Especially now that pages are not stored as plaintext files, some API (use REST) to allow other software to access the plain content of pages (the option to have the page content as HTML, or stripped of markup would also help).
Why? I would like to use a wiki as my PersonalInformationManager?, and this seems like the most generally-useful way make it easy to develop software that talks with wiki. --JohnAbbe?
(discussion of this on MeatBall:XmlRpc and MeatBall:PersonalInformationManager)
Great, Les! Now, could you code a REST API?