You'll see the same problems with huge maps that you currently see with tiny, but only they will be milked games and obviously will take a longer time.
Each of those games I recently submitted took at most an hour to complete. Some of them like the extremely easy low difficulty games took maybe 15 minutes. On chieftain, only 1 civ actually had defenders. Compare that to a larger map when it can take a few minutes waiting for the AI to move, and then easily 15+ minute for you to deal with your civ (like gotm2). I quickly grow bored of this, but as you'll see once the gotm2 scores come up, plenty of people *will* do this for a high score.
So if you do what you suggest you will get some early conquests initially and then a load of milked games. I'm not sure why you are leaning towards the larger maps, why allow huge but not small? I don't play anything greater than standard. And standard only a few times. I have a fast machine, but the micro-management involved is just way too tedious for me.
Anyway, Firaxis' scoring is just broken, its only based on city area and population (which are actually closely linked) so in effect you are just getting x2 score for how many cities you have. Ignoring that very general problem, there is no scale at all for map size. A larger map will allow you to build more cities and obviously have more population and land area which will directly affect your score.
If you want to make the standard size the standard size I suggest you deviate from standard the same in both directions and have a self made scale factor for each. I would just make it simple, figure out a scale factor based only on how many tiles are present in small and large, the goal being if you have an exact civ in either of those maps your score would be the same. In other words a tile in standard may be worth 1 point, a tile in small 1.25 and in large 0.75, or something along those lines. Whats the actual tile size of the maps?
Then once that problem is solved you'll have to deal with random maps, not all maps are created equal. My idea is to have people include the 4000bc save along with the highscore so others can remove the "good map" bonus and compete directly with a current highscore, the intention being to remove any luck factors and create a skill only ranking. If I get a highscore on a map, and then you play it and get a better score, you are obviously the better player and more deserving of the highscore. However, you'll still have to rely on players being honest and not replaying a map forever to improve their "that map only skills".
Thats a fine tradeoff in my mind since there is only so much you can do with any given map, but across all the possible random maps there will be a large variance in scores. That also doesn't change much with map size since the score is an average, if you start with a bad map and have bad land initially you're score will always be hurt by that.
Just take a look at my deity game, that was easily the easiest deity game I've played and that was reflected in the score since I got a quick conquest. Its all because of some wheat flood plains. If you play a deity game and don't have similar resources at the start you simply won't be able to attain a comparable score (in conquest that is).
This is a very good idea at the same time! 
