AI catching up

For topics that do not fit in another sub-forum.

Moderator: Oberlus

Post Reply
Message
Author
Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

AI catching up

#1 Post by Ophiuchus »

this is basically a random AI idea without doing any research if somebody suggested this...


I often quit the game because i dont think AI has a chance of catching up.

How about some mechanism which makes integration more valuable for small gains (e.g. constant amount of population gets integrated vs scaling with size).

So e.g. you have 8AI opponents, and you have the big evil empire outteching and outproducing the others.
3 AI are able to conquer 1 AI each and you are able to conquer 2AI empires because of your strength, the payoff for those only conquering 1 AI should be bigger. So afterwards there are only 3 AI left but it helps catching up for a while (while you are still integrating that 2nd conquered AI).

So with such a mechanism in place if the AI feels there is some overlord empire it switches to a more risky modes in order to speed up taking over neighbors (or being taken over).

Of course AI alliances would also help ;)
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: AI catching up

#2 Post by Geoff the Medio »

Could you elaborate on the "some mechanism" part?

Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

Re: AI catching up

#3 Post by Ophiuchus »

Top of my head would be something like happiness caps a percentage of the max output of colonies (or caps the growth of current output).

And that you need to spend influence to effectively increase happiness. But that also depends how influence gets generated.

Or maybe simpler: happiness naturally growths but slower the further away the colony is from your capital. So that fast expansion does not return fast results (and also loosing and reconquering a colony far from the capital hurts more).
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

defaultuser
Juggernaut
Posts: 854
Joined: Wed Aug 26, 2015 6:15 pm

Re: AI catching up

#4 Post by defaultuser »

I guess I'm not clear how that would preferentially aid the AIs over the hooman players.

User avatar
EricF
Space Dragon
Posts: 357
Joined: Fri Jul 29, 2016 10:12 am

Re: AI catching up

#5 Post by EricF »

Ophiuchus wrote: Tue Nov 20, 2018 2:09 pm Top of my head would be something like happiness caps a percentage of the max output of colonies (or caps the growth of current output).

And that you need to spend influence to effectively increase happiness. But that also depends how influence gets generated.

Or maybe simpler: happiness naturally growths but slower the further away the colony is from your capital. So that fast expansion does not return fast results (and also loosing and reconquering a colony far from the capital hurts more).
With this fast expansion doesn't give fast results but it doesn't hurt you either so it is still a plus in the long run.
You would need some mechanism where fast expansion actually hurt you in some way like a chance for revolt.
And/Or maybe it actually costs you PP's to increase the happiness of a newly conquered planet. And an unhappy
planet does not supply any PP's or RP's to the Empire. So they are effectively a drain on you for some time.

User avatar
megalith
Krill Swarm
Posts: 14
Joined: Mon Nov 19, 2018 1:42 pm

Re: AI catching up

#6 Post by megalith »

Geoff the Medio wrote: Tue Nov 20, 2018 8:49 am Could you elaborate on the "some mechanism" part?
I agree with the OP that AI can't catch up once player reacehs a level of "confortable" gameplay.
This isn't only an issue related to freeorion but in general a problem related to all 4x games. in late game the AI can't catch up with the player.

I suggest that there should be a scalling difficulty modifier instead of flat difficulty, for example the AI starts as tipical in the beginning of the game and ends up as maniacall in late game.

of course if the AI setting is maniacal at the start of the game then only the only way to improve the AI in late game it to make it receive scaling bonus modifiers, ie. at the start of the game the AI has 0% bonus to science and industry and as the game approches to end that bonus is gradualy increasing toward ex. 50%

that would make it possible for the AI to improve in later game stages or for example to better recover from teritory and ship loses.
Last edited by megalith on Tue Nov 20, 2018 8:04 pm, edited 1 time in total.

User avatar
EricF
Space Dragon
Posts: 357
Joined: Fri Jul 29, 2016 10:12 am

Re: AI catching up

#7 Post by EricF »

A lot of people, including myself, aren't real fond of games where
making the AI better is by allowing it to "cheat".
The AI should just play better.

User avatar
megalith
Krill Swarm
Posts: 14
Joined: Mon Nov 19, 2018 1:42 pm

Re: AI catching up

#8 Post by megalith »

EricF wrote: Tue Nov 20, 2018 8:04 pm A lot of people, including myself, aren't real fond of games where
making the AI better is by allowing it to "cheat".
The AI should just play better.
it's not a cheat, it's a scalling difficulty in end game to offset human player. of course the setting should be configurable in game setup and not forced upon a player. so it's up to a player to decide.

User avatar
EricF
Space Dragon
Posts: 357
Joined: Fri Jul 29, 2016 10:12 am

Re: AI catching up

#9 Post by EricF »

megalith wrote: Tue Nov 20, 2018 8:08 pm it's not a cheat
That is subjective.

Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

Re: AI catching up

#10 Post by Ophiuchus »

defaultuser wrote: Tue Nov 20, 2018 5:19 pm I guess I'm not clear how that would preferentially aid the AIs over the hooman players.
The intent is mostly to mitigate snowballing expansion (see the OP).
At the moment having guesstimated one-and-a-half the resources means about conquering twice the number of planets in the same time frame means doubling the gain (linear gain with number of conquered colonies).

If that gain is more constant at least for a while a smaller empire can conquer and gain the same amount as a bigger empire. Or in other words it should help a smaller empire to grow the same rate as a bigger one.

If the game of AI could get tweaked in direction of risky-but-more-likely-conquest that helps the AI in general to build up bigger empires. One sweet spot could to balance conquest/risk up to the maximum gain (i.e. do not invest risky in conquering when continuous maximum gain is already reached)
megalith wrote: Tue Nov 20, 2018 7:56 pmI suggest that there should be a scalling difficulty modifier
From a development perspective we should not take the pressure out to increase AI strength, we should try to get feedback what could really help the AI and where does it suck.
I think making the game optionally harder for oneself is valid for a player. So maybe a valid addition after 1.0 or we have somebody building a player community or we get more useful AI feedback than we can handle.
Another way would be to think up obstacles/content/mechanism which are easy to handle for AI but hard for humans. Maybe a thousand needles strategy - something which is so hard to keep track of compared to the gain that even micromanagers would not do it (but AI does not care about repetition and adding lots of small factors), one could scale up the relevance of that mechanisms in the galaxy setup to increase the challenge for human players.
Or just make those features which the AI cant handle well more expensive.

Or maybe give the AI early access to features. E.g. implement a species exchange (e.g. by mutually "colonizing" a friendly empires outpost with a species the other one needs). As there is no way yet to negotiate that between human and AI, AI would have an advantage in single player.
EricF wrote: Tue Nov 20, 2018 5:25 pmWith this fast expansion doesn't give fast results but it doesn't hurt you either so it is still a plus in the long run.
Well the long run may mean in this case that it will actually never be a plus.
What i mean with this: as long as you expand this will be maxed out. Just if you are not able to expand the gain will catch up (and in the end you could have expanded slower with the same result for you).
Also the enemies could conquer your territory before you have a gain. Maybe they could use influence projects to keep your happinesses down.
EricF wrote: Tue Nov 20, 2018 5:25 pmYou would need some mechanism where fast expansion actually hurt you in some way like a chance for revolt.
Well, that is part of the happiness story i think. Low happiness could leave you more vulnerable to enemy influence projects (e.g. spawning pirates, inciting revolts ..)
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

Post Reply