From delayed choice experiment
The results of all four measurements are then compared. If Victor did not entangle his two photons, the photons received by Alice and Bob are uncorrelated with each other: the outcome of their measurements are consistent with random chance. (This is the "entanglement swapping" portion of the name.) If Victor entangled the photons, then Alice and Bob's photons have correlated polarizations—even though they were not part of the same system and never interacted.
Longer explanation can be found from: Does the passage of time effect a photons entanglement with another?
We have two pairs of qubits, Alice and Bob receive one qubit from one pair (1,4) while Victor gets two (2,3). If Victor does not entangle his qubits, Alice's and Bob's measurements are uncorrelated. That part is clear.
On the other hand if Victor does decide to entangle qubits 2 and 3 then they are projected into Bell state and may take on values 01 or 10 or 11 or 00. As a result qubits 1 and 4 become entangled and have the same previously mentioned set of four superpositions that they might be in. So Alice and Bob can measure respectively 0 and 1 or 1 and 0 or 1 and 1 or 0 and 0.
I don't understand how does this mean that Alice's and Bob's particles are entangled as Alice can't predict in what state Bob's particle is in after her measurement.
Is is possible for Victor to use some kind of local operation (some quantum gate) before or after entanglement so that Alice and Bob only measure 11 and 00 or 10 and 10 (only opposite states or only similar states) so that they would know what each other has after they have performed their measurements?