@MrRadar : you should definitively release these "simple" changes as a mod.

WhyNotBoth.jpg

The owners set the tone of the business and decide what is and is not a priority.

If the message from on high is “yeet it out the door so we can monetize it now” that is what happens

If the message is “don’t bother with fixes, put resources into new civs we can monetize” that is what happens

That being said, something as basic as this fracking typo screams unprofessional to me