[Freeciv-Dev] Re: (PR#11005) Question about Freeciv 2.0.0 (maybe)

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=11005 >

Ed Earl Ross wrote:

>><URL: http://rt.freeciv.org/Ticket/Display.html?id=11005 >
>>
>>Ed Earl Ross wrote:
>>
>>
>>>I'd call this behavior an AI kludge. It is important to human players 
>>>who play against AI, because it gives AI an advantage, and thereby 
>>>changes the dynamics of humans playing AI between 1.14.x and 2.0.x. If 
>>>it is too much of an advantage, it may make AI players too powerful.
>>
>>So to balance AI players being able to cheat we should allow human 
>>players to cheat as well?  That hardly seems logical.
> 
> Clearly, if AI super-human game "knowledge" makes the AI player better 
> than any human, something must be done to let humans play and sometimes 
> win against AI. I didn't specify a solution; though, the better solution 
> is to improve the AI code so it can follow the same rules as humans. 

First of all, the AI cheating doesn't make it super-human.  Compared to 
a good player the AI is terrible.

Secondly, letting the human cheat is not the answer.  Instead the AIs 
cheating should be restricted.  (Side note: there may be different 
degrees of cheating possible.  For instance rather than being shown the 
whole map you may only be shown a few tiles beyond where you could 
normally see.)  The different degrees of AI difficulty are supposed to 
have different handicaps that distinguish them, but this all needs some 
work.  Even fixing the AI to not cheat wouldn't be too easy, I think.

Thirdly, the AI programmers are always trying to improve the AI.  Any 
help you can give them would be appreciated.  One easy thing you could 
do would be to download patches and test or compare them.  Or you could 
keep careful track of the AIs playing and when they do something 
obviously stupid, make a savegame and report it.

jason