All of the good research I've read on online learning is done with self-selected groups of adults who would probably learn well regardless of their modality. All of the research I've read on K-12 studies suggest that computer use in the classroom lowers students' scores (though computer use as a supplement outside of the classroom can improve scores).
Besides that, this ethereal "software that provides a personalized learning experience just right for their abilities" doesn't exist.
I'd really really love for there to be some magical software that can figure out what a child needs and start from there, and we could probably hit a decent percentage of the population, especially those children created in our own (autodidactical) image, but if you spend any time in K-12 schools (I'm a founder and currently on the board of our local charter school and have 4 kids of my own), you'll notice that children have vastly different learning abilities, interests, needs, etc.
I would guess that more than half of what a K-12 teacher does doesn't have anything to do with teaching facts, but teaching about life and how to motivate, how to behave, how to stop flicking boogers on your friends, how to deal with disappointment when you can't "get it", etc. Learning for children is much different than learning for adults. And we're not even addressing more subjective education such as creativity, synthesis, and higher-order thinking skills. These don't translate well to computers because they require direct human judgement and some subjective evaluation.