How to convert 30 bucks into 1200 - get rich now! (starbase exploit)

Exploiting the AI with starbases

from GalCiv3 Forums

So in my current game (1.82 with most of the DLCs) a basic starbase (meaning a just constructed starbase with nothing on it) is worth around 1260 "space dollars" to a factions that loves me - but it doesn't matter where that starbases is located.

Currently I can have 3 constructor modules on a cargo hull.

So what I can do now is go somewhere into a void between stars (insane map with tight clusters gives enough room for that) with 3 module-constructor and build 2 starbases. The constructor has 1 module left so I upgrade it for around 60 space dollars to another construtor with 3 modules - with that I can build 2 more starbases ...

I can sell each of those useless starbases in the middle of nowhere to an AI faction that loves me for 1200 units of money - so basically I can make 1200 bucks for an investment of 30 bucks. That means multiplying by 40!

Now obviously the price for the starbases vary, depending on the relations to the faction and the modules. But I would argue that counts as an exploit.

I even used it to buy a planet - because giving the AI a starbase for a planet is apparently the same than giving it 1200 space dollars, so with enough starbases the AI is willing to trade a planet.

Now I totally love to buy planets, in fact it is probably the way I get most of my planets. But the AI should not blindly value any starbase with the same amount of money. I guess currently the value for a starbase is just calculated by its moduls. But the location is also important. On an insane map there can be huge voids and the only benefit the AI gets from a starbase in a void is maybe more traveling range - but that should not count the same as a starbase close to a planet or a resource. Especially not if there are 5 starbases grouped together as tight as possible - it's just not worth 6000 dollars, especially during a gaming phase where if even the richest factions struggle to have more than 5000 bucks in cash.

 

11,911 views 9 replies
Reply #1 from GalCiv3 Forums Top

The AI should not sell planets, agreed, however as far as buying starbases, well, it is limited by how much money the AI has....

 

I am usually hard pressed to find AI's with enough cash to buy my techs let alone starbases.

 

It could be handy to aid me with buying techs from a more advanced AI though.... so i will keep this 'exploit' in mind for that situation, thanks.

Reply #2 from GalCiv3 Forums Top

Quoting Mystikmind, reply 1

The AI should not sell planets, agreed,

Not agreed. I love buying planets!

Quoting Mystikmind, reply 1
however as far as buying starbases, well, it is limited by how much money the AI has....

No, because they also trade you other stuff for it.

Multiplicity - Control multiple PCs using one keyboard and mouse
Reply #3 from GalCiv3 Forums Top

Quoting Empress_Fujiko, reply 2


Quoting Mystikmind,
reply 1

The AI should not sell planets, agreed,



Not agreed. I love buying planets!


Quoting Mystikmind,
reply 1
however as far as buying starbases, well, it is limited by how much money the AI has....



No, because they also trade you other stuff for it.

 

Ok compromise..... they can make a formulae that the AI will never sell planets within x distance of their homeworld (map size variable as well)

Historically it makes sense.... America can buy Alaska off Russia, but do you think America can Buy Siberia too? NO! of course not, that is what i am talking about.

 

Yea, hence what i said about it being handy for that.

Reply #4 from GalCiv3 Forums Top

Everything should be for sale, at a rational price.

Russia absolutely would have sold the USA Siberia, just not for the same price/acre as they did Alaska.  Heck, Mexico sold us the Gadsdend Purchase  - the small strip of southern Arizona/New Mexico.  That was a part of the core Mexican territory. Yet it was happily sold to the USA, just at a price/acre that was 40 times what the USA paid for Alaska.

The key here is to get the AI to recognize that most of a thing's worth (ship, fleet, technology, planet, starbase) has very little to do with the specifications of the thing in question, and much more with the thing's UTILITY.  How the AI manages to compute utility a hard problem, since humans don't do it very well at all, either.

Clearly, a single starbase far-distant from anything else a player owns is of fairly high utility, compared to one within 10 hexes of another. But also, a fully-stocked military starbase next to a potential conflict zone is worth many, many more times than a single starbase way out in the middle of nowhere, even if that military starbase is right next to several others.  There's a whole lot of math and judgement that has to go into assessing Utility, and it's not at all clear how to make this kind of assessment.

Not to mention that Utility is judged by not only it's usefulness to the seller, but also to the buyer.

Reply #5 from GalCiv3 Forums Top

Quoting trims2u, reply 4

Everything should be for sale, at a rational price.

Russia absolutely would have sold the USA Siberia, just not for the same price/acre as they did Alaska.  Heck, Mexico sold us the Gadsdend Purchase  - the small strip of southern Arizona/New Mexico.  That was a part of the core Mexican territory. Yet it was happily sold to the USA, just at a price/acre that was 40 times what the USA paid for Alaska.

The key here is to get the AI to recognize that most of a thing's worth (ship, fleet, technology, planet, starbase) has very little to do with the specifications of the thing in question, and much more with the thing's UTILITY.  How the AI manages to compute utility a hard problem, since humans don't do it very well at all, either.

Clearly, a single starbase far-distant from anything else a player owns is of fairly high utility, compared to one within 10 hexes of another. But also, a fully-stocked military starbase next to a potential conflict zone is worth many, many more times than a single starbase way out in the middle of nowhere, even if that military starbase is right next to several others.  There's a whole lot of math and judgement that has to go into assessing Utility, and it's not at all clear how to make this kind of assessment.

Not to mention that Utility is judged by not only it's usefulness to the seller, but also to the buyer.

 

Well actually the point of Alaska and Siberia is to demonstrate that some things can be sold and some things are vital to securing ones territory and should not be sold. It should be reasonably possible to get the AI to do this with a simple calculation of distance from home world.

Reply #6 from GalCiv3 Forums Top

Quoting Mystikmind, reply 5

 

 

Well actually the point of Alaska and Siberia is to demonstrate that some things can be sold and some things are vital to securing ones territory and should not be sold. It should be reasonably possible to get the AI to do this with a simple calculation of distance from home world.

Siberia isn't specifically vital to Russia's security, if the price had been right - selling a large hunk of the Western Siberia to the USA for $100 million might easily have been possible, if the US really wanted it.    Conversely, far-flung positions can be immensely vital to national security. Owning Gibraltar is far more important than the Channel islands for the U.K., for instance.

My point was that this metric is useless. Distance from homeworld isn't even a passable way to evaluate worth.  I can sell you Mars, and it's gonna get flipped back to me shortly (or, I can seize it trivally), if I own Earth.   If I own a world over near the frontier where you're fighting someone else, that world is now very valuable to you.

It's all about utility, and utility is extremely hard to programatically define.  At best, I can name at least a half-dozen things that minimally should be considered, with no single one dominating.

Reply #7 from GalCiv3 Forums Top

Quoting trims2u, reply 6


Quoting Mystikmind,
reply 5


 

 

Well actually the point of Alaska and Siberia is to demonstrate that some things can be sold and some things are vital to securing ones territory and should not be sold. It should be reasonably possible to get the AI to do this with a simple calculation of distance from home world.



Siberia isn't specifically vital to Russia's security, if the price had been right - selling a large hunk of the Western Siberia to the USA for $100 million might easily have been possible, if the US really wanted it.    Conversely, far-flung positions can be immensely vital to national security. Owning Gibraltar is far more important than the Channel islands for the U.K., for instance.

My point was that this metric is useless. Distance from homeworld isn't even a passable way to evaluate worth.  I can sell you Mars, and it's gonna get flipped back to me shortly (or, I can seize it trivally), if I own Earth.   If I own a world over near the frontier where you're fighting someone else, that world is now very valuable to you.

It's all about utility, and utility is extremely hard to programatically define.  At best, I can name at least a half-dozen things that minimally should be considered, with no single one dominating.

 

If i had to make an AI to run Russia, i would want it to reject any offer for selling Siberia, but selling Alaska is fine.... If i wanted that AI to work in the best interest of Russia that is.

Also i thought they already had a thing that stops conquered worlds flipping? I heard that somewhere? They could apply that to sold worlds too

Reply #8 from GalCiv3 Forums Top

a starbase from someone else in the middle of nowhere void can be useful because now you have extended your range from that starbase you might have needed a starbase or two  just to reach the one you built, but now you have a huge swarth inbetween your old range & the starbase as well as beyond the starbase.

Reply #9 from GalCiv3 Forums Top

The Ai is still stupidly, laughably exploitable in so many ways diplomatically.  Mostly because it has no concept of what a 'good' and 'bad' trade is.   Remote starbases can have value, but it's entirely dependent on the situation of the game at the time.  That 'good or bad' at game situation X is really hard to predict or program in.  

That said, planets are how you win the game in any victory strategy (niches strats like 1 world godlike challenge excepted), so by default planets shouldn't be considered tradeable by the Ai.  Of course, that relies on shifting the focus from an Ai that doesn't try to win to an Ai that actually wants to win - something that as yet hasn't happened in this game.