Suppose that the many-worlds interpretation of quantum mechanics were true.

Would that rule out trial and error as something that actually happens in reality?

Let's take biology as a case study. According to the Darwinian theory of evolution, random gene mutations occur every so often. The mutations that are best suited for a given environment survive (i.e. natural selection), while the rest end up in the dustbin of biological history.

But through the lens of many-worlds, mutations aren't "random". Every time a favourable mutation occurs, there's another universe in which the mutation does not happen. Taking this logic back to the beginning of life, there should be lots of universes in which the Earth is life-less even to the present day.

Does this make sense? Or is my argument based on a misconception?

  • Doesn't this mean that trial and error does actually happen in reality? You have multiple trials run in parallel and some of them succeed at creating life while others don't. It is like multiple labs trying different mutations, or none at all, and observing what happens in the long run.
    – Conifold
    Aug 30 at 22:53
  • If you define reality as the entire multiverse (again, taking it for granted that the theory is correct), one shouldn't prefer one universe over another. A lab is a purposeful knowledge creation machinery. A universe isn't. Aug 31 at 7:37
  • Where I am trying to get to is this: there are no trials, just a never-ending branching process. Thinking about probability in this way is qualitatively different. Aug 31 at 7:45
  • Reality doesn't have to prefer life to non-life, life emerges by trial and error regardless. Of course, if your definition of "trial and error" involves somebody having a purpose then you'll have to add a deity or a simulator who is running the multiverse to reach it, whatever it is. But even labs' purposes are fungible, they may start looking for one thing and then branch out and explore something else. I do not see how somebody having a purpose has any effect on how probability works though, it's just that their search doesn't have to be exhaustive.
    – Conifold
    Aug 31 at 7:57
  • 1
    @Conifold The lock analogy is not a great one because it only includes variation, not selection. In evolution, mutation creates branching variation, and then the bad branches are "pruned." The question is, how does this translate into QM, in which branches are never pruned? And the answer has to do with the Born rule; though never actually pruned, some branches become unlikely.
    – causative
    Sep 1 at 9:00

Let me see if I understand your point.

You observe that, in many-worlds, there is a branch for every outcome, and no branch has the honor of being the "canonical" one. Let us consider a fictional animal called a "grobling" that is about to be conceived. Let us say that an extra arm would be very beneficial to a grobling, if it had one.

  • There is a branch in which the grobling mutates and gets an extra arm, and prospers in its life.
  • There is a branch in which the grobling is born normally, and prospers.
  • There is a branch in which the grobling gets the extra arm, and dies to a predator without reproducing.
  • There is a branch in which the grobling is born normally, and dies to a predator without reproducing.

Now how do we "prefer" one of those four branches? In many-worlds, no branch is the canonical one, right? So they should all happen at once, and it's difficult to see how there is selection pressure to prefer certain mutations.

That is how I understand your point.

My answer to it is this: in many-worlds, not all possible worlds are equally likely. No single branch is canonical, but some are more likely than others. In quantum physics we assign a probability amplitude to each possible outcome, which is a complex number. The squared magnitude of the probability amplitude is interpreted as a probability.

We would expect that the probability amplitude of branch #1, in which the grobling gets the extra arm and lives, is greater than the probability amplitude of branch #3, in which the grobling gets the extra arm and dies horribly - because the arm is useful to it. So evolution still happens, in the sense that branches in which the grobling is more suited to its environment are assigned greater probability.

There is some controversy over why we should interpret the squared magnitude of the probability amplitude as a probability, especially in the context of many-worlds. See some comments about that here. But the glib answer is, we do make that interpretation, and thus we privilege certain branches over others even in many-worlds.


There is a kind of selection process, similar to natural selection in a way, where some macroscopic outcomes are more likely than others to "survive" (occur). These branches are more common than their counterparts. An example would be firing a gun at a person. There might be 999/1000 branches where the person dies and 1 where they live because the gun misfires.

Since every physically allowable outcomes occurs, there is at least one branch where life on Earth never evolved. Are there more branches where life evolved than not? Hard to say because we don't know how the original spontaneous generation occurred, so teasing out the relative branch weights is not really possible at this time. But we do expect on average to be in the most common branches, so this gives some hint there are more worlds with life on Earth than not.

I think I agree with Conifold that this could be seen as trial-and-error. Say there is a new mutation. If it helps the organism, there will be more branches where they are alive. It it hurts the organism, there will be more branches where they are not alive.

We can't access these other branches, but we expect on average to find ourselves in the branches that are most common (basic probability). So if we continue to see said species, we could say that mutation helped. This would reinforce our idea that the mutation was adaptive. We keep seeing the species around, and since we expect to on average be in the most common branches, we say this trial helped. This trial as in this specific mutation. There are more branches where this species is alive than dead - simple.

So even within a single world we are learning about the trial and error process of the global set.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.