I am not sure this is fixed. In 2022-03-15 Nightly REF does not surrender on declaration. However they do as soon as they land on the shore next to your city. See attached. Just end turn.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
With git.c9eedcf I can load game, hit return, watch the REF try but fail to take New Haarlem, and it does not surrender, yet. The problem was that you have an overwhelming land army (roughly twice the power of the REF) and it was easily discouraged. I have gone back to the behaviour where ithe REF hangs on until either the army or navy is nearly all gone. Not ideal, we have had requests that it give up when clearly overmatched, but then again, it would be hard for the troops outside New Haarlem to know your full strength.
Setting back to Pending.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
What would happen in real life? I suppose with 2:1 military superiority the only realistic way would be for REF to attempt capture of undefended patch of land. Unless AI is reworked to operate better on tactical level this is not going to happen.
The best REF strategy would be to find cluster of colonies with lower defences producing Tools and Muskets try to capture it and grow from there. This not necessarily would mean capturing port first.
With abundance of sea power they could also start by establishing superiority at sea and only then landing.
I am guessing actually coding this is hard :(
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I think the current REF behaviour is "ok". We could certainly do better but there are other parts of the AI that need attention first. ATM, the AI will try to find a good target and attack it. IRL they would not be able to tell that they are badly outgunned, as they can only really see the forces at the colony they are attacking, not that there is a huge force distributed all over the map.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
We need to keep in mind that realism is just an inspiration and not a goal. The goal is having fun. There's no fun in the REF surrendering when the player is looking forward to a big fight.
The same principle is also true for the rest of the AI code. Fun should be our objective -- and not a good AI player and certainly not realism.
I could easily create an AI that would defeat the player almost every time -- with a strategy of selling the initial goods to the natives (muskets and tools), buying artillery and just overwhelm the human player at the start of the game. There's little fun in that.
The goal when I first created the AI was to make high quality colonies for the human player to capture -- with experts, tile improvements and sensible choices for buildings. In addition to token resistance. We need choices like transporting a bit too much tools, so that privateering is more fun for the human player. I can see there have been some regressions in this area that should be fixed.
Keep in mind we have programmed in intentionally silly behavior in order to make the game more fun. For example, in the ScoutingMission we prioritize speaking to the chief over taking lost city rumors (even when having de Soto). This avoids having the AI steal all the fun lost city rumors before the human player has a chance to explore them.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yep, the REF surrender condition was borked, again. Fixed in git.6ed1064.
Note: Duplicate bug reported at https://github.com/FreeCol/freecol/issues/98
I am not sure this is fixed. In 2022-03-15 Nightly REF does not surrender on declaration. However they do as soon as they land on the shore next to your city. See attached. Just end turn.
Resetting to open, clearly needs a revisit.
With git.c9eedcf I can load game, hit return, watch the REF try but fail to take New Haarlem, and it does not surrender, yet. The problem was that you have an overwhelming land army (roughly twice the power of the REF) and it was easily discouraged. I have gone back to the behaviour where ithe REF hangs on until either the army or navy is nearly all gone. Not ideal, we have had requests that it give up when clearly overmatched, but then again, it would be hard for the troops outside New Haarlem to know your full strength.
Setting back to Pending.
What would happen in real life? I suppose with 2:1 military superiority the only realistic way would be for REF to attempt capture of undefended patch of land. Unless AI is reworked to operate better on tactical level this is not going to happen.
The best REF strategy would be to find cluster of colonies with lower defences producing Tools and Muskets try to capture it and grow from there. This not necessarily would mean capturing port first.
With abundance of sea power they could also start by establishing superiority at sea and only then landing.
I am guessing actually coding this is hard :(
I think the current REF behaviour is "ok". We could certainly do better but there are other parts of the AI that need attention first. ATM, the AI will try to find a good target and attack it. IRL they would not be able to tell that they are badly outgunned, as they can only really see the forces at the colony they are attacking, not that there is a huge force distributed all over the map.
We need to keep in mind that realism is just an inspiration and not a goal. The goal is having fun. There's no fun in the REF surrendering when the player is looking forward to a big fight.
The same principle is also true for the rest of the AI code. Fun should be our objective -- and not a good AI player and certainly not realism.
I could easily create an AI that would defeat the player almost every time -- with a strategy of selling the initial goods to the natives (muskets and tools), buying artillery and just overwhelm the human player at the start of the game. There's little fun in that.
The goal when I first created the AI was to make high quality colonies for the human player to capture -- with experts, tile improvements and sensible choices for buildings. In addition to token resistance. We need choices like transporting a bit too much tools, so that privateering is more fun for the human player. I can see there have been some regressions in this area that should be fixed.
Keep in mind we have programmed in intentionally silly behavior in order to make the game more fun. For example, in the ScoutingMission we prioritize speaking to the chief over taking lost city rumors (even when having de Soto). This avoids having the AI steal all the fun lost city rumors before the human player has a chance to explore them.