Dang, I had already written up a reply to this post, but lost it due to careless copy-pasting. Anyway --:
It’s I suppose possible that multiple possible worlds rank equally on the evil scale, and hence, differences between them are not instrumental in determining the best possible one – i.e. if for instance in one world, you had rice-a-roni, while in another, you had baked ziti, and yet both worlds rank equally, then the matter of your meal is immaterial with respect to determining which world is best. However, I’m not at all sure that different worlds can place exactly equally on the scale; in general, I think, everything will matter.
As for your second question, as I’ve stated in my summary, the metric with respect to which to optimise is god’s to choose – and after all, in common Christian belief, he’s what determines goodness (and by extension, evilness) anyway. Furthermore, it’s not at all a problem to optimise with respect to multiple parameters, however, in this case it is likely that optimisation in for instance two parameters will end up with a world placing worse on each of them than if either had been optimised for separately – for instance, god might desire a world in which there is as much free choice, but as little evil as possible, and choose the world that places best on these two parameters combined; however, he will generally wind up with a world that is less free than had it been optimised solely for freedom, and more evil than if the optimisation had been towards minimum evilness exclusively.
As the original context of the discussion was with respect to the problem of evil, evil as understood in whatever framing of the problem you prefer was what I used as optimisation parameter.