"Scared to Death" bug in V1.29

[cracker]

I have been monitoring this problem and can't possibly figure out why it should occur.

When it is the human players turn in the game, then we should not see the population of AI cities changing up or down, unless there is a direct cause that can be found in the human players actions.

Yet, sometimes people in the AI cities will just drop dead from fright and this occurs in between the moves made by the human player when technically the AI players should not be able to move or pop rush or do anything else that might explain the vanishing person trick.

The examples I have of this are cases where a town of pop 2 gets the crapola scared out of it just as we are about to capture the town and mystically one of its people vanishes. This cowardly fright death effectively denies us the ability to capture the town because a pop 1 city is destroyed while a pop 2 city would be captured with a residual population of 1.

Here is a frame by frame replay of one of these events:



If you watch closely you will see our first unit attacks the town of pop 2, but when the second unit crosses the river to attack the town one of its citizens just drops dead from fright even before the 2nd unit can attack. (Note: this wave of attacks eventually succeeds against the lone spearman defender and destroys the town so we know that the vanishing pop point was not just rushed to create an extra defender.)

I have save files to verify this if someone at Firaxis is still fixing bugs. Just PM me with an email address and I'll send it to you since the games cannot be posted here due to spoiler and tournament restrictions.

The bug existed in V1.21 and I have confirmed that it still exists in V1.29 for games that originate in V1.29 as well as games converted up from V1.21.

If I could find any logical reason for the action to be occuring during the human turn then I would not post this as a bug. I think it may be some sort of leaky implementation of some turned base logic or a flaw in the governor code that creates an instant starvation event.

 

[Gastric ReFlux]

There may be an explanation, but I'm not sure if I'll be able to describe it well.

Sometimes, I have observed that there can be a delay before the map has its information updated. For instance, I have noticed that when moving troops into enemy territory, the first unit uncovers the fog of war, but the map doesn't update to show the tile improvements until the second unit reaches the tile the first unit reached.

So what I'm guessing is that Fukushima was already at pop 1 at the beginning of your turn, but the delay of the map being updated kept the display showing 2. Then when your second unit moved into range, the map updates, and reveals that the city is down to 1 pop point.

So what you might want to do when you see this again, is backup to a save, then before attacking with that first unit, investigate the city. See if that would reveal the town's population.

 

[JustListenen]

Every city I sucessfully attack loses one population point; but I can't remember (from last night) when it occurs, after the first attack or the last attach when I capture the city.

IF the city has any culture buildings in it, the Pop 1 city will not be destroyed. I have verified this in my own games.

 

[Pygzilla]

but look at the map. If it has culture it can't be much...no expansion.

 

[Gastric ReFlux]

Originally posted by JustListenen
Every city I sucessfully attack loses one population point; but I can't remember (from last night) when it occurs, after the first attack or the last attach when I capture the city.

IF the city has any culture buildings in it, the Pop 1 city will not be destroyed. I have verified this in my own games.

The loss of a population point occurs with the attack that takes the city.

 

[Aeson]

Could you post the save? Gastric may be right that it's just slow updating. The AI just about always will pop rush a size 2 town if you are at war with it and within 2 tiles. It probably pop rushed a Spearman during it's turn (why else would it have 2 in the city, and 1 just left?), and the number just didn't update for some reason.

 

[cracker]

GR and Aeson,

I thought of the slow update thing also, but I verified the save file is at pop 2 and double checked this by saving the situation and reloading it and playing it through about 15 different ways.
I even went back two turns and replayed it forward varying the the number and type of units in the assault force from 1 up to 10 units.

Here's a screen capture of the town produced by not attacking the town but just advancing the units into all the surrounding squares and fortifying them in place. The FOW is also set to OFF in this screen capture so that the town and all units and improvements are fully revealed.



The pop point does not evaporate even when you send 8 or 10 units into the adjacent squares just as long as you do not attack the town. If you attack the town at any point in time, then the next advancing unit cause the person to evaporate (ie. move in 1 and 2, attack with 3, the the person dies of fright with 4).

I can't post a save file until after the 2nd of August because this extract is from GOTM9. PM me with an email and I'll send you one of the versions privately.

 

[phishphanatic]

I don't think this is a bug. The same thing happened in Civ2. I think the intent is that as in real life, if one attacks a city, there is chance they will kill civilians. I have found that it seems to be random when the city loses a pop point. Either that, or once you build a certain improvement (granary or maybe barracks), you don't lose pop points if attacked. I have attacked many 1 size cities in modern times, and rarely had it dissapear. Now that I think about it, there must be a certain improvement that prevents that pop loss, I just don't know what it is.

 

[Aeson]

I just looked at the save, and that is weird! Investigating the city shows 2 pop as well. Even if you don't attack the city (attacking the wandering spearman instead) it loses pop... has to be a bug.

 

[Gastric ReFlux]

Hmm, still could be a problem--we're still subject to old information maybe? What I'm trying to say is that there may be two areas of information: what we know versus what the game state actually is. So let's say that Civ3 is coded so that there are at least a couple of separate datasets. In the one set, it contains the information about what the human player knows. The other set has what could be called the absolutely current information. The question is this: when does our dataset get updated information from the absolutely current information?

If there are cases where that's delayed, as I suspect, that would mean an investigate city could possibly be gleaning information from the player's dataset, not the current information set. Which could be buggy coding, because one would think that using investigate city should reveal the information from the current info set.

The only way I can think of now to check for this possibility would involve hacking into the savegame, and looking for Fukushima's population points that way.

 

[cracker]

Gastric,

We have "hacked" directly into the save games and verified the pop count externally, so I think we can put that guess to rest.

A key point here is that the pop point eveaporates in the middle of the human players turn and in response to unit movements that do not do anything that would cause the person to die (ie. the person dies NOT due to an attack or bombardment, or by blocking a key food resource).

Our military forces just scare them to death.

I have seen this on a number of occasions in the past because of making targeted strikes against pop 3 and pop 2 cities. This si just the first set of exampels where I have the bug really well documented and bracketed by lots of knowledge abou the target cities and civs.

This bug even occurs if you do an attack and then make certain moves in another completely diffent part of the map.

Firaxis has the "Debug executable" so they can view the actual transactions as they occur to see why the people are dying of Fright. Until then we are just on the lookout for other examples that may be getting missed because of how this occurs in the game.

Send me a PM with an email and I will mail you a save file. Since I am not too familiar with you personally, I will need some assurance that the save will be guarded because it is related to the currently opem GTOM( that will not close until Aug 2.

 

[Lt. 'Killer' M.]

cracker:

just to make sure: did you save the game when it shows pop2 and the units stand next to it, reload, and it still shows 2???

I am sure this is really a bug, just being the usual pain in the ass here....

 

[Bamspeedy]

Well, I just had this happen in my HoF game. I didn't even have to attack. Just declaring war causes 3 cities to lose a population point DURING THE HUMAN'S TURN!

Either they are scared to death of my single mounted warrior, or they are poprushing out of turn!

Here's the save file, just go to the diplomacy screen and declare war. After you move the cursor, the cities will drop in size. Before declaring war you can investigate cities and see that they did have a population of 2. I had just updated the map by trading maps, but you can trade maps again to verify.

http://www.civfanatics.net/uploads/scared_bug.zip

 

[cracker]

Bamspeedy,

Glad you could verify the bug.

I have been doing some further testing and see that this "scared to death bug" is definitely somehow related to declaring war.

In my first examples, I thought it was related to the attacking events, but now see that it must be tied to the outbreak of war and the next human move after war is declared.

I can find no possible programming reason where the population points in enemy cities should be vanishing during the human turn. If the points were being used to rush an improvement or a unit I might view the "scared to death" feature as some undocumented AI cheat. But since I have used the editor functions to verify that the people just vanish, I definitely think this is a bug.

One of the reasons we may not have noticed it before is that in many cases we expect the populations to be effected by warfare. It is just that in this case the effects seem to be happening way outside of the expected sequence of events and in cases where there would be no real logical benefit other than to randomly disrupt the game out of sequence.

 

[Agamemnus]

More buggy code from Firaxis. Amazing.

I shall venture to point out the exact cause of this bug:

1) A pop loss function call has been put into the AI script of what-to-do-when-player-declares war, <i> without </i> adding a defender. Perhaps the defender <i> is </i> added, but then removed due to some new rule <i> out of turn </i>.

The new rule could perhaps be forcing the AI units to disband if there's not enough money in the coffers, although there is, I think, an example to counter that argument.

2) It is, however, unlikely. More likely it is a game mechanics bug more than an AI bug.. perhaps the new rules treat size 2 cities as if they were size 1 cities about to die..

This could be explained by only counting worker citizens (not entertainers, taxmen, or scientists) in figuring out pop loss. Which would be a glaring bug that passes all the tests of Firaxis programming stupidity.

3) The city is next to a flood plains.. perhaps the loss of a population point is somehow tied in with the attack..

Bamspeedy, if you could perhaps characterize those three cities... how many entertainers, are they on flood plains, any other characteristics in common?

 

[SirJethro]

I've noticed this too, but I don't have proof that I can post. There have been times when I'm certain the AI has either pop-rushed or drafted out of turn when I declare war.

Anyone considered using "Investigate City" in this example, both before and after war is declared or the attack happens?

 

[Bamspeedy]

Bamspeedy, if you could perhaps characterize those three cities... how many entertainers, are they on flood plains, any other characteristics in common?

All 3 chinese cities were size 2, before dropping to size 1. One of them was not on flood plains, but did have jungle. Terrein isn't the issue, as I have figured it out, the AI is poprushing out of turn! Investigate the city before declaring war. They have 1 spearman in their garrison, with 4 shields being used towards a settler. Next turn attack. You'll injure the spearman, but now there is another uninjured spearman in the city.

This I guess would hurt the semi-exploit of being just outside their borders, declaring war (saves your rep since you aren't in their territory), but still able to attack their city the same turn before they can respond with more defense.

 

[Lt. 'Killer' M.]

Bamspeedy: if you declared war during your off-turn phase (i.e. AFTER you press end-turn and BEFORE you get your first city-pop-up, then this is nothing new, this simply is the result of the stupid turn-order.

It is NOT as we all expect:

human science turn, human production turn, human move turn
AI1 science turn, AI1 production turn, AI1 move turn,
AI2 science turn, AI2 production turn. AI2 move turn....

but
human science turn, AI1 science turn, AI2 science turn
human production turn, AI1 production turn, AI2 production turn,
human move turn, AI1 move turn...

So, if you delcare war when an AI pops up and ask you to leave or offers a deal, they can rush beforee you move......

same goes for flipping cities, since they flip iun your produciton turn, but the AI can do things there rifght away, they can rush and get the unit in their production turn...



Could this be it here?????

 

[SirJethro]

Originally posted by Lt. 'Killer' M.
Bamspeedy: if you declared war during your off-turn phase (i.e. AFTER you press end-turn and BEFORE you get your first city-pop-up, then this is nothing new, this simply is the result of the stupid turn-order.

It is NOT as we all expect:

human science turn, human production turn, human move turn
AI1 science turn, AI1 production turn, AI1 move turn,
AI2 science turn, AI2 production turn. AI2 move turn....

but
human science turn, AI1 science turn, AI2 science turn
human production turn, AI1 production turn, AI2 production turn,
human move turn, AI1 move turn...

So, if you delcare war when an AI pops up and ask you to leave or offers a deal, they can rush beforee you move......

same goes for flipping cities, since they flip iun your produciton turn, but the AI can do things there rifght away, they can rush and get the unit in their production turn...



Could this be it here?????

Killer....great post. I was aware of some of this, but not all of it. That explains a lot of what I was considering "cheating" (like attacking a city that flipped at the beginning of the turn and finding two conscripted riflemen).

It wouldn't be so bad if they would make these handoffs from human to AI turns more visible and do more to make it clear what things the AI could do in each phase.

 

[Lt. 'Killer' M.]

there can be three defnders even - if they can also draft

also, i cannot see how this is supposed to wrok in PtW