Some people claim that they have seen the image of a pipe on this blog. Obviously, there are many different ways how such an image could have been observed in the many worlds we live in, but in this world there is only one way how it could not have appeared.
So what does this tell us about probabilities?
PS: If you want to think even more about probabilities and many worlds, I recommend this paper and in particular the sections about some toy many-worlds models and
about 'the problem of inappropriate self-importance'.
added later: In my previous example a quantum experiment has two possible macroscopic outcomes, e (espresso) and n (no espresso), and the probability for each is 50%, although e can be realized in N different (macroscopic) ways, while n can be realized in only one way.
Now, one could argue that in both cases the observer does not immediately experience a conflict with the 50% Born probability (the observer in one world e or n does not experience all the other worlds). Therefore we need to consider iterating this experiment.
If the experiment is repeated R times, then observers who see the outcome eeeeeeeee...eeeeee will indeed conclude that something is very wrong. Unfortunately, they are the overwhelming majority with N^R worlds and I would really like to understand how this can be reconciled with e.g. the Deutsch-Wallace interpretation of probability.