Slashdot Mirror
J.K. Rowling Should Try the Voting Algorithm
Rowling confirmed (after the information leaked accidentally) that she had authored a new book, The Cuckoo's Calling, under the male pseudonym Robert Galbraith, which went on to sell only about 1,500 copies before she announced that she was the real author and sales of the book spiked 150,000%.
Stephen King actually tried something similar in the 1970s, publishing a series of books under the pseudonym "Richard Bachman," which he later said was partly an attempt to answer the question of whether his success was due to talent or luck. (The Richard Bachman books sold 10 times as many copies after King was revealed as the author.) Rowling has not said whether she was attempting a similar experiment, having issued a statement that before the revelation, it had been "wonderful to publish without hype or expectation, and pure pleasure to get feedback under a different name."
But if either J.K. Rowling or Stephen King really wants to find the answer to the question of talent vs. luck, the solution lies in the random-sample voting algorithm that I've been advocating in occasional articles for years now, going back to "Censorship By Glut" in 2006. Here's how the experiment could work, for evaluating the quality of fiction writing:
-
Rowling or King could approach a pre-established amateur fiction hosting site with a large number of registered users. Or they could create their own fiction hosting site and announce it to the world for the purpose of running the experiment, which would almost certainly attract a large number of users to sign up. (The experiment only works if the site has a large number of users, for reasons that will become clear.)
-
When a user submits a new short story to the site, the site randomly selects a small subset of other users on the site (say, 20 other users), emails them a link to the new story, and invites them to read it and rate its content. There are several ways you could incentivize those users to read the link and rate the story on a scale of 1 to 10. You could bill it as the "civic duty" of registered users of the site (in the same way that it's the civic duty of registered Wikipedia editors to maintain the quality of articles, even though the editors are working for free). You could require registered users to read and rate any stories that are emailed to them (although of course there'd be no way to stop someone from lazily submitting a rating without even reading the story). You could actually require payments from users who submit stories, and then use that money to distribute small payments to the raters as compensation for reading the story (although that seems like it would be the biggest headache, since you'd have to jump through legal and logistical hoops to set it up, and it would attract cheaters who would try to abuse the system just for the free small payments). But in any case, you don't need every user who gets emailed a story, to actually click through to read the story and rate it. All that matters is that out of those 20 users, enough of them click through to read the story, that you get a statistically representative sample of what users think of the quality.
Optionally, the story raters could also submit written feedback about why they liked or did not like a story. But the important part is collecting the numeric ratings so that they can be averaged into a single overall rating for the piece of content.
-
If a story gets a high enough average rating in the first round of voting, then it gets emailed out to a larger random sample of voters, say, 200. The ratings given by this larger sample can be used to distinguish the very best stories from the merely good. (We expect that for good stories, the ratings would tend to cluster around the high end of the scale, so with that smaller variance, it would take a larger sample size to find a statistically significant difference between the quality of two stories.)
-
The stories that get the highest ratings can be featured on the front page of the site, so that everybody can have the benefit of enjoying the "best" stories. Meanwhile, J.K. Rowling and Stephen King have the benefit of finding out how their stories compare against stories written by unpublished amateur writers.
It all sounds deceptively simple, but the important feature is that you've taken the arbitrariness out of the outcome. As long as your sample sizes are large enough, the rating that a story obtains in this system, will be approximately equal to the average rating it would get from all users across the site. "Luck" is no longer a factor, because you could re-run the experiment twice with the same set of stories, and get approximately the same outcome.
This is important, because numerous experiments and real-world studies have shown that in any environment where users can recommend content to each other and browse content that is already known to be popular — in other words, how most of us discover content in the real world — luck plays a much greater role in which content becomes wildly successful. The generally accepted explanation is that an initial stroke of luck can have a self-reinforcing snowball effect — if a few key influencers happen to discover and recommend a piece of content at the same time, their friends and followers will be drawn to that content as well, and once it crosses that threshold, the content has now become "popular" enough that even more users will be drawn to it just because it's popular.
This is also why any of the existing fiction-rating sites would not work for this experiment — because most such sites allow authors to invite their friends to sign up and give high ratings to their stories, or to form cliques that all give high ratings to each other's writings. It's usually in the site's best interests to allow these tricks, because it gives authors the incentive to promote the site to their friends in order to get them to sign up. But it also means that (a) authors can easily game the system, and the highest-rated stories may not be the highest-quality ones but the ones whose authors simply play the game the best, and (b) even without "gaming the system", the fact that users can see other users' ratings and can seek out "most popular" or "trending" stories, creates the snowball effects discussed above, and introduces a huge amount of arbitrariness into the process.
Duncan Watts' excellent book Everything Is Obvious (Once You Know The Answer) is an excellent introduction to the arbitrariness phenomenon, but if you don't have time to read the whole book, just read about the Matthew Salganik 'many-worlds' experiment", which Watts co-authored and which I've linked to in pretty much every other article I've written about the random-sample-voting algorithm. The gist was that if you divide users into multiple artificial "worlds," where users can recommend content only to other users within those worlds, and seed all artificial worlds with the same content (in this case, songs), then songs which become wildly popular in some worlds will become duds in others.
The whole of Everything is Obvious is at least as insightful as anything ever written by Malcolm Gladwell, and would appeal to the same people, but it never became a bestseller, because — well, probably because we live in one of the many possible worlds of a Salganik experiment, and in the world we happen to live in, the luck of the draw meant that book didn't take off.
But back to the proposed experiment. It is true that the votes of the average users would not tell us anything about whether the winning stories were "artistically" good, however you define that. But in King's case, he was not trying to answer questions about artistic merit. he was trying to find out if his bestselling-author status was due to talent or luck, so the average rating from regular readers would be quite on point. Rowling said that she wanted to write without any hype and receive honest feedback, and it's hard to imagine a better place to do that than writing under a pseudonym for a fiction site that distributes your content directly to the public.
Both King and Rowling deserve some credit for even addressing the question of whether their success was due to talent or luck. It would have been easy for them to assume that their global success was due to their innate skill and hard work, and 99% of the world would have accepted that explanation, so it took no small amount of courage to even raise the question of how luck might have played a role. (We all know plenty of successful people who take umbrage if you even mention "the L word".)
But King did say that he thought he was outed too early to obtain any conclusive results from the experiment (and Rowling also said she wished she could have kept writing under the pseudonym, although she didn't say whether she had any similar "experiment" in mind). The random-sample-voting algorithm would provide instant feedback, not just to King and Rowling, but to any other writer who wanted to see how their writings would stack up against others in the field, from unpublished amateurs to worldwide bestselling authors.
My prediction, if such an experiment is ever conducted: King's and Rowling's writings would be rated very good, but so would many other writers' stories, including struggling writers who have never been published. Or as economist Daniel Kahneman put it: "success = talent + luck; great success = a little more talent + a lot of luck." (That took a certain amount of modesty on his part too, having achieved "great success" himself in the form of a Nobel Prize.) If J.K. Rowling or Stephen King ever launched such an experiment, the biggest favor they'd be doing for the world would not be to boost the egos of a few struggling writers, but to call more attention to the role that luck plays the world.
It's not as if their own egos would have to be bruised in the process. Donald Trump, the last person in the world that I would have guessed to have uttered these words, actually said that "Everything in life is luck," but it didn't seem to deflate his opinion of himself. You don't have to be a jerk like Trump, but just because some unpublished author's story gets a higher rating than yours, doesn't mean you have to let him come live in your mansion.
Now that a famous person's name has gotten your attention, please use our website and give us money!
I believe JK Rowling wants to sell her work rather than give it away. Answering an intellectual question isn't half as nice as proving it to your peers with wheelbarrows of cash.
Which has more power: the hammer, or the anvil?
no ones going to read all this you huge nerd
"But a better way to answer that question would be to strike a deal with an amateur-fiction-hosting site and use the random-sample-voting algorithm that I've written so much about"
If you've developed a method to distinguish the difference between luck and skill then you should really be talking to some scientists, not trying to get Barry Potter and the Sea of Bees to join your amateur-fiction-hosting-hour.
J.K. Rowling was doing nothing of the sort.
http://www.bbc.co.uk/news/entertainment-arts-23366660
I appreciate the the autors may havea neat Idea they want to push but name dropping like this and creating false image of proceedings suggests they're not the brightest or most observant bunch.
I don't care about voting, blind or not.
Socially, the New York Times is supposed to be representative of American media and literacy.
In reality, the National Enquirer is the most _popular_ 'news' paper in America.
imo, you should decide what you want to measure first. Not measure (vote) and then say that is the highest quality because quality is not what is being measured when you count popularity.
This summary is longer than her book.
tl;dr;dc
I went looking for the older reviews on Amazon, and while they were generally positive, it looks like this 'experiment' lasted a month and a half or so. If it were a year or more, that would seem like an experiment. A month and a half seems like a publicity stunt. The reviews there seem to indicate that everybody knew it was a pseudonym and suspected the author had significant publisher backing, so it's hard to even call it a fair experiment in the first place.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
Speaking as someone who shares a sofa of an evening with a publisher, I can vouch that almost every manuscript submitted to any publisher will be dreadful in almost every single department. It starts with embarrassingly poor spelling and punctuation, and moves on through dreadful grammar, choice of paragraph size, layout, and on to issues with plot, characterisation, and general readability.
The average quality hovers somewhere between execrable and toe-curlingly awful, and they get dismissed after a glance through the first page. Sometimes the covering letter is enough and the manuscript can be spared its cursory eyeballing, because if you cannot correctly spell and punctuate in your covering letter then you're wasting everyone's time -- thankfully only 15 seconds of it was the publishers, and two years of it was yours.
Based on that alone, a very useful algorithm with a high degree of accuracy in judging a manuscript's quality is to just throw it straight in the bin -- you'd only be wrong one time in a couple of hundred, which is a pretty good average.
She did not purposefully release that this was her pseudonym, so kind of a bad example. There have been numerous news posts today about how she's mad at the PR firm that leaked the info... http://www.npr.org/blogs/thetwo-way/2013/07/19/203548818/book-news-j-k-rowling-very-angry-that-law-firm-leaked-her-name
"A truly wise man realizes he knows nothing."
after all the first harry potters weren't amateur published, but as pushed books with a publisher with some faith into it to publish in that format.
besides, the quality wasn't the question here, it was if people would buy the books. that's not what was the question in if they had luck in getting the status of a famed author or not - the sales act as the indicator.
a single book though hardly serves as any indicator.. so kings bachman experiment serves more as a guide.
payments.. just forget it and go write a review on amazon. more than that amazon ALREADY has a review pushing system similar to this, doesn't it?
world was created 5 seconds before this post as it is.
My own theory about the stellar popularity of particular authors is that many people just want to share the experience of reading, just as they want to share the experience of watching American Idol or Strictly Come Dancing.
A book that everyone is reading is also a social event, and there's maybe social pressures to read it also.
It's really the only explanation I can think of for the popularity of a book about a teenage wizard in the over-20's demographics.
Why are you mentioning this here? As you say, you've already posted this idea many times to /., and gotten a good amount of feedback. There's nothing new in this latest incarnation that wasn't in the last one. You're not even asking us anything this time; you're just kind of talking at us. Yes, slashdot has insightful people you can bounce an idea off of, but eventually it stops being "bouncing an idea" and starts being "bashing your head repeatedly into a wall".
So instead of rehashing this idea on /. time and time again, why not take it to someone who could actually implement it?
Everything is better with chainsaws.
I don't want a number given out by 20 random people to give a go or no go on what I get to see. I want to have person X and Y, with whom I share an interest for this and that, and despite X hating something I crave, to give out recommandations, saying they liked/hated this because it included A and B.
TFS rambling is flawed. It proposes to hide the author name *and* publish free (or even payed reading !) ebook, and claims to test only the fame factor. Of course the reader will be less averse to spend money in reading unknown author #1138 if the book is free.
This already exists. It's called Lulu Helix review.
http://www.lulu.com/services/helix-review
"The Helix Review provides a detailed analysis of your submitted manuscript by comparing it with all published works within The Book Genome Project as well as making specific comparisons to titles in your chosen genre. "
This is silly. If they had luck back then, it would be wrong to assume they haven't improved their skills. Remember the Matthew effect. By getting that luck, they are now professional writers who spend all day honing their skills while the other mundane details are taken care by others. Those who didn't have luck have to find other way to make money and on top of that manage everything else in life which leaves very little time for developing their writing skills.
tl;dr
There's no way to tell if it's luck or not, because there are too many factors which cannot be controlled or accounted for.
Also, J.K. has already stated that she did it because she wanted to write a book without the massive pressure from fans and industry that she was subjected to when trying to complete the Potter series.
In summary, this submission is just an advertisement for the Voting Algorithm.
It would seem that statistical success != commercial success. Too many dissimilar variables on either side to have parity. Sounds like she went about this the right way.
"J.K. Rowling recently revealed that she was the author of a book she had published under a pseudonym, which spiked in sales after she was outed as the true author."?
No, she didn't.
Some blabbermouth at her solicitors (american, look up lawyers), did.
She's pretty peeved about it, in spite of the extra sales.
when your lawyer can't keep his fucking mouth shut?
"In a statement, it said one of its partners, Chris Gossage, had told his wife's best friend, Judith Callegari, that Robert Galbraith was really Rowling."
But Luck makes you a superstar. J.K.R. is certainly talented, no doubt (no book I wrote would get even 1,500 people to buy it!), but there are lots of talented people. Too many for the average person to keep track of, in fact too many for experts in most cases. There are thousands, maybe millions of great works of art I will never have time to appreciate.
Also, there is a certain amount of luck in creating a classic. The number of artists who can create more than one true classic is extraordinarily rare! Even the best have many mediocre works besides their great ones, if you want to produce more than once classic it requires an insane amount of dedication and time devoted to it along with accepting that most of your work will not really be great.
But one still might need luck to get past the initial 20 reviewers.
She was outed, and she's pissed about it.
And if we used your voting method on this article, it probably wouldn't have hit the front page.
Perhaps she was doing an experiment to see how much luck had played a role in propelling her to worldwide success, and whether she could recreate anything close to that success when starting from scratch.
Perhaps she wanted to release a novel under a different name so that it could be more serious and not be tied to her image of Harry Potter. Perhaps she was just dicking around. Perhaps you could have presented this differently, so you didn't frame the whole thing as if your premise were true so that you could jack it all about your sample-voting blah blah blah geezus get on the front page using a story about an author who probably didn't do what she did for the reasons you suggest.
From what I've heard she did it under the pseudo name so people would buy it on its merit as a good book or not. She didn't want people to buy it because it was written by "JK Rowling zOMG MUST BUY!!!"
Malcolm Gladwell - Spaghetti Sauce (17:34).
It's an awesome TED talk about the problem with asking people what they want. The key point is that you shouldn't try to find out which "thing" want, but that you should ask which "things" people want, and that by offering N choices, you raise the average consumer's satisfaction rating from ~60 to ~80.
He has a longer talk about the Malcolm Gladwell - The Kenna Problem: Why asking people what they like is sometimes a bad idea (48:33). The longer talk spends a lot of time talking about different industries, such as the music industry, the soft drink industry, etc.
I can tell JK Rowling if her success was based on luck or talent after having read her books. Seriously, it's pretty immediately evident.
Oh and I didn't read the last 2? 3?.
That might be a hint at the answer.
(Not to mention she already has the result. If she sold 1500 total copies before she outed herself, then she knows.)
-Styopa
An amateur fiction site will probably have a readership nothing like a broader audience. For one thing, you have to assume that most of the readers are themselves amateur writers, so you have to assume upvotes would skew heavily toward those writing attributes that appeal to what they like to (or would like to) write. The members of a broader audience, however, are generally not interested in being writers themselves and place less value on "writerliness" than they do on simple enjoyment.
How does a person know a book is worth reading until they read it? being an author is a classic "chicken and egg" situation, and an IQ at least in doble figures would allow a person to comprehend this concept.
When Rowling wrote her detective novel under a pseudonym, she was following a classic pattern of behaviour for a famous author crossing a genre barrier (or, in other cases, where an author is just too prolific to release everything under one name). Now this book got very good reviews, but good reviews do not necessarily translate into good sales, for the reason I mentioned at the top.
But, on its merits, the book got good reviews (actually, this itself is a publishing fiction- but let's ignore the fact that book publishers 'buy' good reviews for the sake of the argument). So, book one gets good reviews but poor sales. Now, the publisher announces to the world that Rowling was the author. A popular author is now associated with a well reviewed book in a new genre for her, just as book two is due to hit the selves.
A CLASSIC well-managed PR campaign. This is what makes the best publishers the 'best'. Now, as for whether Rowling's detective novel has true enduring qualities is a different question altogether. How is real quality measured? Here's a clue for you.
In the Victorian age, Britain had annual awards for 'great' artists. Not one of the artists that won the prize is considered worthy of recognition today. Popular with the masses. Popular with the critics of the time. Neither matter a damn. Much later there will be another judgement, and even this may not be a better judge of true merit.
Fashion and luck play a massive part in art success. Yes, we can recognise the garbage (the current wave of 'modern art' is clearly utter trash, for instance) and yes we can recognise the elements of quality in the better crafted material. But beyond this, suggesting there is a reliable way of identifying higher meta-qualities in art is laughable.
JK Rowling should come over to my house for some sexy-time.
-----
Sorry, I'm only a 1336 h4x0r.
She's pretty peeved about it, in spite of the extra sales.
Despite, not "in spite of". Unless you're saying the sales were feeling slighted.
Perhaps Bennett Haselton should use an anonymous blog instead of Slashdot to pour out all his half-baked novel-length thoughts. That might be a better way to see how much luck plays a role in determining whether anyone cares what he's blathering about.
The editors changed that. My original summary block of text began with: "J.K. Rowling recently revealed that she was the author of a book she had published under a pseudonym, which spiked in sales after she was outed as the true author..."
For some reason the editors changed "confirmed" to "revealed", which I agree is less accurate, since it implies some deliberate choice on her part. I have no idea why they did that.
And then in the first sentence of the actual body of the article, I used "confirmed" again, and that one the editors did not alter.
Sorry, I meant to say that my original summary block of text began with: "J.K. Rowling recently CONFIRMED that she was the author of a book she had published under a pseudonym, which spiked in sales after she was outed as the true author." (emphasis added just now, not in the original). The editors changed "confirmed" to "revealed".
We need another experiment:
We need to determine whether the present luck-and-influence-based system leads to a greater total commercial success for the system than would a voting-based-on-perceived-merit system. Remember the current system exists to make money, not good art or social justice. If an experiment can show that the voting system leads to more total money made, then the opportunity might just be seized upon.
In the past modes of publishing that relied on manufactured product distribution (printed books, LPs & CDs, etc.), upfront costs made it advantageous to the system to have few titles and authors, published in great quantities.
There was a hope that the so-called "long-tail" with many published titles and authors in not-necessarily-huge quantities publishing with low upfront cost (e-books, mp3s, etc.) would become successful. It seems the luck-and-influence status quo has largely prevented creators in the long tail from succeeding.
If we could experimentally demonstrate that when augmented by this voting system, the long tail can make more money for the system as a whole, then maybe Amazon and other big players will implement it as a way to increase their total sales.
Anyone can write anything, and the average e-book author makes under $500 per year. The power is in being a gatekeeper, and controlling marketing and branding. The "JK Rowling" brand sold the book, not its quality.
In short, it's a pretty meaningless system based on a flawed average with unknown, but low, confidence in the scores.
She could just get a job with the NSA, and find out what people are really saying about her book.
Sheesh, evil *and* a jerk. -- Jade
John Katz is back! And who says /. ain't what it it used to be?
Plan My Week for iPhone
It's a publicity stunt. Famous author writes a book under an "anonymous" pseudonym, secures some positive reviews (wonder if the reviewers were "in the know"?), puts up with a few months of slow sales, gets her pseudonym leaked, OMG!! Profit. Even better, she can keep the publicity going by then claiming she's disappointed by the whole leak in the first place and just how dare they.
If you want the "Best" books, just use Amazon's "Best Products Based On Reviews" feature.
It takes into account momentum, average and weighted reviews and they even do a pretty reasonable job of removing most fake reviews ( eventually ).
It's also genre-specific with a much larger sample than the OP uses... This is far more important to me as an author ( "Turing Evolved" - #1 Best Technothriller on Amazon, December 2012 to May 2013 - 4.6 average review ) because despite being #1 in the "best" list, the book only sells about 200 copies a month.
In comparison, there are a lot of best-selling technothrillers that don't even rank in the top 100 best list.
Science fiction books tend to have a very small, specialized market and most publishers won't even touch them now. On the other hand, JK Rowland's books have a very wide appeal and many fans. People don't read a book because it's the *best*. They read it because it has a story they want to read, and if it has wide popularity, sales will increase, because it doesn't require such a narrow.
But if they are looking for a very narrow genre? Then they can look for the "best" as a guide - it's far more useful than using the bestselling lists.
GrpA
Enjoy science fiction? "Turing Evolved" - AI, Mecha, Androids and rail-gun battles. What more could you want?
Haselton, you are a cunt of the highest order.
Having struggled through only the first 2 paragraphs of the first Harry Potter book, my guess is that luck is a powerful and unfortunate force in the universe.