Once again, "follow the numbers" is a good idea here. Because...there are no numbers. There are numbers for hits on a page and clicks on a link, and downloads of files (in some cases) but there is no tracking of how many people who read something online then go out and buy it. (Not even self-reporting numbers on websites that carry free downloads: no little checkboxes for "And now I'm going to go buy this book" or "Heck no, I'm not going to buy this book.") There is no evidence--none, zilch, zip--that anyone's shown me that putting up the whole text increases sales of that book. (There is evidence it makes the writer's name better known, but not that it increases sales of that book.)
And--based on long-term marketing and my own experience with my work (things I've done online, things my publishers have done online, and things people have done without permission online) it's highly unlikely that offering unlimited quantities of any product free will produce enough sales to offset the loss of sales. Because of the nature of digital files, availability really is unlimited...it's not like the traditional freebies publishers give away to reviewers and random attendees at conventions, where the publisher could budget x-many copies for advertising purposes.
Cheap goods drive out expensive ones (unless you're rich and buy expensive goods for status.) Most of us have limited spending money. We want more for less. We buy at bargain prices, at sales, when we can find them (have you ever told a store "Oh, no, I'd rather pay full price...never mind the 30% off..."?) Hence the success of Wal-Mart. Lower prices attracted more customers; more customers gave Wal-Mart more clout with producers. The universal desire to get more for less means that American jobs are lost and the metal in our tools is much lower quality than it was fifty years ago (it's expensive to make high-quality steel--remember, I started in a hardware store. I know tool metal and fastener metal by touch, smell, taste--and most of what we get now is crap.) As recent problems with products out of China have shown, "cheap" isn't always "good" or even "marginally acceptable and safe"...but by golly things are cheap. (I don't recall whether it was Thoreau or Emerson who said that there was always profit in putting sand in sugar and chalk in flour...and there is.)
So the existence of unlimited freebie files will--by all the history of commerce since ancient times--result in people choosing the cheaper alternative. It's cheaper to download that freebie file than to buy the book, and easier than going to the library or bugging the library to buy a copy. The file may be adulterated, full of errors from careless scanning, but (like badly made clothes from a sweatshop in Honduras or India, like the mishmetal screws that bend when you put any real stress on them) you may think "Oh, it's good enough." You may tell yourself that one missed sale doesn't matter...you may have ideas about a writer's life that let you feel envious and justified in "sticking it to them" by choosing an alternative that doesn't give the writer a sales credit. And so the slow hemorrhage begins. The death of a thousand cuts, no one of which is fatal, but in combination causes fatal blood loss...just like the independent bookstore, the hometown department store, the local factory that produced high-quality parts for larger companies' products...the writer's career shrinks and finally vanishes.
Where can free downloads be helpful instead of harmful? Early in a writer's career, partial works--the opening chapters of a book, a few paragraphs of a story--do expose that writer's work to more potential readers, do build name recognition, do increase sales of that book. Free samples (from a piece of sausage to a test drive in a car) have long been shown to work in getting people to try new things. My experience with posting one to three chapters was very positive, as I've said. Posting related material on the writer's own website (background, artwork, etc.) is a help, and so are publisher websites that have room for the writer to expand a bit on his/her work. Online interviews, podcasts--all these are parallels to traditional advertising and they do have a favorable impact. Posting an entire book, even for a minimal fee (Against the Odds in Baen's Webscription program) did not produce a "bump" in that book's sales that exceeded what could be expected from putting up three chapters. Moreover, with the increase in illegitimate file sharing, I've seen slower than expected rises in sales with books people express enthusiasm for--the book industry as a whole reports falling sales. It's not a steep dive yet, but it's troublesome. I have had, as I said before, communications from people who expressed glee that they could read my books without profiting me in the slightest...the "gotcha" phenomenon. If one bothers to write me that, I'll bet there are others, and that's not counting the people who, having read the book online or in a downloaded form, just don't want to spend the money on the paper one.
In the long run, a new publishing paradigm will emerge, one that gives writers more control over online publication. Some writers have already experimented with this (I was going to, about eight years ago, but some family stuff ate my time and I didn't ever get it done.) Publication by subscription, on the writer's own website, has now worked for more than one writer. This may be either serial (for writers whose work doesn't change drastically during the writing--mine does!) or all one piece at the end, after the money's been collected. However, right now the income from these experiments does not match (for most of us) what traditional publication produces...and it requires the writer to spend time on tasks which were previously handled by the publisher. (Editing, copyediting, etc. Most of us benefit from editorial help--I know I do.)
In the meantime, readers need to understand the harm that copyright infringement does, and refrain from supporting it. If you like someone's writing, buy their books. Or at least bug your library to buy their books. Or download them from legitimate sources (which will probably cost something, but probably not nearly what a hardcover would.) If the download source is the publisher, or an e-rights subsidiary, the writer will get credit for a sale (not much money, but it counts as a sale when the bean-counters are deciding which writers are acceptable.) Publishers don't give writers any credit for how much or your stuff is downloaded from pirate sites.