[freeciv-ai] Re: long-term ai goals

["Ross W. Wetmore" <rwetmore@xxxxxxxxxxxx>]

At 12:55 PM 02/05/23 +0200, Raimar Falke wrote:
>On Thu, May 23, 2002 at 12:43:12PM +0200, Christian Knoke wrote:
>> On Thu, May 23, 2002 at 11:34:00AM +0200, Per I Mathisen wrote:
>> > I suggest we put the following into README.AI:
>> > 
>> > =====
>> > The long-term goals for Freeciv AI development are
>> >  -> to create a challenging and fun AI for players to defeat
>> >  -> to create modular AI code that can easily be assembled into new AI
>> >     clients
>> >  -> to have multiple different AI clients compete against each other
>> > 
>> > In order to get to this point, the current AI code will be moved from the
>> > server and to the client. This requires that the AI code is separated
>> > completely from the server, and that clients get the possibility of an
>>                                                ^^^^^^^^^^^^^^^^^^^^^^^^^
>> > omniscience cheat.
>>   ^^^^^^^^^^^^^^^^^
>> 
>> That sounds strange to me. It's the way from an 'open' design
>> to a 'closed' one. Thought we wanted "free" client side AI.
>
>Yes from the AI development point of view. However we also have to
>provide an AI which is hard to beat. So it is fun to play against this
>AI. One way to make the AI harder is to cheat.

Only humans cheat - and they really do start with a grossly unfair
advantage. 

The client should have the ability to make requests of the server for
data that is computed or filtered by the server and returned. This 
is what FoW and various unit movement filters do. AI and handicapped
humans should get information which is perhaps less filtered or 
randomly filtered representing sources of information not available
to the better players. The score should be reduced accordingly as
a penalty if one wishes to provide human targetted disincentives.

If the server returns data which is not governed by the strict rules
it may introduce programmed/random errors into this data. There are
viable rationales for this effect which can be introduced into the 
rulesets and/or controlled by console settings - "Bardic travellers
arrive from <blort> giving you information about its countryside",
or just a bardic play level parameter that modifies the known and
seen bits in a pseudo random way and/or to point to pseudo-random tile
locations/contents. It can be easily implemented and rationalized.

When playing solitary against an AI this is much preferable to a
poor AI that lacks both insight and external sources of information.
Adjustable levels make an excellent learning curve. This also fixes
some of the unfair restrictions on the AI of not being able to move 
places it has not yet moved too. A human can figure out how to do 
this easily so such code restrictions don't really hamper them. 

When playing in competitive matches, there will likely be no AI, and 
the option will be turned off, so there is no reason to penalize one
audience or the other - satisfy them both.

I make a distinction between providing information, and unfair battle
odds or special rules. Even these are ok, if they are done as bonuses
for given factions or personality choices, but then they need to be
countered with appropriate penalties for balanced play.

Cheers,
RossW
=====

>       Raimar
>-- 
> email: rf13@xxxxxxxxxxxxxxxxx