Comment by bunderbunder
6 hours ago
And the flood really is overwhelming. This weekend my mom was complaining about having trouble finding anything actually good to read on Kindle Unlimited. I mentioned that the relative lack of slop is one of the major reasons I chose Kobo over Kindle. Even before this latest AI boom I was already starting to view less content as a feature, not a bug, because it seems that on subscription services “more” is increasingly just a polite way of saying “more crap.”
Similar feelings about Nebula vs YouTube, although Nebula straight up doesn’t have entire genres, or videos in languages other than English, so it doesn’t really work as a general recommendation.
I really wanted to like Kobo but the no refunds policy really burned me. I bought a book listed as being in English, with an English title and English on the cover page, but the contents were entirely in French and they wouldn't refund it because of the general no refund policy. I just felt ripped off because what I bought, as advertised, just wasn't what I received.
That would be illegal in Australia, even if the policy was prominently displayed up front.
Claiming otherwise is to treat each book as a packet of Pokémon trading cards, where you know you’re getting some cards, but you don’t get to choose which ones.
I don't follow this rule strictly, but for most of my adult life I've limited most of my book reading to books > 10 years old. If it still seems remotely relevant and worth reading ten years later, it is far less likely to be a waste of my time. Now sure I'm a bit less prepared for water cooler conversations, but overall the policy has served me well.
> having trouble finding anything actually good to read on Kindle
because of AI slop is new benefit of sticking to older texts that I hadn't anticipated.
> reading to books > 10 years old
I've found myself applying this rule, sometimes unintentionally, to almost all media I consume. It's now a very rare occasion that there's something new that I want to read, watch, or listen to. It just isn't that good, tbh. My music library is full of music from ~2018 and older, most books I enjoy are even older than that, and I can't remember the last time a movie came out that I was dying to see since about 2016-ish.
It's not like I'm intentionally filtering out for old stuff, my own tastes just seem to prefer it. Not sure if its due to my own age (I'm not that old, mid thirties), or if we've just so over optimized media for revenue extraction that its become too formulaic and boring.
A significant amount of ebook reading now is romance/erotica or fantasy (or combined "romantasy") genres by readers for whom something a decade old won't appeal. An old book could seem socially "problematic" from a 2026 lens (especially for young people for whom that is before their time), or it isn't what one's peers are reading and one wants to connect with a community of other readers online or in school/university, etc.
Obviously if one doesn't read these genres, this is a whole foreign world, but it is increasingly the state of mainstream fiction reading, and AI slop is a problem for them that you may be asked to help avoid if you are the nerdy loved one of such a reader.
> An old book could seem socially "problematic" from a 2026 lens (especially for young people for whom that is before their time),
Or, you know, you've just read the old books already because they came out 10 years ago and that's a lot of time to read.
I doubt it has anything to do with "romantasy" as a genre, anything that has people actually reading books, on a regular basis (as opposed to the people who mean reading as consuming one "notable" novel a year).
In any case, epublishing has made a lot more books available and filtering through them was a difficult task even before AI increased the output dramatically.
I've been saying for a while now there's a large untapped market for actually effective recommendation systems (almost certainly human driven given the demonstrated limitations of computer systems so far), as mentioned it was a problem to find "the good stuff" even among just self-published pre-ai books, now it's way beyond that.
I guess to some degree it's the same basic problem as spam filtering, but considerably more nuanced and difficult.
1 reply →