What to do about War Score

[omniclast]

To anyone that's been following the LPs coming out, it's becoming clear that war score is broken and could be a big deal breaker for anyone looking to get back into BE with RT. Plus it's a big slap in the face for those of us who've been getting excited about it over the past few months.

In Quill's LPs, he twice tried to make peace with the AI, and both times the war score forced him to demand cities that he did not want and the AI wouldn't give him (likely couldn't give him, since in both cases the AI had to give up all its remaining cities). He got really frustrated that he couldn't change the peace terms to get the AI to take the deal, so he was basically stuck in a war of annihilation.

Beyond the obvious bug that war score shouldn't require AIs to give up their capital, it really highlighted some fundamental flaws in the system:

- The AI assesses whether to accept peace terms separate from its relative score. Even in situations where the alternative is annihilation, the AI will not capitulate -- which means war score doesn't actually make AI decisions any more transparent and they will still appear to be arbitrary from the player's perspective.

- The victor may have to accept peace terms they don't want. Quill didn't want the crappy cities the AI was forced to offer anyway, so he would have happily removed them from the negotiations. He couldn't, and for no justifiable reason.

- War score forces the player to demand too much. One of the big flaws people found with the old system was that AIs would give up cities in odd situations where they didn't seem to be losing that badly. Well, war score doubled down on that. Quill took two cities (technically three, since he lost one of the cities and retook it) and the AI would have had to give him two more for peace. That seems like a huge price to pay for a moderate victory from Quill.

- It looks like war score could be gamed. Quill took a city, lost it on the next turn, and then took it again. For war score this counted as two separate conquests, putting him at three city conquests total even though he only took two cities. Now presumably the AI also got some points for taking it back, so that may have nullified the difference. But if the points are based on the city size, then quill would have gotten more points each time the city flipped since the AI would be taking a lower pop city back. A clever player could deliberately flip the city back and forth to increase the differential and push the AI into worse peace terms.

----------------------------------------------

So, how should Firaxis fix the problem? The most common suggestion I'm seeing is that war score be changed to reflect the maximum possible peace terms, and the player has the option to remove items they don't want. There's two ways this could be interpreted (and I don't think either of them will work).

- War score peace terms become a "suggested" deal, and the player can alter them until the AI will accept them. This system doesn't make a lot of sense to me since we'll be back where we started before war score: adding and removing terms until you land on one the AI will accept. The "What would make this deal work?" button was actually a better system since you only had to click once to figure out the AI wasn't going to make peace with you for anything.

- War score peace terms are changed to reflect the most favorable terms the AI will actually accept. While this would be a more useful solution, it makes the war score pretty much pointless, since peace terms will now be based on the AI's non-transparent attitude instead of the score differential. The maximum *acceptable* terms could even require you to give up spoils when you have the higher war score. At that point you're back to making the "suggested" peace terms basically an automatic "What would make this deal work?" button.

- If we're going the route of a better-implemented "What would make this work" button, the best implementation I've seen so far is in Endless Legend. There are no suggested peace terms, but you have a bar below that shows you how close the AI is to accepting the deal and gives you a rough sense of how much more you need to offer to make it work. I like that system a lot -- it still leaves some uncertainty about how AIs will behave, and you can still occasionally get frustrated by obstinate AIs, but it makes the dealmaking process a lot more efficient overall. But again, there's no use for war score on this model.

I think you can see where this is going: I think Firaxis should do away with war score entirely. It was a bad idea badly implemented, and frankly it seems like trying to cut corners on the challenge of making the AI better at diplomacy. The real problems -- that the AI doesn't make peace when it's rational to do so (or ever), and gives up way too much for peace in bizarre situations -- need to be fixed by making the AI better at diplomatic decisioning, not some gimmick to make players feel like they're stomping their opponent.

In any case, Firaxis needs to deal with this, and quickly. It's a pretty big embarrassment that a flaw which prevents players from making peace with the AI in a lot of common situations got past play-testing, and it has the potential to create a horsehockystorm of bad reviews once the game is released, which is a shame since there are a lot of other really positive features in RT. It's probably too late to see war score fixed or removed in time for release, but perhaps we can hope for a post-release patch.

 

[Ryika]

No, war score is a great idea, it includes the curse of the war into the peace calculations like the old systems could not. It is clearly superior if done properly and I want them to do it properly. If you copy an idea from another game, copy it COMPLETELY or make your own system around it - don't just copy the most basic function and then do nothing with it, how stupid is that?

My solution is simple: Go full EU, period.
- Everything you could ask for in the old system should have a (scaling) "warscore-requirement"
- The Victor should be able to pick and choose choose if and what he wants to use his warscore for to take stuff from the other player
- If the Victor is ahead far enough the AI (and maybe also players) should be FORCED to accept a peace deal

There you go. You have a working warscore-system. Why the devs think the current system is a good idea however is beyond me. If it really goes live like that - even if the bugs that make it so the AI doesn't accept deals - people will hate the system.

 

[SupremacyKing2]

My suggestions:
1) Change the formula for how war score is calculated to fix the exploit of losing and then recapturing a city counting twice. Also, the war score should take into account more than just how many units/cities each side has killed/captured. It should also take into account relative military strength. If done right, the war score could be a useful metric to tell the player how well they are doing in the war. In fact, I would argue that is what the war score should be: a metric for showing which side is "winning" the war at that moment.
2) Give the players different options that would go up to the maximum allowed based on the war score. So players could offer more generous terms to the loser which would give the player a boost in respect. This would also allow the player more flexibility and would not have to get cities if they don't want them. It would also fix the bug of the AI unable to give up their last city.
3) Peace terms should have a lot more variety. Players should be able to transfer energy, science, resources, not just cities in exchange for peace.

 

[GenEngineer]

I don't think they should do away with war score entirely, but they should actually fix the system. Frankly, what Ryika suggested is exactly what it should be, which is why it's so galling that it's broken so badly when it would take so little to fix.

 

[SupremacyKing2]

I would add that a sufficient high war score should force the AI to accept terms even if those terms means giving up their last cities. After all, that is what unconditional surrender is. For example, in WW2, Nazi Germany was totally defeated so they were forced to accept the allies' terms and in civ terms, give up all their remaining cities to the allies. This would a great way of reducing the tediousness of warmongering when you are so far ahead militarily. The AI should realize when defeat is inevitable and be willing to give up its remaining cities and be eliminated. And that way the player does not have to literally capture every city which is tedious.

 

[KrikkitTwo]

I would add that a sufficient high war score should force the AI to accept terms even if those terms means giving up their last cities. After all, that is what unconditional surrender is. For example, in WW2, Nazi Germany was totally defeated so they were forced to accept the allies' terms and in civ terms, give up all their remaining cities to the allies. This would a great way of reducing the tediousness of warmongering when you are so far ahead militarily. The AI should realize when defeat is inevitable and be willing to give up its remaining cities and be eliminated. And that way the player does not have to literally capture every city which is tedious.

That would be ok IF there was a decent pacification cost. Otherwise that leads too much into snowballing. (after all even the tedious elimination uses up unit-turns that could be used against the next target).


I definitely think
1. It should be a way to Force peace
2. It should differ based on who is requesting the peace (if you propose peace you get less (presumably because you are doing it on your turn after you just killed a bunch of their units and they haven't counter attacked yet)..so I could declare war, kill a few of your units and then ask for immediate peace, but I would have to pay a lot to get it.
3. It should include an 'army strength factor' and a time factor
4. The proposer can Craft the peace deal with each thing worth a certain amount of 'war score points'
5. left over War score points on a peace deal give you free Diplo Capital (the other side doesn't lose it, you get them for proposing a better peace deal)

 

[SupremacyKing2]

That would be ok IF there was a decent pacification cost. Otherwise that leads too much into snowballing. (after all even the tedious elimination uses up unit-turns that could be used against the next target).

I agree. I am certainly not suggesting that a player could take just a couple of cities and force the AI to surrender his entire empire. But if you have a ridiculously high war score like 4000-4, and the AI is down to 2 cities, I think the AI should surrender rather than make you go through the motions of attacking each remaining city. This might not be an issue in BE but I can remember in past civ where you would be the undisputed victor but still have to hunt down that last size 1 city somewhere in the arctic circle that the AI built before the war would end. That was very tedious and unnecessary. I want the game to avoid that. And yes, captured cities should produce sufficient unrest and unhealth that the player would need to deal with before getting the full benefit of the captured cities.

 

[SupremacyKing2]

1. It should be a way to Force peace
2. It should differ based on who is requesting the peace (if you propose peace you get less (presumably because you are doing it on your turn after you just killed a bunch of their units and they haven't counter attacked yet)..so I could declare war, kill a few of your units and then ask for immediate peace, but I would have to pay a lot to get it.
3. It should include an 'army strength factor' and a time factor
4. The proposer can Craft the peace deal with each thing worth a certain amount of 'war score points'
5. left over War score points on a peace deal give you free Diplo Capital (the other side doesn't lose it, you get them for proposing a better peace deal)

I like it!

 

[Ryoga]

No, war score is a great idea, it includes the curse of the war into the peace calculations like the old systems could not. It is clearly superior if done properly and I want them to do it properly. If you copy an idea from another game, copy it COMPLETELY or make your own system around it - don't just copy the most basic function and then do nothing with it, how stupid is that?

My solution is simple: Go full EU, period.
- Everything you could ask for in the old system should have a (scaling) "warscore-requirement"
- The Victor should be able to pick and choose choose if and what he wants to use his warscore for to take stuff from the other player
- If the Victor is ahead far enough the AI (and maybe also players) should be FORCED to accept a peace deal

There you go. You have a working warscore-system. Why the devs think the current system is a good idea however is beyond me. If it really goes live like that - even if the bugs that make it so the AI doesn't accept deals - people will hate the system.

So in the end, their bad habit of making things worse through their half-assed attempts to improve them hasn't quite changed, uh?

This looks like a very game breaking issue

 

[KrikkitTwo]

I agree. I am certainly not suggesting that a player could take just a couple of cities and force the AI to surrender his entire empire. But if you have a ridiculously high war score like 4000-4, and the AI is down to 2 cities, I think the AI should surrender rather than make you go through the motions of attacking each remaining city. This might not be an issue in BE but I can remember in past civ where you would be the undisputed victor but still have to hunt down that last size 1 city somewhere in the arctic circle that the AI built before the war would end. That was very tedious and unnecessary. I want the game to avoid that. And yes, captured cities should produce sufficient unrest and unhealth that the player would need to deal with before getting the full benefit of the captured cities.

Well the size 1 city is OK with the current Victory mechanics (ie you've destroyed their victory wonder, and you have the capital for your conquest victory), elimination of the other side is not necessary (and is discouraged by warmonger mechanics)

So you can offer them peace and they survive and you get a lot of diplo capital

 

[Psyonif]

- It looks like war score could be gamed. Quill took a city, lost it on the next turn, and then took it again. For war score this counted as two separate conquests, putting him at three city conquests total even though he only took two cities. Now presumably the AI also got some points for taking it back, so that may have nullified the difference. But if the points are based on the city size, then quill would have gotten more points each time the city flipped since the AI would be taking a lower pop city back. A clever player could deliberately flip the city back and forth to increase the differential and push the AI into worse peace terms.

I am of the opinion that this is a feature, not a bug. This is how it should work. Going full Stalingrad on 'em, having the opposing side see the crater-strewn techno-boulevards, the ruined shells of biodomes, the streams of gene-bombed civilian refugees abandoning what was once a jewel in the crown of the enemy civilization... it is entirely legitimate that you should be able to pump up your warscore by turning their city into an attritional charnel house.

 

[Barathor]

Okay, I'm going to play the devil's advocate.

Is the only issue with the War Score system that AI's with only one or two cities left won't negotiate peace? Were all the examples of this in the videos during the later part of the game or were there earlier examples?

If all we saw were later game examples -- what if it's intentional? Maybe Firaxis wants you to finish them off if you're going to beat them that hard. They don't want you to cheese it, like in the past, where you just leave them with a single city and hopeless. This would be especially strong now because not only could you eventually continue to trade with them, but you'll be able to continue to get their powerful, leveled-up perks from them and you could also gain more diplomatic capital if they rent any from you (if they have DC saved-up before being obliterated) -- all while it's too late in the game for them to even have a chance and be a threat. They're reduced to a harmless cow that everybody else can milk.

Above, where I was asking about early vs late game examples is because maybe there's a threshold built into the system which triggers and makes an AI fight until the death.

Earlier in the game, this would definitely be a problem. A lot of times you're not warring to wipe out a civ, you just want to grab some territory and weaken them; you don't have the army to wipe them out. But, if you're forced to continue attacking to the end, it could really hurt you. Plus, earlier in the game the AI can still have a chance at getting back on their feet and becoming a threat. Plus, traits probably aren't that plentiful or leveled-up yet either.

- - - - - - - -

I was going to suggest maybe some type of capitulation later in the game, where an AI with no chance of holding off the attacker and making a comeback just surrenders and gives you their last one or two cities to save you the trouble. Though, that would also save you time or rather, turns, and turns are valuable. Wrapping up a war and getting back to things a few turns sooner could mean the difference between a win or loss in a really tight race.

Also, another thing against capitulation, what if you're not the attacker and it's another AI beating on an AI that's useful to you? If peace isn't allowed later in the game when you're reduced to very few cities, you wouldn't want them to just give up and hand over all their remaining cities before you can get a chance to come to the "rescue".

- - - - - - - -

Finally, realism aside, maybe we're to accustomed to the old ways where you war and then you always settle for peace. Peace is always the end result, unless you wipe them out. If you obliterate another civ and don't wish to finish the job, maybe you have to accept that they'll be hostile towards you for the rest of the game, like another hopeless barbarian faction on the map that you have to deal with. Though, it would be best to finish them off, since they're probably still trading with the other civs and giving them some nice bonuses.

 

[KrikkitTwo]

- It looks like war score could be gamed. Quill took a city, lost it on the next turn, and then took it again. For war score this counted as two separate conquests, putting him at three city conquests total even though he only took two cities. Now presumably the AI also got some points for taking it back, so that may have nullified the difference. But if the points are based on the city size, then quill would have gotten more points each time the city flipped since the AI would be taking a lower pop city back. A clever player could deliberately flip the city back and forth to increase the differential and push the AI into worse peace terms.

.

Except it can't... lets say a city is initially pop 32 and loses 1/2 of population in attack and is owned by PAC

ARC takes city
city: 32->16
ARC: 32
PAC: 0

PAC takes city
city: 16->8
ARC: 32
PAC: 16

ARC takes city
city: 8->4
ARC: 32+8=40
PAC: 16

PAC takes city
city: 4->2
ARC: 32+8=40
PAC: 16+4=20

ARC takes city
city: 2->1
ARC: 32+8+2=42
PAC: 16+4=20

ARC is always ahead, but they WERE ahead by 32, and now they are only ahead by 22


I do agree that sometimes it would bake sense for the thoroughly beaten AI to just become a barbarian faction, you can give them peace, but they won't surrender any of their few remaining cities, and they will sanction you... and immediately redeclare war

 

[GenEngineer]

Okay, I'm going to play the devil's advocate.

Is the only issue with the War Score system that AI's with only one or two cities left won't negotiate peace? Were all the examples of this in the videos during the later part of the game or were there earlier examples?

If all we saw were later game examples -- what if it's intentional? Maybe Firaxis wants you to finish them off if you're going to beat them that hard. They don't want you to cheese it, like in the past, where you just leave them with a single city and hopeless. This would be especially strong now because not only could you eventually continue to trade with them, but you'll be able to continue to get their powerful, leveled-up perks from them and you could also gain more diplomatic capital if they rent any from you (if they have DC saved-up before being obliterated) -- all while it's too late in the game for them to even have a chance and be a threat. They're reduced to a harmless cow that everybody else can milk.

Above, where I was asking about early vs late game examples is because maybe there's a threshold built into the system which triggers and makes an AI fight until the death.

Earlier in the game, this would definitely be a problem. A lot of times you're not warring to wipe out a civ, you just want to grab some territory and weaken them; you don't have the army to wipe them out. But, if you're forced to continue attacking to the end, it could really hurt you. Plus, earlier in the game the AI can still have a chance at getting back on their feet and becoming a threat. Plus, traits probably aren't that plentiful or leveled-up yet either.

Any situation where peace is impossible is a bad thing. And we're being generous to assume it's only with only 1-2 cities left they won't - all scenarios have had them fail to accept peace. While the more recent builds have only shown peace after their left with a city, watch Quill's NSA LP and he points out a major break where the AI wouldn't accept peace period - it's starting to look like an issue they knew about for a while and never fixed. Likewise, the early vs lategame point doesn't work because all of the wars where we've seen this happen is in the early game. This is you take out a border city in the first 50 turns, want peace because you can't take the capital, and are tuck in the war for the rest of the game. If that's not fixed, then the entire game is badly broken.

The argument about preventing 'milking' of the AI would have some value, if there weren't a way of implementing it that doesn't specifically break what the system was designed to do - presumably, the AI can deny you deals in the same way that you can deny it to them. Just have them refuse to give you deals while their still angry at you for your aggression. We know that it's tracked and eventually goes away from how it works in vanilla BE and CiV, why not do that instead of breaking a system specifically designed to allow for good peace deals.

 

[Barathor]

...
Likewise, the early vs lategame point doesn't work because all of the wars where we've seen this happen is in the early game. This is you take out a border city in the first 50 turns, want peace because you can't take the capital, and are tuck in the war for the rest of the game. If that's not fixed, then the entire game is badly broken.
...

Then it is certainly broken.

Though, while fixing this, I hope Firaxis considers the "milking" factor for the later part of the game (not too late, because then it doesn't matter either). It's a different game now with the new diplomacy system and the trading of powerful traits, which don't care if you're an empire or a single city.

 

[omniclast]

A lot of these ideas are really good and Ryika and Krikkit have convinced me war score could be a good addition to the game, with some pretty big alterations. (Though it sounds like it would requires some judgment calls - like what is a relative score that makes sense for forcing peace? -- which I honestly don't trust the devs to make.)

Even if they are able to fix war score eventually, I don't think it would be realistic for it to be implemented in a post-release patch -- the devs need a fix to make the game playable for people buying RT, and they need it two months ago. I say yank war score for now, bring back the old deal mechanic, and then take more time to flesh out a war score system that makes sense. I'd rather see no war score system than the game-breakingly flawed one they have now.

 

[SupremacyKing2]

I say yank war score for now, bring back the old deal mechanic, and then take more time to flesh out a war score system that makes sense. I'd rather see no war score system than the game-breakingly flawed one they have now.

with the new diplo system, I don't think it is realistic to put the old deal table back in. It would require too much coding. For one, you would need to completely alter the UI to put the negotiation table back in. I think it would be best for them to change the war score formula to at least make the score work better and fix the bug of the AI always saying no. Then at least the war score mechanic would be playable and a bit less broken. Then in a few months they can release a full patch that makes bigger changes to the war score, to bring it to its fullest potential.

 

[Westwall]

My suggestion is to hold off on buying Rising Tide/cancel your pre-order until the devs actually confirm that the issue has been resolved, or otherwise release a patch sooner rather than later upon realizing that not doing so will directly affect sales if enough people boycott this expansion based on what's clearly a game-breaking feature.

Call me stark raving bonkers, but I think my plan would have the benefit of getting the most results in the least amount of time if people actually follow through with it.

 

[SupremacyKing2]

I think the solution that would require the least amount of coding for the greatest benefit would be to tweak the war score formula as follows:
- Have the war score only start after 10 turns of war. Before that, the war score would stay at 0-0
- Include relative military strength in the war score calculation
- Include war duration on the war score formula
- Change the peace terms so that the war score would give the winner energy, science proportional to the war score differential and only the highest war scores would give you 1 city max (this alone should fix the bug of the AI refusing peace if I am understanding what is causing the bug correctly).

This should fix most of the bugs and at least make the war score mechanic playable.

 

[MadDjinn]

That would be ok IF there was a decent pacification cost. Otherwise that leads too much into snowballing. (after all even the tedious elimination uses up unit-turns that could be used against the next target).


I definitely think
1. It should be a way to Force peace
2. It should differ based on who is requesting the peace (if you propose peace you get less (presumably because you are doing it on your turn after you just killed a bunch of their units and they haven't counter attacked yet)..so I could declare war, kill a few of your units and then ask for immediate peace, but I would have to pay a lot to get it.
3. It should include an 'army strength factor' and a time factor
4. The proposer can Craft the peace deal with each thing worth a certain amount of 'war score points'
5. left over War score points on a peace deal give you free Diplo Capital (the other side doesn't lose it, you get them for proposing a better peace deal)

Since units and cities have a warscore value, you could easily have the ai calculate it's own potential loss value (sum it's cities and units) vs an opponents current war score and if there is at least a 10x difference, then the ai should capitulate when asked. Now Ofc, this also requires some time in the war and actual combat losses.

And on that note, if the war was only unit vs unit and no city was ever attacked, then no city should ever be given up. but Ofc, you'd have to include things that aren't cities in the list of things to make peace with.

I also don't think the ai should just give up everything (last city stuff if original capital etc). That's a snowball issue that shouldn't be allowed.

Also, the previous example of ww2 didn't end up with others owning Germany's lands, but rather split it and forced changes. That's quite different than giving away total control.