It does not apply if you're an altruist trying to help people as much as they can be helped.

First I would like to note that I don't disagree with you in practice, though I remain sorely tempted to donate to Village Reach.

If your goal is not to maximize altruism, but rather to ensure a certain minimum level of altruism given massive uncertainty about the effectiveness of charities, I could see it being reasonable to split donations.

Let's imagine there are two competing existential risk reduction charities. We could call them, say, the Singularity Foundation and the Future of Humanity Cooperative. Neither of them are rated by Give Well because there are no real metrics for evaluating them. If your main concern is not to maximize altruism but to minimize the chance that you give all of your money to something practically useless, why not split? I think it's possible that timtyler means something like this by diversification, though of course I don't think that risk aversion trumps altruism.