What are the options?

   a) noone looks at the code
   b) they make random user-facing changes
   c) random churn that makes the code better (preferably with as few user-facing modifications as possible)

IMO the best option is c). Guy is literally doing the single best thing possible.

Better than current Xorg activity ;) (which is very low)

Eh, I don't know. There's an argument to be made that no changes is preferable to breaking changes with little or no benefit.