VP not finished yet: The AI can't win by domination

[glider1]

I've tried everything. With any version of VP including the one that is already been officially released back in May, the AI cannot win by domination unless you set up the most trivially easy map for it.

The closest I got was with the latest version 9-15. It was an amazing thing to witness. Here are the test conditions.

• Duel size frontier map continents
• Babylon verses India
• advanced start information era
• domination only victory condition
• quick speed
• everything else default settings
• no mods

Babylon and India slugged it out beautifully. They colonised and fought over the colonies. Babylon was simultaneously attacking the colonies while at home it took an Indian city and a citystate adjacent to the Indian capital by naval attack. It nuked the capital and moved an airforce into the frontline cities and hammered the Indian capital really hard with Armour, death robots, XCOM and stealth.

But here is the problem that happens every time no matter whether the map is duel or huge.

The game crashes before the domination win happens. For some reason Babylon could not take the capital and then turtled for over a hundred turns until the game crashed......

You can't load the save game either.

This is on default settings, no mods.

I have tested it on autoplay probably more than 30 times in all sorts of different ways.

VP always CTD.

HERE IS THE WEIRD PART:

If I turn off VP and run the community patch only, the carpet of doom in vanilla has to be seen to be believed it is so horrible because there is no unit supply cap......but at the same time the game does not crash....it is just that the map is carpeted in units with an AI that can make no progress in domination through the deadlock.

But the community patch is rock solid.

So what is going on with the VP mod and domination? If there is a memory problem, why does the game engine not crash with a carpet of doom when every tile is taken up with a unit, yet in VP with the beautiful small armies thanks to the supply cap concept.....it crashes and cannot be reloaded.

How can we say this mod is finished if there is one whole victory condition that might as well simply be turned off? Is it because of the burden of the other VP infrastructure not present in the community patch? I cannot see how that could be true.

It is easy to test. You just autoplay two civs on a duel continents type map using the conditions I listed. Takes 15 minutes.

TEST SETUP:
Windows 10 16GB i7700k on a 1080GTX

 

[Moi Magnus]

About crash: If you ahve time, please make a bug report on Github (https://github.com/LoneGazebo/Community-Patch-DLL/issues), with the CvMiniDump file generated by civ 5 when crashing. This is not expected behavior. (and though that's not the first time I've heard about it, it was never really reported without being together with mod conflict issues)

End game isn't as tested as the remaining of the game, since most game finish when the player win or lose. (and auto-test are made starting in ancient era, since VP is not balanced with latter era starts).

And lastly, the mod is not finished. The mod is:
+ Feature complete (up to what is needed for balance reasons)
+ On the way to gold edition for balance (but still heavily discussed)
+ Still with bugs.

 

[Gazebo]

The game can CTD for a number of reasons. Keep in mind that the AI carpet of doom often ignored 90% of its units. In VP we run simulations for positions (often into the thousands) for the whole army every turn before deployment.

G

 

[Gothic_Empire]

The AI can't think. That makes something complex like capturing a specific city to win the game hard. The other victory conditions are buckets to fill. Conquest is much more complicated.

 

[tommytoxen]

I play VP alongside dozens of other mods and have never ever crashed. After about 6 hours of playing I start getting graphical glitches though, but restarting the game fixes that.

What map are you using? It's either your map script, GPU or you have a memory fault.

If I try to start on a small map with too many city states I crash, but that happens even without VP. Going a map size larger always fixes it for me.

Also I really don't understand your title, nobody has ever claimed 'VP is finished', that's why they still do beta releases and updates..

They can only do so much with the AI. Without VP it just sits there drooling down its self. VP is practically the only mod that makes the game playable and enjoyable.

since most game finish when the player win or lose.

I love to play right into the future most of the time i turn all victory conditions off lol

In VP we run simulations for positions (often into the thousands) for the whole army every turn before deployment.

Wow I actually never knew that, that's awesome!

 

[chicorbeef]

The AI can't think. That makes something complex like capturing a specific city to win the game hard. The other victory conditions are buckets to fill. Conquest is much more complicated.

https://forums.civfanatics.com/threads/monty-prince-ai-domination-victory.629024/

 

[SpankmyMetroid]

I have noticed in the last few weeks of patches that virtually no AI capitals have been traded. I’ll admit there’s definitely some inherent bias on the player’s part to harry domination leaders over other leaders, since they’re the most salient threat and are easiest to snowball, so it’s difficult to observe in a typical game. It’d be interesting to hear Gazebo’s observations on what kind of victories the AI are typically achieving.

Are you also testing with randomized VC’s on, Gazebo? I wasn’t sure if you thought it should be a ‘standard’ setting for VP and are balancing around that.

 

[glider1]

Thanks for the suggestions but if it were my hardware, then the community patch simulation should also crash but it never does it is rock solid.

Also, the same AI exists in the community patch as in VP, so that eliminates the AI as a cause.

It is likely to be something unique to VP other than the AI. My tests aren't exhaustive but I have run each case a few times some cases dozens of times.

VP might run through to completion in the test setup of the OP on duel/tiny map but I have never had it run through on bigger maps than that.

I'm reluctant to post it to github because there are still too many variables.

But the point remains. The AI doesn't win by domination. It may have in February. But it doesn't from the May release onward.

At the end of the day I guess all this doesn't matter if people are playing the game normally and it is not CTD for them.

The AI can't beat itself at domination, but it still can stomp a human with it.

I'm just trying to imagine is getting the DLL to a point where it is so robust that it can perform an amazing variety of possibilities for future mods.

 

[Moi Magnus]

I'm reluctant to post it to github because there are still too many variables.

The CvMiniDump file generated by the crash allows you to track the reason of the crash. If moreover you have a setup to reproduce the issue without mods other than VP, then that's worthy of a post on GitHub. Even if this post is closed with a "cannot be fixed".

 

[Gothic_Empire]

https://forums.civfanatics.com/threads/monty-prince-ai-domination-victory.629024/

I didn't say impossible. I said hard.