Starting location

[Catwalk]

Isn't it a problem for an island AI that it's easier to isolate them?

[Nelphine]

Maybe but its not due to their lack of expansion. Its the lack of a third wizard in the area. Whether the island is good with multiple pop 25 dark elf cities, or bad with a badly placed pair of lairs blocking the island so the AI only gets two tundra cities, they seem to expand to other islands just as easily.

And I wouldn't want to try to prevent AI from ever being on an island alone.

[Catwalk]

I'll agree to disagree and move on to other matters. I'll keep an eye out for games where the AI is impaired by a bad starting location. Now that I think about it, the last Nature game I posted had Lo Pan (Blue) get off to an awful start both because of bad terrain and being blocked off by another wizard.

[Catwalk]

My current game (which I'm restarting, I want to play Sorcery instead) showcases weak AI expansion as a result of starting on an island:



Blue player starts on an island and has bad expansion. Green player is blocked by a sorcery node and cannot expand south at all, so is also effectively on an island. He has very bad expansion, and has chosen awful sites far away instead of much better sites on the same continent. Maybe this is a special case where he doesn't realize that he's blocked and gets continents mixed up? Purple has a nice starting location, it looks like his expansion was stunted by blue taking a neutral city north of him and using that to spam settlers in his backyard.

Conclusion based on the above example: AI fares poorly starting on an island.

[Seravy]

Green player is blocked by a sorcery node and cannot expand south at all, so is also effectively on an island.

I can't detect this sort of situation, so this one wouldn't be fixed by stopping island starts.
I assume the settlers used ships for transportation. The AI cannot use ships to "transport" within the same continent, so it had to go for another one. The rest of that continent is pretty much invisible for AI ground troops - another AI will eventually settle there instead, unless the node gets cleared.

Not sure what year this is, but purple and blue seem to be doing ok. Same amount of cities despite one being on an island. Sure, 5 cities are not 10 but neither of them are blocked, and they'll be able to fill the entire plane over time from the looks of it.
Yellow is doing crazy well, I wonder if he was accelerated by an early neutral or plays some fast race. Green is definitely hit hard by the location, but he is not on an island from the perspective of coding - disabling island starts would not help him in any way - if anything, it would make this happen slightly more often.

[Catwalk]

Keep in mind this is June 1403, 5 cities is quite low by this time. In my recent game all AIs have 4 cities by April 1402.

I figured the green player was probably screwed in a way that would be difficult to do something about. It shouldn't happen very often, so I guess we can ignore that one. Purple and blue are not losing, but they're by no means doing well.

[Seravy]

Keep in mind this is June 1403, 5 cities is quite low by this time. In my recent game all AIs have 4 cities by April 1402.

o_O
5 cities in 1403 is not bad at all. I still remember when 5 cities in 1408 was the best the AI could do.

[Nelphine]

I'd say.. 6 cities by 1403 is average for an AI that doesn't encounter another wizard. Some are faster, but rarely slower than 4.