> The prime difference here is that most browsers do not throw markup errors or halt execution upon a bad tag; they're extremely lenient. Conversely, direct par

> The prime difference here is that most browsers do not throw markup errors or halt execution upon a bad tag; they're extremely lenient. Conversely, direct parser access via JavaScript can throw a host of errors and halt execution (as I noted with IE). In practice, this just isn't really a problem. The situations in which it can happen are specific and limited enough that I don't know if you got really unlucky sometime or if you're exaggerating for effect, but I've been writing JavaScript since the '90s and have never found this to be much of an issue — particularly since libraries can easily smooth over it. And if you think innerHTML can throw errors, you should check out the DOM API. The set of valid inputs to the DOM API is far more limited than for innerHTML. document.createElement("oh no") will throw an error in every browser around.