I came across this post by Brian Ball and it really made a light bulb glow in my brain. Brian posed a question on what makes a good book review.
I heavily rely on book reviews and recommendations. I wish we had a much better version of Goodreads, I would gladly pay for it.
I am not a big fan of Amazon reviews. I know Goodreads is owned by Amazon, but Amazon reviews are a different beast compared to Goodreads. I have seen people give a 1-star review and write praising comments about it. These are the ones who are confused about the system. There are people complaining about delivery issues or damaged copies or some pesky things and give 1-star reviews. So I don't fully trust the Amazon reviews.
Coming to the Goodreads reviews, I find the following patterns:
People who just collect books in their "Want to Read" queue. This is a medium-quality signal. It just tells me that people heard about the book and they want to read it. It's good but that alone is not enough. It is good because you can discover new or interesting books through this.
People who just give star ratings alone. They are quite consistent and no-errors in understanding how it works. In this area, ratings are a high-quality signal. I trust these reviews based on their previous recommendations and reviews.
People who give one-or-two line reviews and star ratings. This kind of review always has my attention, because people have not alone spent time to read and rate a book, they have also taken pains to write about it. This is a high intent, very high-quality signal. I better pay attention to it. This helps me to weed out bad books and keep good books in my queue and reading list.
People who write more 100 words of review (along with rating). These are a delight. There are a few in my network, whom I blindly trust on this. People like Gwern, Jason Furman, Manu. They have done a lot of hard work in reading as well as writing about it.
I really don't have any qualms over the content of the review. I like people tagging spoiler warnings but that doesn't really matter to me. I am more interested in the above-mentioned signals than the actual content itself.