Stabilization is way too easy, especially with spare the dying. I've been DMing since the beginning of the public playtest, and I've only seen four character deaths: One when a player foolishly ran off on his own during an invasion of a fortified beachhead, two from a critical failure on a death save while the cleric was incapacitated, and one from energy drain (which kills you at 0, bypassing negative hp).
The question unasked is how common PC deaths should be. Given the plan for modular play, it seems reasonable to set a core rules default, and include modules which alter the lethality.
The yo-yo effect is easily avoided if the cantrip instead stabilizes the character, but requires he heal from zero before being able to rejoin combat, which is the Pathfinder approach.
If I were designing the game with the objective of growing the player base, then I'd lean to core rules that carry a limited risk of PC death, and greater lethality being a module which can be adopted with full knowledge that it will mean more frequent character deaths. It's not a lot of fun for the new players to spend more time making new characters than playing the game. I'd also be focusing on the potential for loss without death - if you survive, but the task you set out to achieve is not accomplished, that is still a loss.
But I've never been a fan of meat grinder play where the players keep rolling up new characters to throw at the near-impossible challenge until someone gets lucky and makes it through. Some gamers prefer a much more lethal style, which can be accommodated through modules upping lethality. The option of a very lethal default and modules to enhance survivability also exists, of course, but the default should target enjoyment of the game for new players. To me, that's a pretty low lethality setting.