The many world's interpretation is based on the assumption that both the observed system and the observer is modeled by a wavefunction. In that case observation is entanglement, which seems weird to us, but which makes perfect logical sense.

Other interpretations of quantum mechanics start with the assumption that the world is fundamentally like our naive experience leads us to think of it as. Therefore when we observe something, we couldn't really be thrown into a superposition of states, each of which observed a different thing and which cannot meaningfully interact due to quantum mechanical principles. If we can't be thrown into such a superposition, then there must be some sort of underlying reality, or quantum collapse, or other weirdness that is not explained by QM.

All of the weirdness of interpretations other than many worlds can be understood as the conflict between how they would like to understand the world (there is a reality that happened), and how quantum mechanics described things.

All of the weirdness of the many worlds interpretation can be understood as, "We can't perceive the process of quantum superposition, so it seems really, really weird to us."

Lets say you have some experiment which gives you a 1 or a 0, and you have two entangled particles which will give the same result depsite the result being random and the measurements being taken outside each other's lightcones. In Collapse QM there was some instantanious communication making that possible.

In Many Worlds you'd find that for each measurement the wavefunction split into two branches, one corresponding to 1 and one to 0, in each location. From the sites of both measurements the decoherence spreads outwards as particle interactions happen, and when they finally touch the two 1 segments and the two 0 segments merge, and there is one area split along 1/0 in the wave function instead of two.

That's all horribly simplified, but should get the idea across.