In March 2003, I almost ruined myself financially by giving a book away. Jeff Carlson and I decided to because sales had been so poor, and we thought it would be a nice gift to the world.
My mistake was in not calculating the bandwidth bill ahead of time. I hosted the file on a server run by a friend who had bought my Web hosting company, and his co-location provider, Level 3, charged based on sustained average usage across a month, dropping out about five percent of the busiest hours and using the 95th percentile as the threshold for what to bill. I nearly pushed the monthly bill from $2,000 to $15,000.
I managed to pull the plug in time, but only after about 10,000 downloads of a nearly 20 MB PDF. Through hearing about my plight, people contributed about $2,000 which I sent (with their permission) to Project Gutenberg when it turned out I wasn't being bitten by Level 3's pricing. You can read the whole story in Adam's article for TidBITS, "" (2003-03-24), or .
I had this all in mind recently when Adam and I decided to give away "," a book we last updated in 2004. The book sold okay - we earned out our advance from Peachpit Press - but not well enough for Peachpit or us to feel it was worth another revision in print.
We had adapted part of the book into "" (which we just released in significantly updated form last week), and I wrote a separate book, " ," which has since evolved into " " (also heavily revised and re-released last week).
But we felt that the bulk of "The Wireless Networking Starter Kit" was still useful, with advice on planning a network, putting the pieces together, troubleshooting network problems, long-range wireless connectivity, and more. We thought people around the world might enjoy a free book on the subject, and if they needed current information on security and other topics, perhaps the release of the free book would translate into a few extra sales of our Take Control books.
In this case, I created a 14 MB PDF, and ran the numbers: Where could I host a book that might be downloaded 10,000 or even 100,000 times without being cut off or breaking the bank? Though I'm a .Mac subscriber, for my regular account could have been exceeded, although I would have paid nothing extra for it. I host my own Web sites with , also TidBITS's Web host, and could have put the file there and paid digital.forest for the transfer if I went over my monthly allotment. But that seemed to be a bad bet if the transfers went sky high, and it seemed a silly way to spike traffic on digital.forest's networks if that happened.
In the end, I opted for service, which allows arbitrarily large amounts of storage with straightforward fees: $0.15/GB stored each month; $0.18/GB downloaded up to 10 TB/month (at which point prices drop); and $0.10/GB for uploads. (I use  to manage my S3 storage. S3 is a bit obscure, but it's very much like FTP in its basics, once you overcome the XML and other abstractions that Interarchy handles for you.)
At those rates, 10,000 downloads of a 14 MB file is 140 GB, which translates into a whopping $25. While 100,000 downloads seems unlikely, the $250 that would cost will hopefully be outweighed by sufficient Take Control book sales. So far, we've sold a handful of additional Take Control books and have hit over 50 GB of downloads, for less than $10 in charges.
There are services and methods of hosting content at no cost, too, but all of them have tradeoffs, and I felt the S3 cost was low enough to provide robustness and not abuse anyone's free network.
In the space of four years, we've gone from $15,000 for a sustained level of high bandwidth to $25 for a large hunk of bandwidth. Rates that used to be $10 to $40 per gigabyte four years ago are now commoditized in part due to Amazon's S3 and other services.
To me, this sort of change is just as significant - and perhaps more so - as the growth in computational power and the sophistication of online media. With faster computers, more video, and cheaper bandwidth, the power of individuals to disseminate their ideas in any form has increased as well.
The real question is whether cheap bandwidth means more mass-produced content brought to us faster, or a continued growth in personal expression. While giving away an ebook isn't the same as composing a new opera, filming it, and having an audience of a million Web watchers, the difference in infrastructure is slight.