Camestros Felapton

FELAPTONL

Just some links to Ersatz Culture’s detective work

Ersatz Culture has been doing some deep data-driven detective work on Goodreads sockpuppet accounts. It’s covered in three (so far) Twitter threads and I want to discuss some of the things being revealed but not quite yet. I’m putting links to them here for reference:

[ETA one more thread from EC]

I will note one thing as it is pertinent to both EC’s examples and also the troll examples I discussed earlier this year. To register an account with Goodreads you have to give an email address BUT unlike most websites these days there is no email verification step i.e. you don’t NEED multiple actual email addresses to set up multiple accounts. The system is wide-open for abuse.

It is trivially easy to set up a fake Goodreads account. How easy? https://www.goodreads.com/user/show/111117965-sock-puppet (I will delete Monsieur Sock Puppet later and he won’t be rating anything obviously)

camestrosfelapton

23 responses to “Just some links to Ersatz Culture’s detective work”

  1. I’ll be interested to hear who this was benefiting, but I’m in no way surprised to hear it’s happening. On the GR lists I try to contribute to there are always a few outliers that look like they’ve been brigaded up the charts, although whether by actual fake accounts or just by enthusiastic fans following an author’s lead is hard to tell

    Liked by 1 person

    • Mark, it’s one of those GR lists you’re referring to that originally alerted to me to all of this, per the comments from a year ago on that list.

      I’ve been saving a copy of the HTML of that GR list every week or two for over a year, primarily with the intention of showing the progress of books over the year. A side effect was seeing some of those books suddenly jump in a couple of metrics in early November, which caused me to dig into all this.

      Liked by 2 people

  3. Interesting that they add and rate books since you don’t need to vote in the GR awards or to vote and add to lists. But I suppose that’s the easy part if you’re doing it via bulk import.

    (Monsieur Sock Puppet is very cute in his profile pic.)

    Liked by 1 person

  4. So there’s a book that leapt out at me on the list – which I’ll follow the lead on not naming as there’s no evidence of the author participating in this – which I’d previously noticed appearing a bit further up the GR Hugo-eligible novel lists than I’d really have expected it to be.
    It also made the GR award write-in stage. As GR publish vote totals then I can say it had approx 1700 votes (from two rounds), and wasn’t the lowest vote-getter, so either there was a lot more brigading than EC has spotted, or this was at most a “top up” to existing popularity.

    Liked by 2 people

    • Mark: I pondered writing a bit more about that particular title in my Twitter thread. I have saved stats for all of the GR Choice nominees from when each round’s nominees were announced – hopefully grabbed quick enough that their stats hadn’t been been distorted too much by those announcements. When that book was announced as a write-in nominee, it wasn’t lowest ranked of the 5 write-ins in that category in any of the metrics like #ratings, #reviews, etc. As such, it felt unfair for me to call it out – unlike the other successful write-in from the slate, which is a blatant outlier.

      However, I was similarly surprised by that fact it was on a par with the other write-ins – it struck me as a much lower profile/less talked about title than any of those others in that category.

      In that title’s defence, its metrics on GR are much higher than most of the other titles in the “slate”, some of which have #ratings in the low 3 figures – I haven’t done the analysis, but I strongly suspect the ~80 ratings from dodgy accounts are a large proportion of all the ratings they’ve received.

      I’m sure I’ve only uncovered the tip of the iceberg. I did contemplate at one picking a bunch of random numbers as Goodreads user IDs, and seeing what user page it served for those ideas. If a sample of say 100 arbitrarily chosen user IDs returned a non-trivial number of dodgy looking accounts, then that might give an idea how bad the sock puppet problem is.

      Of course, this is likely against some Terms of Service, and cause me to have my account banned. (Assuming all the data analysis I’ve already done hasn’t already triggered that process.) That would be ironic, given that ultimately I’m a muppet who’s done hours of work for free for Jeff Bezos, because he’s too tight to fork out the cash to give his subsidiary company a proper security/moderation team…

      Liked by 2 people

      • Of course it’s metrics could also be inflated….
        But it is a title I’d heard of independently, and it’s not impossible it’s been chosen as a “cover”.
        Yes, with stuff like this along with GR’s pathetic response to recent trolling, they clearly need to up their security game.

        Liked by 1 person

  7. Given how slapdash GoodReads seems to be about so much stuff, I always wonder why people use it when a perfectly good substitute like LibraryThing exists.

    Liked by 3 people

    • I have separate use cases for Goodreads and LibraryThing: I use the former as a reading tracker and the latter as an ownership tracker. That way I can pretty easily tell the difference between “I read this from the library / via ebook” and “I own a physical copy”.

      WorldsWithoutEnd is actually closer to my ideal Goodreads use case but I do read things other than SF/F now and then….

      Liked by 2 people

Blog at WordPress.com.