Really? GoodReads reviews have devolved to complete garbage. The first few pages of reviews are always from the same people trying to be GoodReads famous.

Everyone gives everything 5-star reviews because they get more likes for positive reviews. The user reviews mean nothing. I'd like to have a way to see what people actually think about books. But on GoodReads that's almost impossible to find.

I'm surprised there isn't a rotten tomatoes for books. I've considered starting it myself, but I don't see much money to be made from it. Maybe that's why...

Interesting point about reviewers skewing positive for popularity. My big issue with most recommendation engines is that they're biased towards false positives rather than false negatives. The other way would lead to fewer, more focused recommendations.

In particular, I'd like to see an engine take a reviewer's dislikes into account more. If someone likes (switching to film for an accessible example) Cinderella, The Little Mermaid and Mulan, their opinion on Brave isn't going to carry much information as they're pretty indiscriminate about Disney films. But if they love Mulan, dislike Little Mermaid and hate Cinderella, their opinion on Brave is more likely to line up with my own.

I would think that the challenge is to find good recommenders rather than specific recommendations.

If I can find similar tastes, maybe a little bit different, then that will result in good recommendations.

So I would want to find people who both like similar things but also dislike similar things. It will be interesting to find people my similarity and see which are best at finding books I like.

I follow my friends, and my friends tend to have good taste! This is essentially my strategy. I do have a few friends who have recommended books that I didn't like, so I just don't listen to their recs anymore. Solved problem!

That's one of the problems I have with ML-based recommenders. I can't identify any particular ML algorithm, and therefore I can't say "This one has consistently given me bad recommendations, so I'll ignore it. But I do like this one over here." Whereas with my friends, I know that Lori and Nicole gives good recs and Brad doesn't, so I'll ignore Brad's and read recs from the other two.

The "smile or die" mentality propagated by sites like FB, where you only have likes but no dislikes, is throwing half the information away. I think it stems from a misguided desire to avoid adversity.

Or possibly, it stems from becoming "super reviewers", and then being paid to review works for compensation?

Or the thought that such things, if not happening, may happen? Just like on Amazon and other sites?

In my experience this really depends on genre. Some sub-genres are absolutely awful in this regard, while others are pretty good. In my experience, reviews on a lot of postmodern books are good as long as you mentally filter out reviews which are basically "I didn't want to read a postmodern book" :).

I wish for better reviews and scores too. The animated reviews are simply awful. In some genres virtually everything has near 5 stars, even the most derivative badly written dross. I now ignore any positive reviews and look at the negative reviews instead. I look at those which seem the most level-headed and mention things I care about.

I think they're fine? I didn't say great. In fact, I did say that I don't think ML will ever live up to my friends' recommendations, and that's primarily what I look at on Goodreads. My news feed thing is a list of books my friends want to read, are reading, or have read and rated, and that's more than good enough for me.

As far as the recommendations goes, I mean, I guess you could say I skip past most of them and only get the interesting ones, but at least the algorithm put it in front of me.

Again, my bigger problem with Goodreads is the performance. I'd use it a lot more and interact with my friends a lot more (and thus get better recs from them!) if it was a speedier site.

I remember metacritic used to have a books section, but they eventually dropped it.

My most hated part of Goodreads is that if you type something in search and try to click the middle mouse button to open the resulting, highlighted book in the new window... it just copies the page you're already on. It's a site for discovering books, why do I have to open the landing page 5 times if I want to open 5 books? It's so counterintuitive to me.

It isn't just me!

> It's unfathomable to me how something that has such a monopoly on the domain can have such terrible performance.

You’ve answered your own question there, the keyword being ‘monopoly’.

If there is no competition pushing them to improve performance or site design, it’ll inevitably fall by the wayside until they get around to it or more accurately until somebody can convince a manager/business that they should spend time working on it, which, without using ‘look how much better the competition is’ that’s gonna be a hard sell.

I’m looking at this with broad strokes though, but as a best guess I’d say that’s pretty close.

I don't think there's anything monopolistic about Goodreads .. we just convince ourselves of it. But as we can see here, people would be highly willing to try out some competition.

It's a lot harder than people give it credit it for as well, I imagine. "why don't they get a recommendation engine already" ... that's the secret sauce of companies like Amazon and Netflix who have a financial interest in every successful recommendation. Goodreads doesn't have a profit model in place afaik to support honing the engine to that degree

I'm glad the recommendations work for you, but I agree with the blog post author and echo their experience. I think Goodreads recommendations are awful. All of the other recommendation engines I use regularly (Facebook, Twitter, Amazon shopping, Amazon Prime, Google, targeted web ads, Netflix, Steam) work significantly better than Goodreads does.

god yes, I just google author name or book goodreads to avoid using their search

Discovery engines are pretty easy if you just use clustering unless you allow transitivity (book in two clusters can be used to suggest a book in either cluster)

If it is so easy, why most of them sucks?

Because it's only actually easy of you remove the irremovable thing that makes it hard. This is one of those tropes that honestly really hurts the credibility of ML in the eyes of non-practitioners. Practitioners get really excited about some 75% solution, and talk about how this thing which is bad everywhere is really easy. The average listener doesn't realize how hard that last 25% (why would they? They were just told it's easy!) and end up writing off the whole thing as snake oil.

To be honest, 90% of the code takes 90% of the time and the remaining 10% takes an extra 90% of the time.

That's the point... if you allow transtivity, you get discovery but your recommendations start sucking.

But my recommemdations sucks pretty much everywhere.

Spotify pretty reliably finds me one or two interesting tracks in my discover weekly. That still leaves dozens of tracks I don't particularly care for (although I generally don't actively dislike them) but I still think it's a decent result.

I guess the obvious problem is that for books the equation is different. If you get recommended 20 books and after reading them you only ended up really liking 3 of them, then you'll feel like you've wasted a lot of time reading books you don't care for. Meanwhile if you listen to 20 audio tracks in the worst case scenario you'll have wasted an hour, and you can do that in the background while you do something else.

I've found that Last.fm's recommendations are infinitely better than Spotify's. I feel like Spotify's recommendations have too much influence from their partners and sponsors.

2 interesting tracks out of a 100+, that's not a great result. I'm sure if you listen to 100 random tracks, you will like a couple.

If all recommendation engines can do is essentially try to find the right cohort for you (which is how they all work because they can just try to go "people who liked A also tended to like B"), then you'll never like them if you have narrow tastes. Recommendation engines are blunt instruments, after all. Just think of how fundamentally unreliable "people who liked A tended to like B, so you might like B too" actually is for such a personal topic.

You'd have to wait until recommendation engines can read your mind and run your simulation against books.