Re: Here's more sensible analysis...
Well having reviewed as much as I can of the data, and trying to ignore as much as I can of the bullshit generated internally in the media, here is my currently favoured hypothesis (based in large part on what Chris Goodfellow was saying on his Google+ page) - most of which fits the few known facts, and none of which requires anything currently known to be impossible:-
There were two 'events', separating the flight into three phases.
Phase 1: Normal, routine take-off and climb out of KL. Having reached cruise altitude (about FL350), the aircraft settles into cruise mode, and as they depart the Malaysian ATC zone, the First Officer says 'goodnight' to the Malaysian controller.
Event 1: A severe event occurs which disables a large fraction of the aircraft's electronics. A fire, structural failure, meteorite impact, or similar - there are plenty of things that can go bad, killing the electronics but not causing the plane to crash immediately. Perhaps an oxygen tank ruptures and takes out a cable bus under the cockpit. Whatever it was, either the event itself, or the aircrew's response to it, results in the loss of transponder and VHF (a fire might lead the crew to pull a number of circuit breakers, for example).
Phase 2: The captain is an experienced local pilot. He decides that rather than turn back to KL, which is a busy airport on the other side of some high terrain (and bearing in mind that he cannot raise ATC to have a slot cleared for his aircraft to land), he will head for the slightly closer, larger, and far less busy airport at Pulau Langkawi. He knows that airport, and he knows the area; but with much of the avionics shot, and flying at night, he keeps the navigation a simple as possible, turning left and crossing the Malay peninsula somewhat North of the target airfield, at the point where the terrain is at its lowest, and knowing that even if he can't locate the airport electronically, he can simply follow the coast south once he reaches the Strait of Malacca, until he sees the runway lights. Unbeknownst to the pilots, they are tracked some of the way by military radar; however none of the radar operators are aware that a civillian aircraft is missing at this point, and so no particular notice is taken at the time.
Event 2: Having reached the straits, the pilot attempts his Southerly turn; but at that point, something else goes wrong. Perhaps weakened by the earlier incident, the cabin pressurization falls to the point where the crew succumb to hypoxia; perhaps the control systems for the aircraft go the same way as the R/F gear and simply stop working; but the aircraft, now no longer under the control of its crew, becomes stuck on a southerly heading.
Phase 3: With all on board deceased, incapacitated, or (less plausibly) simply unable to do anything to influence the heading of the aircraft, it flies on into the night, until it runs out of fuel. The Satcom transceiver, the only remaining functional comms gear on the aircraft (which will work as long as it has power, even if no signals can be sent from the cockpit to the transceiver) continues its hourly handshake 'ping' with the INMARSAT bird at 64E, the last 'ping' coming shortly before the crash, at a point on the Southern 'arc' currently under investigation by SAR.
Of course, there are lots of other possibilities; but at this stage, none fit the known facts as well as the above, with a minimum of speculative additional 'facts' not currently in evidence.