Huh?

A documented API is considered by Apple to be like a contract. By documenting it, they are promising that they will try their hardest not to change it in any incompatible way in the future, and any deviation from this would be considered a bug (and, depending on the impact, a showstopper).

There are lots of reasons why an undocumented API may remain in that state. Often it’s because the API as it currently exists only works properly in a limited subset of the scenarios which the parameters suggest it might. Or it might be that it’s a poor fit with the rest of the API in a given area. Or it’s because the developers intend to change it in the not hugely distant future.

Apple regularly opens up its private APIs once they release maturity if there’s a solid case for doing so, and did precisely this when Snow Leopard was released (Quick Look, for example, moved from PrivateFrameworks to being part of QuartzCore with the arrival of 10.6).

As the old adage goes… don’t make a promise you can’t keep.

Errr

Is “the Complete Works of Shakespeare”, packaged in printed book form copyfraud?

No.

What about when it's an iPhone application? Or in nicely-formatted PDF?

No.

Actually, the only point at which it's copyfraud is if somebody takes a public domain work, republishes it in some way, and then uses that to claim copyright over the original public domain version. The author here doesn't seem too bothered by that scenario, however; more the fact that Google's system makes the assumption that if a copyrighted edition exists, then that precludes a public domain version existing, until some manual intervention tells it otherwise. This is called “making sure you don't get sued for unauthorised redistribution of copyright-protected works”. Google Books is not Project Gutenberg.

In Flickr's case, it's taken—what—4 years for this to crop up? And even then, it's an issue with ONE site with one very specific class of images—those which legally CANNOT be copyrighted at any point. That's a pretty rare occurrence in the world of Flickr, and it's good that they both recognised it and made provision for it.

Creative Commons’ stance is quite obviously about recognition more than anything. Huge numbers of people don't know what “Public domain” actually means, and this is confused even more by the use of terms like "copyleft" (which is often assumed by the layman to mean 'not copyrighted' rather than 'a license which deliberately subverts the usual protections of copyright'). In contrast, "Creative Commons" has become recognisable as "you may, possibly with some conditions depending upon who you are, do various things with this work including modification and redistribution"—with different variants of CC indicating who is permitted to do things and what they're permitted to do. CCPD, therefore, neatly fits into this with a variant that indicates “anybody” and “anything” respectively.