Washington Post Revises a Year’s Worth of Best-Seller Lists, Assures us the Problem Will Not Recur
It has long been acknowledged that getting on a best-seller list has little value to authors; it won’t help you sell more books, and is really only useful for things like burnishing your professional credentials so you can earn higher speaker fees (this is why paying someone huge sums to get on to a best-seller list is a waste of money for the most part).
Now the Washington Post has shown us that even if you can make it on to a best-seller list, you might get kicked off again later. The Post published a notice on its site on Friday that informed us that the Post’s Amazon-derived best-seller lists had been wrong for over a year:
In February 2018, The Washington Post launched new lists of best-selling books. In an attempt to combine online and in-store sales, some of the new lists incorporated data from both NPD BookScan and Amazon. For the first three months, all the lists were published correctly. But in late May 2018, the software that merges the two data sets began to experience substantial, intermittent errors, rendering the lists inaccurate. Two causes have been identified. One was related to how the software ingested and combined data from the two sources. The other started in late December 2018, when a change in one data source led the software to begin inadvertently removing the top bestseller, causing the books to be listed in the wrong order and ignoring some sales from top-selling titles. These errors were caught in June 2019. All errors have been corrected and the lists have been republished online and labeled with corrections. The Post is now hand-checking every list before publication to ensure accuracy.
While one could have fun at their expense, that would be a bad idea karmically speaking because it magnifies one’s own errors.
Instead one should merely note that The Washington Post
Amazon has discredited its lists for all time. It’s not just that they had an error, but that they missed it for a year.
Who know what other errors have not yet been found?
image by clasesdeperiod via Flickr
Peter Winkler July 28, 2019 um 8:49 pm
"It has long been acknowledged that getting on a best-seller list has little value to authors"
According to whom?
Nate Hoffelder July 29, 2019 um 8:08 am
Pretty much everyone?
Getting on a best-seller list can be a result of selling a lot of books, but it will not cause you to sell a lot of books.
Disgusting Dude July 29, 2019 um 8:48 am
There’s articles on buying the way onto lists; of list compilers gaming them openly, excluding sources and manually boosting preferred titles; the NYT openly admitting to Publishers Weekly their list is mrant to be promotional, not accurate; there’s the dumping of months of pre-orders into first week sales; there’s the horde of one week wonders; and above all the reported decline in sales needed to get on a list.
Now add improper processing of incomplete data.
jmurphy July 28, 2019 um 10:09 pm
This says nothing about the quality of Amazon’s lists, it demonstrates that WaPo’s software was not fit for purpose. There was nothing wrong with Amazon’s lists, the problem was when WaPo’s software combined two lists, it threw away parts of lists, and merged them incorrectly.
Nate Hoffelder July 29, 2019 um 7:50 am
Thanks for pointing out that this is WaPo code; I was thinking Amazon programmed it, but I am probably wrong on that point.
That said, I still think this discredits the lists.
jmurphy July 30, 2019 um 8:56 pm
"This discredits the lists"
The WaPo combined best seller list? Absolutely.
The Amazon list? Again, there is absolutely nothing in the quote you published that indicates a problem with Amazon’s lists. WaPo wanted to combine Bookscan data *and* Amazon data.
WaPo failed miserably. The WaPo lists are discredited, yes. For the sake of argument, assume Amazon’s list is garbage: WaPo combined it with Bookscan incorrectly and made the combined list WORSE. For the sake of argument, let’s pretend Amazon’s list is the gold standard of bestseller lists: WaPo combined it with Bookscan data incorrectly and also made the combined list WORSE.
The point of the article you quoted is that WaPo incorrectly combined an online bestseller list (Amazon) with an offline list (Bookscan) and completely screwed it up. So, yes, "this discredits the lists" published by WaPo. This is a WaPo story, not an Amazon story. From your text: "the Post’s Amazon-derived best-seller lists had been wrong for over a year". Well, not exactly. The WaPo best seller list is not "Amazon-derived". It is synthesized from Bookscan and Amazon. And WaPo (more likely the code monkeys they outsourced to) synthesized it incorrectly.
A Washington Post észrevette, hogy már egy éve rossz az algoritmus, ami a bestseller-listájukat számolja – 444 July 29, 2019 um 3:00 pm
[…] A Washinton Post könyves hírlevele és tartalmai széles körben elismertnek számítanak az Egyesült Államokban, és azon kívül is. (The Digital Reader) […]
Episode 279 – Keywords, Pirates, and Getting Over Yourself | Sell More Books Show August 7, 2019 um 7:03 am
[…] News #4: Cancelled Due To Drama (1) News #4: Cancelled Due To Drama (2) News #3: Sober Advice News #2: Scratch That Off The List News #1: Five Figure […]