Overextension warnings

[Nelphine]

I don't think turn count matters. I think number of enemies matters. Turn 200 for a slow strategy could be as relevant as turn 60 for a fast strategy. If you've taken out 2 wizards, the remaining ones are still limited in how fast they can expand no matter what turn it is.

However, we don't want people rewarded for keeping one AI alive with one city. So it can't actually be based on number of players. So it's based on assumed number of players, which is simply land size.

Or to put it in reverse - this feature also inherently makes it difficult to stay at peace when you are the last one on your plane fighting the myrran AI. That should be kept.



I do agree that charismatic aura of majesty wizard pacts should certainly overcome this penalty.

[Seravy]

I think this should be the threshold : 

(4+(turns-12)/10)*(2+land size)/4 cities.

For Fair land size, this means the following amount of cities :

1401 4
1402 5
1403 6
1404 7
1405 8
1406 10
1407 11
1408 12
1409 13
1410 14
1411 16
1412 17
1413 18
1414 19
1415 20
1416 22
1417 23
1418 24

For Huge it's 50% more, for Tiny it's 50% less.

Turn count is definitely relevant. We want the penalty to happen if the player controls more territory than what we assumed they should be holding when we did all the economy calculations. On the other hand, if the player plays a slow strategy and doesn't hold many cities, they are not a threat the AI needs to attack. Basically, fast strategies are more powerful due to the snowballing nature of the game, so they need to meet stronger enemy resistance to drain more resources from the player.

If we do go with this city amount, we'll also need to rethink the penalty amount. I think this is functionally equivalent to the lower city count threshold in the other system I posted. The higher one could be the same but with (10+) where it has (4+) now.

Or to put it in reverse - this feature also inherently makes it difficult to stay at peace when you are the last one on your plane fighting the myrran AI. That should be kept.

Maybe, but that would be "goal 2" in my above posts. It has nothing to do with actual overextension.
I never really found it hard to stay at peace with the last person - we had contact earlier, so the possible outcomes were usually

a. They attacked me and I lost, game already ended
b. They attacked me and I won, they are no longer the last "strong" person.
c. We remained at a prolonged peace due to a treaty or other diplomacy reasons - in this case the overextension will almost never be enough to escalate into a war.

[Nelphine]

For c) I disagree. I think that should still fall under my second last post (where I discuss what I think the goals are and how difficulty affects this).

However, in addition to that.

I'm not sure those numbers are good. For one thing lunatic AI will consistently beat the early numbers - even before counting neutrals. A lunatic ai can easily have 8 cities in 1402. 

Speaking of neutrals, while I agree this is a major source of snowballing I don't think we should penalise the player for getting lucky and having 6 cities in 1401 without building a settler. Penalizing good luck inherently feels bad.

I'd suggest using lunatic production rates as a guide for early years. So probably just changing the base to 7 instead of 4. Except that leaves you at too high in 1410.

Ideally, I'd actually make it more like 8 + 2/3 years (modify for size).

So 1409 would be 16? Still too high.

10+1/2 years. Makes 15 in 1410. 20 in 1420 is a bit low, but I'd much rather err on too low than too high.


Also, if huge is +50% I assume large is +25%? And similar for small?

[Seravy]

Also, if huge is +50% I assume large is +25%? And similar for small?

Exactly.

Okay, so let's set the formula for these values :

8 cities in 1402 and 24 cities in 1418.  16 cities in...16 years. How convenient.

(8+(turns-24)/12)*(2+land size)/4

This would mean
1401 7
1402 8
1403 9
1404 10
1405 11
1406 12
1407 13
1408 14
1409 15
1410 16
1411 17
1412 18
1413 19
1414 20
1415 21
1416 22
1417 23
1418 24
1419 25
1420 26
etc

This seems to allow way more cities than we planned in the early game though so it kinda defeats the point of having the system. Early game is where overextension really mattters...It roughly normalizes around 1412 and gets a very harsh in 1418 (where we assumed the player has 30-35 cities...)

Still, it's not a bad formula.

Let's check my previous game. I was pushing for territory like crazy, to the point that my garrisons in my home continent were minimal, but I still did my best to hold the conquered cities. So by my own definition I was slightly overextending, but not by the amount that should be penalized much (I started on a smaller continent, and I had to fight hard for new cities, it wasn't an easy conquest, but I did expand way more aggressively than what I normally do).

1406 I had 9 cities. Formula allows 12, seems about right.
1407 I had 11 cities. Formula allows 13, getting closer to the limit but still safe.
1408 I had 15 cities. That's 1 more than the limit, and by this time I most definitely was overextended. I couldn't have held more if I hadn't made a peace treaty with the other two wizards.
1409 I had 17 cities. 2 over the limit. 
1410 I had 21 cities. 

..Turns out I was actually overextending in this game. Other wizards only held 15 cities total at this point... and I couldn't make any progress past this for the next 10 or so years.
I'm amazed by the capabilities of the AI, being able to pose a threat when I held 60% of Arcanus.

Ok, I think I can agree to this new formula. Maybe reduce the numbers all by 1? 8 cities in 1402 sounds a bit unreasonable, even for Lunatic AI unless they found multiple easy neutrals. They can have 2 settlers in play at a time. Eh, might as well test it. Turn 24, with me taking no spots from the AI, highest city count I see out of 4 players is 7 cities. That includes a conquered neutral and the AI has no real way to accelerate the settler production so assuming there is land available, they will generally have 5-6 cities. With luck they might have 7.

..but there is rounding. turns/12. That always rounds down, so in December you still get 1 fewer cities allowed even though you used up almost the entire year. I think I'm fine with the 8 then.

So (6+turns/12)*(2+land size)/4 or greater to trigger the weaker warning.

And for the higher threshold, same but add (land size+2) cities? So +2 for Tiny, +4 for Fair, +6 for Huge? Or do we want more?

Or maybe we go back to the old system of having the penalty proportional to the amount you exceed the threshold?
Like penalty of (3+difficulty) REL plus an additional 20% for each 1 extra city you have?

[Seravy]

One more question, are we ok with the penalty applying to every wizard at the same time, or do we want to roll the 1/6 chance separately for each of them?
In one case you either anger everyone or no one at all, in the other case, you might get lucky with one wizard and unlucky on another. One is consistency between wizards within the same game, the other is consistency between games (you don't get games where rolling lucky gets you no penalties and rolling bad drags your relation to bottom with everyone)

The third option is to throw away random and have it apply on every 12th turn guaranteed.

[Nelphine]

Well, I can accept trying those city numbers. I think the late numbers will be too high but that can be seen later if it's true.

I'd personally like to do random by wizard, mostly so I don't get a message from everyone on the same turn - it always feels like they're banding together when that happens, which is odd in too many situations. When you banish someone it makes sense everyone gets mad together, but for overextension I think random per wizard will 'feel' better. Given how often it gets checked I think that it would even out pretty safely in either case, so I'm satisfied with basing my decision on perception of the player for this.

In general I definitely prefer to keep the random in.