Science in the Sands reports how a US Professor, Dianne O'Leary, wrote to the publisher, Springer, asking for a PDF of her own paper. They refused and as her university doesn't subscribe to the journal, because it is prohibitively expensive, she couldn't get a final copy of her own article.
A later update reveals that, she could get the copy, but you have to ask the right person. This is something I have often encountered – getting the definitive copy, as quite often the last version you hold on your machine is not one that has been through final edits. So even when she could get the copy, it took a long time to obtain it, you had to find the right person to ask, and it often comes with limitations on use (such as you can't post it on a website).
Now, contrast this with a recent experience I had with my Digital Scholar book, which is published by Bloomsbury under a Creative Commons license. I realised the other day that I too, didn't have a PDF version of the final book. The version they make available is in HTML. I knew that the book torrent site Library Pirate had made a version available. As it is a non-commercial license, this is perfectly legitimate (I think). So I went to their site and downloaded a 'pirate' version of my own book. Within minutes I had a PDF (Download TheDigitalScholar), which I could do with as I wish.
As an author, which process is preferable?