[freeciv-ai] Re: README.AI

["Per I. Mathisen" <Per.Inge.Mathisen@xxxxxxxxxxx>]

On Mon, 29 Apr 2002, Raahul Kumar wrote:
> Greg, go ahead and edit other people's typing. I don't care.

Agreed.

> Greg, can you do a prettier version using MathML. This calculation looks butt
> ugly. Just post a link to a pretty version.

Or write it in floating point C.

> > also too hard on novice players and too easy for experienced players.
>
> Not sure this is really applicable anymore. Per's diplomacy patches makes it
> a lot easier for easy players and a lot harder for experienced players.

The team patch won't be in this release. Also, we're really talking 1-on-1
here.

> > didn't care about code readibility a lot. After he left the project,
> > various people have contributed their own mostly unfinished hacks
> > without really fixing the main issues in the AI code, resulting in
> > even more mess.
>
> Is this a dig at us? The code was in a far uglier state before the crew
> got here.

Do you really think I meant that? I'm hurt.

> >   This file aims to describe all such problems, in addition to various
> > not entirely self-describing constants and equations used in the code
> > commonly.
>
> All constants?

It says "various".

> > THE AI CREW
> > ===========
> >
> > If you wish to get in touch with the Crew about AI problems and have
> > save games and a method for reproducing the problem in hand, here are
> > the email addresses:

Shouldn't Ross be here, too?

> > These are ideal numbers, your mileage while travelling through the
> > code may vary considerably. Technology and diplomats, in particular,
> > seem to violate these standards.
>
> In parts of the code, if want exceeds 200 it is lowered below that value.

Ok.

> >            N*q^Y + N*q^{Y+1} + N*q^{Y+2} + ...
> >            = N*q^Y * (1 + q + q^2 + q^3 + ...)
> >            = N*q^Y / (1-q)
>
> Unreadable. Please have a prettier version linked to on the Freeciv website.

I suppose another math person could understand it...

> > * AI doesn't understand when to become DEMOCRACY. Actually it doesn't
> > evalute governments much at all.
>
> Not just democracy. Fundamentalism is a great gov for the AI.

Or whatever govt a modpack might have.

> > * Cities don't realize units are on their way to defend it.
>
> Not a bug!! We don't want cities to expect units that may never turn up.
> The AI code for pathfinding/ finding stuff to kill is insane.

Agreed. Unfortunately.

> > * AI doesn't understand that some wonders are obsolete, that some
> > wonders become obsolete, and doesn't upgrade units.
>
> All of this is easy to fix. Why are we lacking patches?

Good question.

> > * AI builds cities without regard to danger at that location.
>
> What do you mean by this?

I think Syela meant that a city shouldn't be built if it would get
danger > 0. I disagree. Building cities is good.

> > * City tile values are not cached; wastes CPU time.
>
> Get rid of it. There's just no benefit in caching tiles. Raimar already
> showed this.

Agreed.

> > * AI won't build cross-country roads outside of city radii.
>
> Mention that there is a patch that does this.

That patch was rejected, for reasons I think I disagree with, but until we
have an agreement on this point, it shouldn't be in a TODO list.

> > * Settlers won't treat about-to-be-built ferryboats as ferryboats.

I don't actually think this is something that should be changed for the
time being. See my comment on this in the settlerscleanup patch. I think
this point should go out.

> > * AI autoattack is never activated (probably a good thing too) (PR#1340)

This should be a comment in the document above, not a TODO item. The AI
doesn't need autoattack. Its own calculation functions are far superior.

> > * It may be correct to starve workers instead of allowing disorder to
> > continue.
>
> No. In order of best to worst actions:
>
> create specialist/increase luxury rate
> build temple/happiness improv
> buy/build happiness wonder
> buy/build engineer/settler
> finally allow city to starve

Yes, but comment is still correct.

> > * struct choice should have a priority indicator in it.  This will
> > reduce the number of "special" want values, thus reducing confusion.
>
> What the hell is this? If we are going to have priority indicators, what is 
> the
> use of want?????

I wonder the same thing.

Yours,
Per

Ask not what you can do for your country
Ask what your country did to you
 -- KMFDM, Dogma