Speaking personally, I share your views that “all rights reserved” is closed even when these resources might be accessible at no cost. You are certainly not alone in holding this view.
Sadly, licensing is not always an effective mechanism to regulate intent. Remember Flatworld Knowledge the open textbook publisher? From recollection, most of these texts were published under a CC-BY-NC-SA license (under different versions of the license, ie. V1.x to V. 2.x). When the copyright holder decided to change their business model from free online access, a few folk from the free culture took the initiative to archive digital versions of these openly licensed texts (see for example http://2012books.lardbucket.org/). Creative Commons licenses are irrevocable and as you point out, it is possible to make copies to your own repository.
However, under Creative Commons licenses, the copyright holder may request removal of attribution in the case of a collection or adaptation up until V3.0 of the license. Under 4.0 the removal of attribution is on request of the Copyright holder (in other words it is not required to be a modified work before the request is permissible.)
In this example the licensor requested the lardbucket site where copies of the texts are hosted, to remove the attribution requirement from the collection of books hosted on the site.
It may seem trivial to remove the attribution requirement because access to the original text is still available on the site. However, in practical terms, this is a good example of a form of enclosure of one of the most restrictive Creative Commons licenses because it reduces the perceived value of the text. I am aware of a course development here at Otago Polytechnic which was based on one of these open textbooks. The legal requirement by the licensor is to remove attribution. References to this collection should be attributed as anonymous. As a respected education provider, we did not feel comfortable prescribing a text authored by anonymous authors. Moreover, as we will be charging assessment and student fees, any derivatives of the original materials may be deemed by the licensor to be used for commercial activity and therefore infringing on the original license. Granted, this is an unusual case, and we should also acknowledge that “all rights reserved” copyright holders would also have the rights to remove the attribution requirement.
When I develop open courses, as an open educator, I refrain from using OA texts unless they are OER - but this is a personal choice. I don’t mind putting in the extra work to ensure the future freedoms of the course materials by finding or developing OER substitutes. However, I think there are pragmatic reasons why course designers may choose to develop open courses which use OA resources. For example, many university libraries provide access to “all rights reserved” tutorials on citation methods. I think its unlikely that these institutions would enclose these materials and there will be a large number of alternatives available on the open web. Similarly, the growing number of Governments which require research outputs funded by taxpayer dollars to be published in open access journals will reduced the risks of possible enclosure in the future of these OA peer reviewed journals.
I guess what I’m saying is that designing and developing OER courses is more than just choosing the “right” license. Educators need to assess the risks of enclosure and how this may impact on the learning experience.