Why the Many Worlds Interpretation of Quantum Mechanics isn't Deterministic

It's often said that the Many Worlds view of quantum mechanics simplifies things because it's purely deterministic - as all outcomes of quantum possibility do actually occur.  This is a mistake.  There are many problems with this view of the world, and one of them is that it doesn't in any way help with deterministic predictions.  

Consider the good old Schrodinger's Cat situation.  A cat is enclosed in a box with some poison gas and a radioactive source, which, if it emits a particle, will result in the release of the poison gas.  We wait long enough that the radioactive source has a 50% chance of emitting a particle.  Is the cat alive or dead?  The Many Worlds view says that the cat is both alive and dead, but in different realities.  This is in contradiction to the conventional 'Copenhagen Interpretation' which states that, at some point, the possibilities collapse so that the cat becomes either dead or alive in one reality.

Now suppose there is an observer O who tries to guess if the cat is dead (D) or alive (A).  In the Many Worlds interpretation the Observer O splits into two observers O1 and O2, each of which observes one of the possibilities D or A.

So, suppose the observer guesses that the cat is alive.  The possible results of the experiments are:

1. O1+D and O2+A => observer 1 sees dead cat and observer 2 sees alive cat
2. O1+A and O2+D => observer 1 sees alive cat and observer 2 sees dead cat

Which ever of these outcomes is what happens, there will always be one copy of the observer who has guessed wrong, and one who has guessed right.

Like all mainstream quantum mechanical interpretations, the Many Worlds interpretation leaves observers with points in their history which were unpredictably random.  No observer either has a fully deterministic past or a fully predictable future.