Although the idea of a multiverse has been around for a while, no one has ever really come close to making the sort of calculation outlined in the previous section.2 Instead, those wishing to make predictions in a multiverse context have made strong assumptions about which parameters ai actually vary
2 The most ambitious attempt is probably the recent one by Tegmark [3,5].
across the ensemble, about the choice of O-object, and about the quantities PM (ai) and NO,M (ai) that go into predicting the measurement probabilities. Some of these short-cuts aim simply to make a calculation tractable; others are efforts to avoid anthropic considerations or, alternatively, to use anthropic considerations to avoid other difficulties.
I would not have listed any ingredients that I thought could be omitted from a sound calculation, so all of these short-cuts are necessarily incomplete (some, in my opinion, disastrously so). But by listing and discussing them, I hope to give a flavour of what sort of anthropic arguments have been made in the literature and where they may potentially go astray.
Was this article helpful?