Is there any hope of getting the Population Transport minister a new brain?

So I took a look at the problem last night. Programming is my Day Job, and I'm not sure how much of my fun time (which is on a budget now that I'm married) I want to dedicate to this but I think I can do a little bit better, possibly at the cost of performance on lower-end machines.

The current algorithm is not specialized for population in any way, nor for really trying for any sort of overall improvement in the planet population spread. It just says "if I'm not 75% full, look for a planet I can fill up from; If I'm mostly full, look for a planet I can dump on"

My thoughts on an algorithm that's not as efficient as a human on large transports but will work much better than the current one are as follows. (The immediate actions at the beginning handle cases that arise when the ship goes to a planet that can be improved and needs to juggle cargo; there is no per-ship state memory to handle these at the point where the planet in question was chosen as a destination.)

* If I have cargo of mixed atmospheres:
** If I am over a domed planet where some of my cargo are natives, immediately drop them and pick up non-natives, repeat until out of natives in me or non-natives on the planet.

** If I am over a non-domed planet where some of my cargo are natives, dump the natives.

** If I'm still mixed, find the nearest planet where the smallest part of my cargo are native where they will all fit, and dump them there.

** If that doesn't work, try the next smallest part of my cargo.

** If there just isn't a way to get unmixed, find a domed planet for which my empire has no natives, and dump my smallest group there.

* If I'm empty:
** Look for a planet with a mix of native and non-native population, and go take non-natives from it.

** If none of those, count the number of domed colonies by atmosphere type, and starting with the atmosphere most represented, look for a colony with lots of that kind of natives to fill up from.

** If none of these, look for a domed colony that is near or at max pop, weighting distance less significantly than normal.

** If none of those, look for any colony to fill up from, with the original weighting of distance vs overpopulation.

* If I'm not empty but less than 75% full:

** Look for a planet with lots of population of the same atmosphere as I already have, and fill up from it, with the original weighting for distance and population among candidates.

** If none, look for the nearest domed planet where my current cargo is native, go there, pick up non-natives to fill me, dump natives. (This should trigger the mixed stuff when I get there, resulting in maximal progress toward undoming.)

** If none, fall through and pretend I'm full.

* If I'm full:

** Look for the nearest domed planet where my current cargo is native and fits, and dump them there and pick up non-natives.

** If that fails, look for the nearest domed planet where my current cargo is native and there is at least some room, go there, dump natives, pick up non-natives. (This should trigger the mixed-cargo stuff.)

** If that fails, look for an undomed planet where my population is native that needs population and dump there, with the usual weighting for distance vs. need.

** If that fails, look for a domed planet that needs population and dump there, using the usual distance vs need weighting.

** If that fails, clear my "minister controlled" flag. I need human intervention.

Hmm, given the limitations of the scripting language it's hard to have more than one thing be an absolute priority, as I'm making undomed planets be. But if you want to define a function to replace the current one, for rating a given donor or recipient planet numerically, you could. For that matter, such a formula for the existing one would be a much more trivial change than the large amount of work I'm talking about.

I am actually hoping that somebody much more familiar with the scripting language than I am, maybe one of the folks already working on the AI, will take this algorithm and run with it. But more likely I'll try to hack it up myself.