TEPCO still doesn't know what caused the power failure and how, therefore does not have measures in place to prevent any future power failure. Pressure from the national government was clearly on the operator to just start cooling and not worry about minor details like what caused the failure.
Ad hoc has been the name of the game for the past two years. Why change now?
Asahi Shinbun (3/20/2013; part):
The power outage this time was the most severe since the start of the accident. However, the cause of the outage hasn't been identified. According to TEPCO, there was no visible sign of abnormality in the temporary switchboard that is connected to Reactors 3 and 4, where a trouble that led to the power outage was thought to have started. The company had initially planned to identify the cause, install measures to prevent future problems, and then turn the power back on. However, restoring the power was given the priority, and the cooling systems were connected to a different switchboard, circumventing the temporary switchboard.
The particular temporary switchboard is "3/4 M/C (A)" (M/C stands for "metal clad"). The location (or connection) of the switchboard in the overall electrical system looks different in the two different diagrams below. Either way, I don't quite understand why Reactor 1 should be affected by the failure of this switchboard, as the Reactor 1 cooling system does not appear to be connected to this switchboard.
As TEPCO's spokesman Ono explained it (via Ryuichi Kino),
[The trouble at "3/4/ M/C (A)"] had the ripple effect on Plant-wide common M/C (2A)(2B)
3/4 M/C (A), in the diagram from March 2012, from then-Nuclear and Industrial Safety Agency (NISA, also from Ryuichi Kino: