It's a little more complicated then this so I wanted to quickly expand so this conversation has a few facts which have been missed.

There are two elements to this code.
1) Link hijacking and history API (generally stable)
2) Loading a page through API and rendering it. (has bugs)

We use 2) after making an edit to avoid a page reload. This is good. This has been in stable but a lot of the bugs it has go unnoticed and usually get reported in alpha. So your assessment that it is not stable enough for beta is not quite accurate.

I don't think I personally waste time on bugs here (most get back logged) but alpha helps us identify a lot of issues in the editing interface that may go unnoticed.

Also if we ever want to serve a Firefox OS app via the mobile site we will need to keep iterating on this code.

I don't think we waste significant resources on this code but others feel free to disagree.