Solver's unofficial BtS 3.17 patch

Linking to who has the tech would be better in real terms - but it wouldn't be working quite as well for the AI. The less assumptions you have to make, the better, so the current implementation would probably lead to better AI performance overall. Like most AI stuff, it has a random element to it and is seriously dependent on personality.
 
Would the implementation of a negative diplomacy modifier for "disposal" of nukes already built after having the resolution passed require a lot of effort? I think this is really needed to reduce the counter-intuitiveness of this whole thing.
 
You know, I'm not a fan of that just because I think the entire system of diplomatic modifiers for nuke use is wacko anyway. But on the upside, here the description at least does what it says: "can not build nukes". Rare enough for the description to actually be what happens ;)
 
That wouldn't really work, either. Vetoing hurts and isn't supposed to be always necessary. Also, even after you've built the Manhattan, the ban still hurts. If all you can build is 1 ICBM before nukes get banned, you're stuck with that 1 ICBM, the ban also prevents construction of any additional nukes.

Well, it 'works' in a ham-fisted way, but isn't very elegant. I just thought it might have been the intention of the developers to discourage developemnt and use of nukes across-the-board.

Again, I don't agree with it, but I could see it being a 'legitimate' game design choice. After all, nukes aren't absolutely necessary to win the game, either.

But I am interested to see how your probability-based AI ban decision script works. :goodjob: I definitely don't want the final system to be too complex. That's always a recipe for disaster.
 
Does SDI affect tactical nukes as well? If not, then the SDI effect Solver is proposing doesn't work too well. In any case, it seems to have a rather large effect.
 
Linking to who has the tech would be better in real terms - but it wouldn't be working quite as well for the AI. The less assumptions you have to make, the better, so the current implementation would probably lead to better AI performance overall. Like most AI stuff, it has a random element to it and is seriously dependent on personality.
I see what you're saying.

I guess the concern I have is that there definitely is a difference between someone who can build nukes but hasn't yet, and someone who can't build them at all.

Wodan
 
Oh, if the AI can build nukes but hasn't yet, chances are that it won't stay that way for long and the AI will get its paws on some nice exploding warheads ;)
 
At any rate, version 0.19 is coming up. Please download before throwing rotten bananas at me ;)

  • AI plane/ship retreating logic when city is in danger extended to Forts
  • Vassal's spies no longer caught in master's territory
  • AI logic on banning nukes tweaked to account for AI leader personality and the global situation
  • Undid galley production changes from Better BtS AI
  • Non-native collateral damage tweaked (thanks guys)
  • Civilopedia will now display "replaced by" lines for buildings that get replaced by UBs
  • AI guided missiles will lean towards destroying resource-producing tiles as opposed to improvements like Towns
  • Damaged AI attack planes may choose to continue attacking if no defending interceptors are around
  • Rivers running through deserts will result in floodplains
  • Start location tweaks per SevenSpirits
  • Workers not as lazy, per Bhruic's patch
  • Glance screen reactivated

EDIT: Some of you requested that I carry over the "stolen techs can't be traded under no tech brokering option" change - turns out that was already included in 3.17, just not documented.

I have also received a request for a 50-civ version. Here is the dll file which is tweaked for 50 civs - download version 0.19 and then replace the dll with this one. Well, the ones who wanted this know what to do ;)
 

Attachments

  • Vassal's spies no longer caught in master's territory

Not throwing bananas, but what's the reasoning behind this change?

  • Civilopedia will now display "replaced by" lines for buildings that get replaced by UBs

Cool!
 
Not throwing bananas, but what's the reasoning behind this change?

Because it was in BH's patch and people demanded it :D It did lead to annoying situtations - your vassal might want to get his Spy somewhere else, but would never be able to make it through your territory if your espionage security was considerably better.
 
There can only be 18 civilizations in a game normally, that DLL allows there to be up to 50 civilizations. It is mainly needed for compatibility with certain custom scenarios that add a huge amount of civilizations to it.
 
I am going to go with logic for AI nuke bans for now:

[The old logic stays, AIs that really want to get nuke happy won't ban and AIs that are behind will not ban]

The AI will look at whether it has the SDI and will consider its own personality. Has SDI - less likely to ban.... errm, what the heck is up with SDI having 75% interception probability? I could swear they wanted to make it 50%. The more the AI likes building military units, the less likely a ban. Warmonger AI personalities always less likely to ban. If the game is set Aggressive AI, less likely to ban again. As an additional consideration, if the AI civ has built some nukes and is the only civ in the world with the SDI, it's much less likely to ban.

Without modding the leader personalities, here are some values.

Gandhi, no SDI: will ban 85% of the time.
Gandhi, SDI: will ban 60% of the time.
Monty, no SDI: will ban 30% of the time
Monty, SDI: will ban 5% of the time
Ragnar, no SDI: will ban 20% of the time
Ragnar, SDI: will never ban

With Aggressive AI, they become considerably less likely to ban.
Aggressive Gandhi without SDI will ban 70% of the time, with SDI he'll ban only 20% of the time. So if they have the SDI, most leaders won't ban under Aggressive AI.

I have a question as to how the decision is made.Does the AI make a random choice that's modified by the factors you listed, and in that case does it make the choice again the next time the vote comes up or does it only change it's mind if one of the factors has changed? Or does the AI have a specific number of "pro ban" factors it has to reach and being above of below that line determines it's decision?
 
It's a random decision influenced by those elements. So in my examples, Gandhi has an 85% probability of voting to ban, though there's a 15% chance that he would vote not to ban. I have no idea what happens if the vote is brought up again... can it be reversed anyway? But the AI will make a new choice then, with those same probabilities.
 
There can only be 18 civilizations in a game normally, that DLL allows there to be up to 50 civilizations. It is mainly needed for compatibility with certain custom scenarios that add a huge amount of civilizations to it.

Ah, that's what I thought, but I wanted to make sure, because 50 civs is a lot in a game. :eek:
 
Because it was in BH's patch and people demanded it :D It did lead to annoying situtations - your vassal might want to get his Spy somewhere else, but would never be able to make it through your territory if your espionage security was considerably better.

How do we know the vassal simply wasn't spying on his master?
 
Back
Top Bottom