I really don't see a huge problem with the magical generation of a unit of population. It goes without saying that we'll be able to balance this with the cost of the colony ship, especially when you consider the fact that there will be (IIRC) no manual movement of colonists. Furthermore, the need (or want, for that matter) for migration totally eludes me. In MoO3 it was just another one of those things that made the game more confusing and less fun, and if the player has no control over it, why have it in there at all? IMO, we should just KISS the whole migration issue (which isn't really in question here, but people seem to be assuming it will exist for the purposes of their arguments) and have no migration whatsoever.
However, I wouldn't mind an implementation like Geoff suggested where there is a secondary resource requirement that needs to be filled. IMO, the objections to that suggestion are invalid because a) the build queue is a fairly important part of the game. Having more than one resource to keep track of per build item is not micromanagement by any means, and b) you won't have to wait longer for the colony ship to be built if you build it at a place where you have people. In a sense, its a good limiting factor for where and how often you can build colony ships, which is probably good for balancing in certain cases, for example if you have a highly productive race with low reproduction, does it make sense that they'll be expanding like mad?
Another thing I like about the secondary resource idea is that it would allow easy implementation of this
idea, which I am quite fond of. Overall, I'm in favour of the secondary resource requirement suggested by Geoff, whereby the planet on which the colony ship is being built must fill a population requirement. Magical population generation is my second choice. Anything that relies on migration is un-KISS and unappealing to me.