![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Popular SF
Jan. 8th, 2015 12:39 pmI see (via File 770) that the Sad Puppies slate looks as though it will be back for another year.
The logic involved is somewhat tortuous, but a big chunk of it seems to be tied up in the view that because the Hugo awards are a fan-based majority vote, popularity as shown in sales should match up with ranking in the nominations, and that therefore the usual nominees are somehow rigged by a clique… something like that.
The Hugos (as I've noted before) are not simply a popularity contest; an examination of the general pattern of nominations shows that voters do filter, to some degree, by a perceived quality in a way different from simple sales. Selecting "the book I liked best from the year" is not the same as selecting "the book that I read that sold the most copies last year". Every year, pretty well, there's a scuffle over the nominees regarding their relative perceived or real inferiority to a commenter's preferred slate, and that's the usual context for this question. The Sad puppies complaint is a little different: they want the vote to reflect simple popularity, and think that the books that they like are popular enough that their exclusion is an injury to them.
What do we mean by popular?
When people talk about a book being popular, we can mean two things: first, (usually) that a lot of people buy it (i.e. it's a "bestseller"), and secondly, that people like it a lot. These are not the same thing.
Bestseller lists get manipulated or otherwise mangled. Most involve some form of human judgement to filter them, even if only in selecting sources; Amazon generates its top 100 lists automatically, but it has to use some form of algorithm to age entries so that recent sales of new books are significant but ongoing sales of popular books also count, and once one starts to get into categories it gets a bit blurry -- books will sometimes be allocated to a "wrong" category and there may be multiple instances of the same book in different formats. Measures of how much people like things involving ratings are extremely difficult to calibrate (some people rate almost all books at 5/5 or 4/5, and this may reflect merely a reluctance to grade down, not that they think that hundreds of books are all the best thing since sliced bread).
However, LibraryThing has a "Top Five Books of 2014" set of lists, with (as of now) 962 books entered on members' lists. If we ignore internal rankings, and compare against a snapshot of the Amazon lists where (again) we don't worry about fine rankings but simply about presence on the list, we can come to some limited conclusions.
(I've taken the SF and Fantasy separate lists from Amazon, giving me a potential set of 200 books, although in fact (once I get rid of media tie-ins and duplicates) it's a lot shorter. I also eliminated Kindle books at less than 8.99, where popularity is artificially pushed up by comparison to other books by a low price.)
First, a lot of books appear on only one list and not the other. More precisely (because given the constraints we're comparing a list of 960 to a list potentially of only 200 items and actually of a little over 100 titles), most of the bestselling novels do not appear on people's favourite lists. Only 17 books are within the intersection of the two sets.
Secondly, although both lists have a heavy mixture of non-recent books, only a few books are older than the last five years (Oryx and Crake, The Sirens of Titan, The Hobbit, The Lord of the Rings). (Older books on the Amazon list look suspiciously, as a class, as though their numbers are inflated by course assignments; this is obviously not a major concern for the LibraryThing list.)
Thirdly, although the LT list had most of my preferred fiction of last year (it missed out of Saunders' The March North, which wasn't surprising, given its low readership), only one of those books was also on the Amazon lists: Ancillary Sword.
As an interesting additional datum, The Goblin Emperor was Bakka's number one hardcover bestseller for 2014, and Watts' Echopraxia is #6, but neither appear on the recent Amazon rankings (I know, apples and oranges, as one is cumulative over a year and the other is a cross-section at one point of time).
If I apply the Bakka cumulative list as a third filter, I'm down to "countable on the fingers of a mutilated hand": Ancillary Justice and The Slow Regard Of Silent Things.
(If I match the Bakka data against the LibraryThing Data and don't consider Amazon, 5 of the books are in the top 500: The Goblin Emperor, The Slow Regard of Silent Things, Skin Game, Lock In, and Ancillary Justice.)
The first thing that this all tells me is that except for narrow purposes the Amazon data is largely, but not entirely, useless. It is, in one context, useful to know that Tolkien and Martin continue to be highly popular authors, or that Star Trek / Star Wars spin-offs continue to be a licence to print money: especially if you translate that into the general conclusion that most of the profitable print media sales are driven by exposure on TV and in the cinema.[1] However, if you're the sort of person for whom that is useful (e.g. a Managing Editor at an SF imprint trying to work out a model for weighing publications in the next year's budget) you probably already have much better data than Amazon provides, e.g. via Bookscan and your own sales figures. For anyone else, the redundancy in products and the limited scope of the list -- a lot of very popular novels might make the list in a blip, if at all -- makes it not very useful. The further manipulation by including low-priced Kindle books further distorts any signal as regards quality or trends in taste.[2]
The second thing is that once one uses two or more filters of a different type, it tends to select works of high or at least highish quality (although not necessarily works to everyone's taste), but it cuts off too high. If you're looking for advice as to what to read and you follow any general review site at all or are aware of what is managing to make the NYT extended bestseller list you'll already be aware that all of the works selected are hugely popular, and (with a few exceptions, such as Leckie) tend to come from writers with a track record of years of popularity.
If I were to select books from any of these lists with an eye to ones which I think will last (ignoring ones like those by Tolkien and Vonnegut which already have lasted), I'd certainly be discarding a good number of even the top entries and probably promoting a number which don't make the very top. (It can be illuminating to look at the Locus awards runners up for, say, the 1970's and 1980's and realize that many of these books/authors were popular enough to make an end-of-year best list but have sunk without a trace since; on the other hand, some that are relatively low down the list are still around.)
Amazon top 100 overlap with LibraryThing best books of 2014
Science Fiction:
1. The Martian: A Novel; by Andy Weir; Paperback; LT2
8. The Bone Clocks: A Novel; by David Mitchell; Hardcover; LT39
35. Oryx and Crake; by Margaret Atwood; Paperback; LT732
52. Ancillary Sword (Imperial Radch); by Ann Leckie; Kindle Edition LT334
53. Red Rising; by Pierce Brown; Hardcover; LT44
67. Ready Player One: A Novel; by Ernest Cline; Paperback; LT20
91. The Sirens of Titan: A Novel; by Kurt Vonnegut; Paperback; LT740
100. Ancillary Justice; by Ann Leckie; Paperback; LT4
Fantasy
1. A Song of Ice And Fire; by George R.R. Martin; Mass Market Paperback; LT118
6. The Hobbit; by J. R. R. Tolkien; Kindle Edition; LT147
13. The Lord Of The Rings; by J.R.R. Tolkien; Paperback; LT147
15. The Ocean At The End Of The Lane; by Neil Gaiman; Paperback; LT5
16. Written In My Own Heart's Blood; by Diana Gabaldon; Hardcover; LT82
18. The Slow Regard Of Silent Things; by Patrick Rothfuss; Hardcover; LT65
23. Colorless Tsukuru Tazaki and His Years of Pilgrimage; by Haruki Murakami; Hardcover; LT171
62. The Way of Kings; by Brandon Sanderson; Kindle Edition; LT113
76. Lies of Locke Lamora; by Scott Lynch; Paperback; LT72
Is the assumption that less popular books than the Sad Puppies are being selected true?
Now that we've established that the idea of "popularity" is slippery, let's look at the question of whether, in fact, less popular books are being preferred for the Hugos and whether the Sad Puppies preferred books are in fact more popular.
One can note from the list above that last year's winner, Ancillary Justice, is still in the top 100 Amazon bestsellers, and was ranked fourth on the LibraryThing list. Now it has undoubtedly had an awards boost: I doubt it (or its sequel) would be anywhere near as high on the current Amazon list if it had not won four major awards last year. On the other hand, the winning of four awards, and the high ranking by people who had read it, reflects an initial popularity (in both senses) which has merely created a feedback loop.
So let's look at the current popularity, as reflected by Amazon sales, among last year's novel nominees. To level the playing field as regards format (some books are MMPB and some only trade) we'll compare only Kindle rankings, which are more similar in terms of the price points.
(Order is the ranking of position 1, position 2, as determined by different runoffs.)
So Warbound, generally considered the best candidate on the Sad Puppies list, is fourth out of five by current popularity. (As regards the fifth, there's a reason Charlie is emphasizing The Laundry Files and the world-walker series: NB was considered relatively "difficult" and sold less well, for all of its favourable reception by a number of critics, Paul Krugman included. For that matter, it came in third on the Locus Award list, which is as close to a straight-up popularity poll as SF has, so it can't be considered all that unpopular.)
It's also clear that "popularity" in the sales sense isn't reflected in the order.
If we extend the consideration to all the nominated novels as shown in the final statistics, and order by Kindle store popularity, we get:
If the sad puppies want a champion, Ringo is a far more popular writer than Correia. By the same token, Sarah Hoyt is probably not a good choice. Just sayin'.
If the Hugos were a genuine popularity contest based on sales, Sanderson would be taking home the rocket, probably on a regular basis (his two Stormlight books are currently at 1197 and 801, for comparison to the above list). He's not, but seems to be quite happy writing books he wants to write. Of course, he's laughing all the way to the bank. That being said, Steelheart gets a boost in sales as a YA book, but that's a negative factor in Hugo contention.
I think we can close the book on the complaint that the sad puppies are being deprived of a prominence their popularity entitles them to.
The overall sense I get is that their complaint comes from an alternate universe, coloured heavily by a subjective sense that all those other fans out there can't actually enjoy works X, Y, and Z which are talked up and then nominated, and must be insincere. De gustibus, and it's hard to believe that you're in a minority. (One of the things that emerged from some of the online discussion last year is that there really are engaged readers who authentically like Carreia's and Day's work, and equally authentically dislike the Leckie and Stross novels. I can even see why this might be true. But projecting your own tastes onto the population as a whole is risky, and when feedback tells you that the two don't match and you still persist in doing so, it's a sign of cognitive dissonance.) [3]
[1]This is actually probably the most important takeaway from the entire exercise, although it's so big and obvious that any glance at bestseller lists would confirm it.
A quick scan of Twentieth Century bestsellers will show a mid-century pattern whereby some bestselling novels generate later films (e.g. for the 1950's: From Here To Eternity (1951, 1953, film 1953), The Caine Mutiny (1951, 1952, film 1954), The Silver Chalice (1952, 1953, film 1954), The Robe (1942, 1943, 1944, 1945, 1953, film 1953), Auntie Mame (1955, 1956, play 1956, movie 1958, musical 1966, film of musical 1974), The Man in the Gray Flannel Suit (1955, film 1956), Peyton Place (1956, 1957, film 1957, TV 1964-1969), On The Beach (1957, film 1959), Doctor Zhivago (1958, 1959, film 1965), Lolita (1958, 1959, film 1962), Advise and Consent (1959, 1960, film 1962), Mrs. 'Arris Goes to Paris (1959, TV 1958). Dates are for bestseller status, not date of publication.
With a few exceptions (From Here To Eternity, The Robe, Mrs. 'Arris...) the film appearances do not generate a (major) later bump in readership (I'm sure that in fact every book benefits somewhat when an adaptation comes out, but it's not generally reflected at this level.) The pattern is far more Sydney Carter's "I'm waiting for the film to come": substituting the more accessible film for the less accessible book. Novelizations of works which first appeared as films, if they existed, (the novel of 2001 (not a top 10 bestseller) would be an early instance) never hit the bestseller lists (the only entries I can find are for "ET" (1982), "Return of the Jedi" (1983), and "The Phantom Menace" (1999)).
The newer pattern is of books which sold well when they first came out, but have been pushed into semi-permanent bestsellerdom by TV or cinematic adaptations. (It's not that there aren't other kinds of bestsellers, but this kind and direction of feedback is relatively new.)
To refer to Amazon again: Martin's ASOIAF is at rank 3 for 2014, 10 for 2013, 5 for 2012, 3 for 2011, and not on the list at all for 2010 (the TV series began in 2011). That's out of all books, by the way, not just SFF, as Amazon doesn't break out past bestsellers in fine detail. (Jeff Vandermeer must be hoping for a good adaptation of his Southern Reach Trilogy.)
North America is becoming a less print-driven society and a more "media"-driven society. This is actually a much bigger issue for publishers than the advent of e-books. The workers who used to take books to read on vacation are now as likely to take DVDs, or, more likely, expect to be able to stream video on demand.
The same observation can be made regarding YA series, beginning with Harry Potter, with a book series -> film -> books feedback loop.
[2]I once worked at a second hand bookstore, and am well aware that some customers will read just about anything in a category (e.g. romance, western, space opera, gaming-based fantasy) as long as it meets pretty minimal standards of readability and supplies a similar experience to the last book read. These people made up a good part of our bread and butter -- they would purchase N books a week and return the N books from the previous week -- rather as functional alcoholics provide the core market for the LCBO, accounting for the dominance of cheap just-drinkable product on their shelves. Very low priced e-books and subscription services cater to this market.
[3]That's one reason I reference the LibraryThing figures: there's no obvious skew in favour of fandom per se in their makeup. For the record, there are by my count 19 SF/F books in the top 100 of the collated top five books of 2014 lists; looking at the comments to Torgersen's post, about 3 of these would fall within the ambit of the works which they would aim at promoting (Sanderson, Butcher, and Weir).
The logic involved is somewhat tortuous, but a big chunk of it seems to be tied up in the view that because the Hugo awards are a fan-based majority vote, popularity as shown in sales should match up with ranking in the nominations, and that therefore the usual nominees are somehow rigged by a clique… something like that.
The Hugos (as I've noted before) are not simply a popularity contest; an examination of the general pattern of nominations shows that voters do filter, to some degree, by a perceived quality in a way different from simple sales. Selecting "the book I liked best from the year" is not the same as selecting "the book that I read that sold the most copies last year". Every year, pretty well, there's a scuffle over the nominees regarding their relative perceived or real inferiority to a commenter's preferred slate, and that's the usual context for this question. The Sad puppies complaint is a little different: they want the vote to reflect simple popularity, and think that the books that they like are popular enough that their exclusion is an injury to them.
What do we mean by popular?
When people talk about a book being popular, we can mean two things: first, (usually) that a lot of people buy it (i.e. it's a "bestseller"), and secondly, that people like it a lot. These are not the same thing.
Bestseller lists get manipulated or otherwise mangled. Most involve some form of human judgement to filter them, even if only in selecting sources; Amazon generates its top 100 lists automatically, but it has to use some form of algorithm to age entries so that recent sales of new books are significant but ongoing sales of popular books also count, and once one starts to get into categories it gets a bit blurry -- books will sometimes be allocated to a "wrong" category and there may be multiple instances of the same book in different formats. Measures of how much people like things involving ratings are extremely difficult to calibrate (some people rate almost all books at 5/5 or 4/5, and this may reflect merely a reluctance to grade down, not that they think that hundreds of books are all the best thing since sliced bread).
However, LibraryThing has a "Top Five Books of 2014" set of lists, with (as of now) 962 books entered on members' lists. If we ignore internal rankings, and compare against a snapshot of the Amazon lists where (again) we don't worry about fine rankings but simply about presence on the list, we can come to some limited conclusions.
(I've taken the SF and Fantasy separate lists from Amazon, giving me a potential set of 200 books, although in fact (once I get rid of media tie-ins and duplicates) it's a lot shorter. I also eliminated Kindle books at less than 8.99, where popularity is artificially pushed up by comparison to other books by a low price.)
First, a lot of books appear on only one list and not the other. More precisely (because given the constraints we're comparing a list of 960 to a list potentially of only 200 items and actually of a little over 100 titles), most of the bestselling novels do not appear on people's favourite lists. Only 17 books are within the intersection of the two sets.
Secondly, although both lists have a heavy mixture of non-recent books, only a few books are older than the last five years (Oryx and Crake, The Sirens of Titan, The Hobbit, The Lord of the Rings). (Older books on the Amazon list look suspiciously, as a class, as though their numbers are inflated by course assignments; this is obviously not a major concern for the LibraryThing list.)
Thirdly, although the LT list had most of my preferred fiction of last year (it missed out of Saunders' The March North, which wasn't surprising, given its low readership), only one of those books was also on the Amazon lists: Ancillary Sword.
As an interesting additional datum, The Goblin Emperor was Bakka's number one hardcover bestseller for 2014, and Watts' Echopraxia is #6, but neither appear on the recent Amazon rankings (I know, apples and oranges, as one is cumulative over a year and the other is a cross-section at one point of time).
If I apply the Bakka cumulative list as a third filter, I'm down to "countable on the fingers of a mutilated hand": Ancillary Justice and The Slow Regard Of Silent Things.
(If I match the Bakka data against the LibraryThing Data and don't consider Amazon, 5 of the books are in the top 500: The Goblin Emperor, The Slow Regard of Silent Things, Skin Game, Lock In, and Ancillary Justice.)
The first thing that this all tells me is that except for narrow purposes the Amazon data is largely, but not entirely, useless. It is, in one context, useful to know that Tolkien and Martin continue to be highly popular authors, or that Star Trek / Star Wars spin-offs continue to be a licence to print money: especially if you translate that into the general conclusion that most of the profitable print media sales are driven by exposure on TV and in the cinema.[1] However, if you're the sort of person for whom that is useful (e.g. a Managing Editor at an SF imprint trying to work out a model for weighing publications in the next year's budget) you probably already have much better data than Amazon provides, e.g. via Bookscan and your own sales figures. For anyone else, the redundancy in products and the limited scope of the list -- a lot of very popular novels might make the list in a blip, if at all -- makes it not very useful. The further manipulation by including low-priced Kindle books further distorts any signal as regards quality or trends in taste.[2]
The second thing is that once one uses two or more filters of a different type, it tends to select works of high or at least highish quality (although not necessarily works to everyone's taste), but it cuts off too high. If you're looking for advice as to what to read and you follow any general review site at all or are aware of what is managing to make the NYT extended bestseller list you'll already be aware that all of the works selected are hugely popular, and (with a few exceptions, such as Leckie) tend to come from writers with a track record of years of popularity.
If I were to select books from any of these lists with an eye to ones which I think will last (ignoring ones like those by Tolkien and Vonnegut which already have lasted), I'd certainly be discarding a good number of even the top entries and probably promoting a number which don't make the very top. (It can be illuminating to look at the Locus awards runners up for, say, the 1970's and 1980's and realize that many of these books/authors were popular enough to make an end-of-year best list but have sunk without a trace since; on the other hand, some that are relatively low down the list are still around.)
Amazon top 100 overlap with LibraryThing best books of 2014
Science Fiction:
1. The Martian: A Novel; by Andy Weir; Paperback; LT2
8. The Bone Clocks: A Novel; by David Mitchell; Hardcover; LT39
35. Oryx and Crake; by Margaret Atwood; Paperback; LT732
52. Ancillary Sword (Imperial Radch); by Ann Leckie; Kindle Edition LT334
53. Red Rising; by Pierce Brown; Hardcover; LT44
67. Ready Player One: A Novel; by Ernest Cline; Paperback; LT20
91. The Sirens of Titan: A Novel; by Kurt Vonnegut; Paperback; LT740
100. Ancillary Justice; by Ann Leckie; Paperback; LT4
Fantasy
1. A Song of Ice And Fire; by George R.R. Martin; Mass Market Paperback; LT118
6. The Hobbit; by J. R. R. Tolkien; Kindle Edition; LT147
13. The Lord Of The Rings; by J.R.R. Tolkien; Paperback; LT147
15. The Ocean At The End Of The Lane; by Neil Gaiman; Paperback; LT5
16. Written In My Own Heart's Blood; by Diana Gabaldon; Hardcover; LT82
18. The Slow Regard Of Silent Things; by Patrick Rothfuss; Hardcover; LT65
23. Colorless Tsukuru Tazaki and His Years of Pilgrimage; by Haruki Murakami; Hardcover; LT171
62. The Way of Kings; by Brandon Sanderson; Kindle Edition; LT113
76. Lies of Locke Lamora; by Scott Lynch; Paperback; LT72
Is the assumption that less popular books than the Sad Puppies are being selected true?
Now that we've established that the idea of "popularity" is slippery, let's look at the question of whether, in fact, less popular books are being preferred for the Hugos and whether the Sad Puppies preferred books are in fact more popular.
One can note from the list above that last year's winner, Ancillary Justice, is still in the top 100 Amazon bestsellers, and was ranked fourth on the LibraryThing list. Now it has undoubtedly had an awards boost: I doubt it (or its sequel) would be anywhere near as high on the current Amazon list if it had not won four major awards last year. On the other hand, the winning of four awards, and the high ranking by people who had read it, reflects an initial popularity (in both senses) which has merely created a feedback loop.
So let's look at the current popularity, as reflected by Amazon sales, among last year's novel nominees. To level the playing field as regards format (some books are MMPB and some only trade) we'll compare only Kindle rankings, which are more similar in terms of the price points.
- Ancillary Justice, 1,473
- Neptune's Brood, 34,494
- Parasite, 23,439
- The Wheel of Time (as represented by The Eye of The World), 2,006
- Warbound, 30,604
(Order is the ranking of position 1, position 2, as determined by different runoffs.)
So Warbound, generally considered the best candidate on the Sad Puppies list, is fourth out of five by current popularity. (As regards the fifth, there's a reason Charlie is emphasizing The Laundry Files and the world-walker series: NB was considered relatively "difficult" and sold less well, for all of its favourable reception by a number of critics, Paul Krugman included. For that matter, it came in third on the Locus Award list, which is as close to a straight-up popularity poll as SF has, so it can't be considered all that unpopular.)
It's also clear that "popularity" in the sales sense isn't reflected in the order.
If we extend the consideration to all the nominated novels as shown in the final statistics, and order by Kindle store popularity, we get:
- Steelheart, 390
- Ancillary Justice, 1,473
- The Wheel of Time, 2,006
- The Golem and the Djinni, 2,033
- Under a Graveyard Sky, 2,628
- The Republic of Thieves, 4,109
- London Falling, 56,720
- Abaddon's Gate, 7,990
- The Shining Girls, 12,805
- Parasite, 23,439
- Warbound, 30,604
- Neptune's Brood, 34,494
- A Stranger in Olondria, 48,153
- River of Stars, 101,135
- A Few Good Men, 354,878
If the sad puppies want a champion, Ringo is a far more popular writer than Correia. By the same token, Sarah Hoyt is probably not a good choice. Just sayin'.
If the Hugos were a genuine popularity contest based on sales, Sanderson would be taking home the rocket, probably on a regular basis (his two Stormlight books are currently at 1197 and 801, for comparison to the above list). He's not, but seems to be quite happy writing books he wants to write. Of course, he's laughing all the way to the bank. That being said, Steelheart gets a boost in sales as a YA book, but that's a negative factor in Hugo contention.
I think we can close the book on the complaint that the sad puppies are being deprived of a prominence their popularity entitles them to.
The overall sense I get is that their complaint comes from an alternate universe, coloured heavily by a subjective sense that all those other fans out there can't actually enjoy works X, Y, and Z which are talked up and then nominated, and must be insincere. De gustibus, and it's hard to believe that you're in a minority. (One of the things that emerged from some of the online discussion last year is that there really are engaged readers who authentically like Carreia's and Day's work, and equally authentically dislike the Leckie and Stross novels. I can even see why this might be true. But projecting your own tastes onto the population as a whole is risky, and when feedback tells you that the two don't match and you still persist in doing so, it's a sign of cognitive dissonance.) [3]
[1]This is actually probably the most important takeaway from the entire exercise, although it's so big and obvious that any glance at bestseller lists would confirm it.
A quick scan of Twentieth Century bestsellers will show a mid-century pattern whereby some bestselling novels generate later films (e.g. for the 1950's: From Here To Eternity (1951, 1953, film 1953), The Caine Mutiny (1951, 1952, film 1954), The Silver Chalice (1952, 1953, film 1954), The Robe (1942, 1943, 1944, 1945, 1953, film 1953), Auntie Mame (1955, 1956, play 1956, movie 1958, musical 1966, film of musical 1974), The Man in the Gray Flannel Suit (1955, film 1956), Peyton Place (1956, 1957, film 1957, TV 1964-1969), On The Beach (1957, film 1959), Doctor Zhivago (1958, 1959, film 1965), Lolita (1958, 1959, film 1962), Advise and Consent (1959, 1960, film 1962), Mrs. 'Arris Goes to Paris (1959, TV 1958). Dates are for bestseller status, not date of publication.
With a few exceptions (From Here To Eternity, The Robe, Mrs. 'Arris...) the film appearances do not generate a (major) later bump in readership (I'm sure that in fact every book benefits somewhat when an adaptation comes out, but it's not generally reflected at this level.) The pattern is far more Sydney Carter's "I'm waiting for the film to come": substituting the more accessible film for the less accessible book. Novelizations of works which first appeared as films, if they existed, (the novel of 2001 (not a top 10 bestseller) would be an early instance) never hit the bestseller lists (the only entries I can find are for "ET" (1982), "Return of the Jedi" (1983), and "The Phantom Menace" (1999)).
The newer pattern is of books which sold well when they first came out, but have been pushed into semi-permanent bestsellerdom by TV or cinematic adaptations. (It's not that there aren't other kinds of bestsellers, but this kind and direction of feedback is relatively new.)
To refer to Amazon again: Martin's ASOIAF is at rank 3 for 2014, 10 for 2013, 5 for 2012, 3 for 2011, and not on the list at all for 2010 (the TV series began in 2011). That's out of all books, by the way, not just SFF, as Amazon doesn't break out past bestsellers in fine detail. (Jeff Vandermeer must be hoping for a good adaptation of his Southern Reach Trilogy.)
North America is becoming a less print-driven society and a more "media"-driven society. This is actually a much bigger issue for publishers than the advent of e-books. The workers who used to take books to read on vacation are now as likely to take DVDs, or, more likely, expect to be able to stream video on demand.
The same observation can be made regarding YA series, beginning with Harry Potter, with a book series -> film -> books feedback loop.
[2]I once worked at a second hand bookstore, and am well aware that some customers will read just about anything in a category (e.g. romance, western, space opera, gaming-based fantasy) as long as it meets pretty minimal standards of readability and supplies a similar experience to the last book read. These people made up a good part of our bread and butter -- they would purchase N books a week and return the N books from the previous week -- rather as functional alcoholics provide the core market for the LCBO, accounting for the dominance of cheap just-drinkable product on their shelves. Very low priced e-books and subscription services cater to this market.
[3]That's one reason I reference the LibraryThing figures: there's no obvious skew in favour of fandom per se in their makeup. For the record, there are by my count 19 SF/F books in the top 100 of the collated top five books of 2014 lists; looking at the comments to Torgersen's post, about 3 of these would fall within the ambit of the works which they would aim at promoting (Sanderson, Butcher, and Weir).
