Re: [Rails-devel] Automatic route calculation

[brett l. <bre...@gm...>]

On Sun, Mar 21, 2010 at 10:45 AM, John A. Tamplin <ja...@ja...> wrote:
> On Sun, Mar 21, 2010 at 1:21 PM, Stefan Frey <ste...@we...> wrote:
>>
>> I have only a very basic knowledge about graph theory, but from my
>> background
>> as econometrician I wonder, if there is a need to search for the true
>> maximum. Is there no equivalent to solve such a problem similar to the
>> search
>> for the global maximum of numerical functions, where one can employ
>> simulated
>> annealing or genetic algorithms to get close to the optimum?
>
> Close to the optimum run is going to pretty annoying for a player who needed
> the optimal run to buy their permanent train.
>>
>> Usually that is as good enough for macro-economist to optimize social
>> welfare,
>> should that not be sufficient for a 18xx player ;-)
>> And if one player thinks he can do better, still let him do so?
>
> I'm just not sure there is much point including it if it can't be correct.
>  If it is accurate most of the time, then people will depend on it and not
> notice when it is wrong, and if it is rarely right people will just turn it
> off and ignore it.


There are cases where SimTex's 1830 route calculator is wrong and
doesn't find the optimal route. It's not perfect, but it's very
usable.

I think the right approach is to have route calculation as an option,
and iteratively improve it after we've got it functional.

It makes sense to also take a modular approach to route calculation,
and allow ourselves the space to develop multiple calculation
algorithms that can be used.


> --
> John A. Tamplin


---Brett.