The learning curve happens over time.
The ships up to the NX-class have single bodies with the warp cores inside. It might be either experience with the Warp Five engine causes them to move the newer cores to a secondary hull, or the next warp core design after it proves more likely to have issues than the Archer Warp Five core, and thus the extra safety of separation is installed during the refit of the NX-class and many later classes of starships.
The fluff from Drexler suggests that the new core was not as tested as the Archer Warp Five core and thus to be save and have the means to get the ship home, they added the secondary hull with the new core inside, while leaving the original warp core in place in the primary hull. Thus, if the new core failed, they could drop the secondary hull and later start up the old core to get at least some form of warp travel back to a Starfleet base. They would probably not be able to go at Warp 4 or 5 with the secondary hull gone since there extra pod thing on the back would have been ditched with the secondary hull, so they'd be limited to maybe warp 3 without it. It was supposedly there to help balance the warp field at higher speed and to compensate for some sort of time related effect (Drexler makes reference to a line from "The Cage" about breaking the "time barrier" with the newer ships like Pike's Enterprise).