To paraphrase Steven Hawking's publishers, every turn made by flight 370, later than say 5-10 minutes after the transponders went off, cuts the plausibility of accident theories roughly in half.

That does raise the question of how many turns there really were, though. Earlier reports based on radar suggest that the plane made several course changes as it flew in a zig-zag around the Malacca Strait and Andaman Sea west of Malaysia. If enough of those apparent sightings can be knocked down then an accident becomes more likely. Though even one change of direction onto one of the final SATCOM arcs, apparently without any effort to communicate at the same time, is a real head-scratcher if you assume no foul play at any stage.