Crusader Kings II is a game with a long and proud tradition of “bugs that should be hung in the Louvre”, but I think this new one I’ve learned about today is my favourite.
And this is a game where a horse could become Empress of Rome, so that’s something.
As Paradox Interactive’s Henrik Hansson explains in this Twitter thread, in which devs were asked “What are your favourite bugs and weird edge cases in video games???”, he cites the example of a version of Crusader Kings II in which the Greek culture got a little too mean.
In CK2, before 2.4, if the Greek culture grew too large, the game would grind to a halt since every AI character of Greek culture went through every other character in the realm and considered castrating or blinding them. They repeated this process every day. Aspirational!
As he says, that particular bug was patched out long ago (2015, to be exact), but this is the first I’m hearing about it, and I love it. To provide some more background, here’s Hansson again, writing about the bug when it was first discovered on Paradox’s forums:
But what really hits performance are large empires because it increases the amounts of realms (every ruler technically holds his own sub-realm) meaning more evaluations per character has to be made. For instance I discovered late game that huge greek/byzantine empires were slowing down the game because each greek person was evaluating against each other person in the realm “can I castrate?”, and this took up like a huge chunk of the performance of the AI. I think 70% of the AI demands were about castrating or blinding someone when I loaded late-game byzantine saves. Luckily this is fixed in the coming 2.4 patch. A fun little anecdote.
Crusader Kings II is a game where the AI must constantly be calculating almost every aspect of medieval political and military life, from the smallest sieges to the largest plots to usurp a ruler. To have 70% of a huge empire’s AI demands be only about weighing up whether to blind someone or cut off their balls is very, very funny to me.
Leave a Reply