Well, look at it this way. If I want to own a copy of Wuthering Heights (for example) I can go and buy one at the used book store, or I can go and buy a retail copy in a chain bookstore. My costs will range from $2-3 for a 'thrift' edition to$15 or so for an annotated or footnoted scholarly edition, all the way up to first editions or whatever which I am not counting in the equation.
If I want to *read* a copy of Wuthering Heights, I can rent it from the library at zero cost, or I can download or from Project Gutenberg or Feedbooks or Manybooks or the interface of my preference for 100% free (assuming one already has the internet and computer, of course---if one were buying them solely for the purpose of acquiring Wuthering Heights, that would be a different story).
So if I take this free copy of Wuthering Heights and read it on an e-device, I am reading a book on that e-device and getting that use out of it, so it is fair to count that in the equation---and reading a certain percentage of these free books does help to amortize the cost of the device over time and subsidize the money you spend on other books because if under the old scheme I had to pay $5 for Wuthering Heights anyway and now I don't, I have that extra $5 to spend on something different
But there are many books in the public domain and it would be unfair to say that *all* of them count as part of your library just because you *can* download them. For example, I have no plans at all to read Dante's Inferno in original Italian (I don't read Italian). so it being available on-line for free is not subsidizing my library costs at all because there is no $5 in my head that had ever been allocated which is now free. However, there are many works I do plan to read, and if I get my legitimate use out of my device reading them, I do count that in my total.