Terrible ideas as that would all but eliminate a weak opponent from ever getting the spell off. Nope, once it's started the only way to stop it is to kill the caster. I do agree with a certain amount of turns to do this, but, nothing else.
I more or less agree, but I think there needs to be significant heads up beforehand. The problem with magic and tech victories in most games is that by the time you find out that someone is on their way towards a tech/magic victory, it's almost always too late to do anything about it. This is annoying when you play a game and then after 300 turns see "Player A is researching Tech Victory," and then 20 turns later by the time your forces finally reach Player A, the game ends, you lose. However, it's just as bad the other way around - I have never lost a game after starting to research the tech victory research (the AI has never had time to kill me once it gets that warning before I finished the research).
So my point is, there needs to be enough warning so that something can actually be done about it, but the spell has to be done in such a way that weaker players still have a shot at pulling it off. One thing I can think of is to have the spell take a very long time to finish casting, but maybe give you incremental bonuses as it progresses. So even if it's going to take a relatively weak player 40 turns (arbitrary, hypothetical large number) to research, their power would grow every turn - so the longer they hold out, the stronger they become. Balanced right, this could give enough warning to do something about it without totally screwing over weaker players pursuing this victory condition.
I am also not opposed to giving methods of impeding the casting of the Spell of Making. I'd rather not be able to just stop it outright, but to be able to slow the casting down - by how much would be determined by the relative magical prowess of the opposing channelers. And to prevent the whole world ganging up on that one player and bringing the casting progress to a complete halt, there could be a limit (more channelers opposing could have quickly diminishing returns, for example).