If you take a Chomskyan view of syntax, it's not possible to get a VSO order without some sort of movement. The arguments to a verb are always Merged in first, and then the subject attaches somewhere higher up; you can't get the subject between the verb and its arguments without something moving.
(In other words, like with many parts of the traditional Chomskyan framework, we assume that all languages work fundamentally like English, and then apply some extra rules on top to explain why they look different. This is a common criticism of the program.)
So according to a Minimalist, the underlying structure would look something like this. (Don't worry about the "little v", it's not relevant to this question.)
Then a movement rule takes our verb and moves it somewhere to the left, perhaps to T.
And now we have a VSO order, "likes Alice Bob".
So according to this Minimalist, why is it likely that we'd also see SVO structure in this language? Well, because to get SVO order, all we'd have to do is put an exception in this movement rule. For example, perhaps this movement rule doesn't happen in a particular type of clause. Then in that type of clause, we'd see SVO order instead.
Or in other words—we already need one movement rule to get VSO order, and we don't need to add any more rules to get an alternation with SVO order, we just need to add a loophole to the one we already have. This makes it very easy for this situation to evolve.
For a specific example, older forms of English had exactly this sort of structure: SVO, with V moving to T in interrogative clauses. Thus, "you do like him" alternates with "do you like him?" (In modern English this still happens, but only certain verbs are allowed to move: *"like you him?") A single rule that applies only in certain contexts thus gets us an SVO~VSO alternation.