Testing the Current Diplomacy Changes

[sven]

I'm working on a number of different inter-connected updates to diplomacy, and I've just pushed my current working draft (r37620) out to the test servers.

For starters, there are now a number of different causes that may motivate a declaration of war. All of these causes take faction attitude into account, so AIs should be far less likely to suddenly start a war with someone they have good relations with. Conversely, low reputation now has a much higher chance of translating into open hostilities.

AI's are also far more likely to stop fighting wars that aren't going terribly well for them -- which I'm hoping will have secondary benefits for mid/late game balance, as it should now be much less likely that AIs will exhaust themselves fighting unproductive wars. (Though the "sue for peace" events themselves still need work to better balance the terms of peace treaties.)

And I've also made a number of changes to the way that alliances factor into war/peace behaviors.

There are still many bigger and smaller TODOs on my list for overhauling diplomacy-- but, some feedback on how the current system is playing would be helpful. So, if anyone would like to checkout the latest dev build, play through a game, and let me know what feels like it's working, and what's feeling off, I'd appreciate it.

[zolobolo]

Great work so far on the changes:
- There seems to be a notecable drop in perfromance but it is worth it when considering the below:
- Alliance actually matters now: joining war on side of the player and dragging the player int owar both works well and is a huge boost to immersion: great change
- After ally has joined the war of the player they correctly dispatch forces to assist them and have sent the ships to a logical defensive position where player fleer has been amassing
- Border tention is cool and hop to see it dynamically evolve but it seems to disappear once alliance is established between the parties - is this intended?

One issue I say so far is a missing variable in the below instance:

[sven]

- There seems to be a notecable drop in perfromance

Yeah, I'm not thrilled about this. Between merging in some of harpy eagle's more sophisticated mods, and adding a whole bunch of more complex diplomatic logic, end-turn times are really spiking. I'm going to do some profiling to try and optimize stuff; but, realistically, I think it's probably inevitable that game is going to get a bit slower as a side effect of all the new logic that's going in.

- Border tention is cool and hop to see it dynamically evolve but it seems to disappear once alliance is established between the parties - is this intended?

Yes, it's intended. (Not sure it's actually a good idea -- but, my theory is sortof that once you trust each other enough, having a long border stops being a problem. Um, kinda like with the US and Canada.)

One issue I say so far is a missing variable in the below instance:

That should be fixed in the next patch...

[zolobolo]

- Border tention is cool and hop to see it dynamically evolve but it seems to disappear once alliance is established between the parties - is this intended?

Yes, it's intended. (Not sure it's actually a good idea -- but, my theory is sortof that once you trust each other enough, having a long border stops being a problem.

I figured as much...lots of games use this logic and it does make common sence

But gameplay wise I would still consider rethinking: SiS games tend to have "only" a handful of factions on a map, and removing this tension between allies will cause the game to get stale very quickly after the early alliance wars (currently only an assumption). There needs to be some forces that pull apart alliances from the inside: border tention and voting for someone else in Galactic Council seems like the logical mechanics to take for this (third pillar could be population opinion based on your behavior against similar pops like bombing, enslaving)

As long as it is transparent I don't think people will mind and it will force you to not consider alliances as a done deal till a new game is started

Um, kinda like with the US and Canada.)

Ah there is your problem I am coming from a European country and can tell you that having a long border and alliances does NOT mean you will not be going into war with your neighbor

What if alliance would cap the amount of border tension in half instead? Though not sure you can display this transparently
In the end, if you end up with only two allied superpowers that have a looong common border, they should still look at each other with suspicion to set the stage for an end-game war. The player would not have the feeling of able to get away with reckless colonizing and conquering into and around allied territory without any diplomatic tension penalty

[zolobolo]

First time I have engaged a battle with an ally and started on the same side: humans and ally: Imperials against Yoral
Ally sends proper forces to assist player

The common starting position and the changes to tactical AI worked perfectly: ally was moving smart and efficiently: small crafts reserved, shuttles capturing enemy ships, it was a blast

Fighting alongside allies is the cooles I have seen in any game so far - I am not sure if the positioning of ally ships next to each other is new to this change since it was almost impossible in the past to get our ships together but works realy well

[zolobolo]

Additional findings for mid to early-late game:
1. AI does not yet sue for peace even after loosing almost hteir entire fleet and most of their planets while being the agressor who declared war on the player due to low military might - maybe the casus belli should not play a role but euther when losses are considerable proportionally or attack capacity is nullified it would make sense to trigger it
2. AI seem to be prioratizing assistance to allies over senf defence when it comes to sending reinforcements. Have seen this in a huge map of 170 stars: Yoral were sending the bulk of their fleet to help out their ally on the other side of the galaxy (17 Turns travel time) while suffering heavy losses and loosing planets in the region from where the ships have been sent from leaving the regaion defensless
3. Might be worth adding a relathionship penalty when ally does not assist an attack
4. Please make the autocombat option available for prompts to assist ally: currently the battle screen needs to be initiated in all cases (workaround is clicking autoturn there untill battle is over)

[nweismuller]

OK, so, I have to say watching how the diplomatic game unfolds has made for a much different behavior in known space. My current game, space has polarised into two major alliances clashing in a savage war to the finish, and my own Phidi sitting on the sidelines and trading with both sides... and being resented for their neutrality. I like this. This feels a lot more convincing already than earlier diplomatic behavior.

[wminsing]

Oh this all sounds extremely promising; I will try to fire up a game today and see if I have any useful feedback.

-Will

[nweismuller]

Not sure when this change came around, but just noticed the Gremak actually enslaving conquering populations in my latest game! This has been something I've wanted to see for a while, so nice.

[harpy eagle]

Sounds really promising! Won't be able to try out the new changes for a bit (in the middle of moving house, and some projects on the go) but I'm really looking forwards to playing my next game.

[zolobolo]

This is why I think there needs to be some sort of errosional force for alliances:

Current mechanic results int oalliance war untill mid game with 2-3 sides duking it out. One side will innevitably be victorios before the mid-game is over, mupping up the remnants.

If existing alliances are static, players are left to win the game withouth their allies recognising that they are in a competition and not "fighting" back - this will yuickly become boring for the player and end-game is unlikely going to be played (colonising, building up and ruh-attacking weak allies if necessary is basiaclly mopping up)

If alliances are fragile hovewer, as soon as the initial alliance war is over, the winner side can quickly become unstable withouth a common foe to fight against and the resulting chaos serves as the endgame

Already mentioned mechanics to make allainces fragile are:
1. (High) Upkeep cost for alliances in influence
2. Increasing border tension based on leght of common border - with the asumption that these graudally increase among alliance members
3. Large relationhip penalty if voting against alliance member if they have more pop (This is probably the best way to emulate the will to "win" for the AI as it serves as a soft mechanic to make the AI "aware" of the competition that serves as the key motivating factor for player behaviour)

[zolobolo]

When player invades human refugee colony and conquers it, the allies of the player show up with a message asking the player to join their war against the already conquered refugee colony:

[harpy eagle]

3. Large relationhip penalty if voting against alliance member if they have more pop (This is probably the best way to emulate the will to "win" for the AI as it serves as a soft mechanic to make the AI "aware" of the competition that serves as the key motivating factor for player behaviour)

This one in particular stands out at me because it applies some erosional pressure on the alliance, keeping it from getting stale, while also giving the player some agency.

I think the trickiest part of handling how AIs break alliances in any 4x game is that depending on how it is done, it can feel one-sided for the player.

The best approach IMO is to give the player control over the matter with some kind of cost/benefit decision they can chew on.

[zolobolo]

The best approach IMO is to give the player control over the matter with some kind of cost/benefit decision they can chew on.

For this case my suggestion regarding alliance upkeep cost would be fitting

Cost would be of course influence and would need to be considerable enough so that one allied empire cannot produce enough influence to maintain it (otherwise all alliances would be self-maintaining). Tying it to relative pop strength would also do the trick of scaling: AI empire with 2X the pop would need 2X as much influence to keep in an alliance

Benefits are obvious and rectify the costs:
1. If they are stronger- they support their ally with effective military might
2. If they are weaker, they will vote for the player on the galactic council

[zolobolo]

Turn times are around 1:50 on a map with 170 star systems in mid game

Performance hit is not actually measerable on the system level, so it could lie in the complexity of the lagorithms that take that much time ot run their sequental course?