[freeciv-ai] Re: (PR#8992) Patch: Building ferries

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

<URL: http://rt.freeciv.org/Ticket/Display.html?id=8992 >

On Thu, 4 Nov 2004, Gregory Berkolaiko wrote:
> My experience in AI tells me that simple WAGs usually work at least as
> good as long and complicated models. Civ is a too complicated game to be
> analyzed exactly. As an example I can mention old settlers code which was
> simulating city's growth and new code which doesn't do any simulations but
> achieves, on average, better results.

I do not think your example is very good. The reason why the old settler
code was bad, and the new one better, is mostly that the new one uses the
same codepaths for evaluating the city as is used for calculating the city
benefits. The old settler code has its own calculations which was full of
silly assumptions and dubious guesses. The downside is that using the same
codepaths for evaluating and for final calculations is that it is slower.

However, I think the old settlers code did something right when it
modelled the growth of the city, and I did something wrong made I
calculated the benefit of a city from a size 1 city. The problem is of
course that modelling is also slow, so using the new settlers approach
with a growh model would be extremely slow. But I still think it should be
done, only that it needs to be used fewer times and the resulting data
used (cached) more intelligently.

Anyway, you are problably right that the ferry patch tries to model too
much, and that a simpler model would be more effective.

  - Per