[tialys]

This is why I hate the 'star' review system. It very often leads to users trying to 'punish' the maker of the product, and creates an incorrect perception that the product itself is flawed. This might be a great book, but someone browsing through titles will only see that it has a 1-star rating and assume it sucks. I really think there'd be a lot of potential in some sort of hybrid review system that could weed out these sorts of garbage reviews.

[alex_c]

Pricing is an important factor in a purchase decision, so shouldn't it be a valid factor in a review?

A 1-star review saying it's completely overpriced, while not perfect, is still better than reviews I see in the app store like "Amazing app, love it! 1/5" or "Pretty cool app, but it crashes on load and I can't use it, 4/5" (and those are the intelligible ones).

[tialys]

I made a comment below about price changes and how it hurts the usefulness of such reviews, but I also wanted to add that what you mentioned is another example of where the system fails. A 1/5 or 4/5 review that contradicts itself with a single sentence 'this rocked/sucked' isn't useful. We should be forcing people to provide a real review, or not allowing them to do so at all.

[nickolai]

Well they review it as 1-star saying in essence that it is poor in terms of value for money, while throwing some flak at the publisher for not selling it cheaper. Im not sure what you could hate there - or why you would want to remove it.

If someone found something is poor value for money I'd rather know it before I buy. That kind of reviews saved me from a few bad purchases already.

I do agree there must be some part of "activist reviews" there too, not that they are worth the trouble of weeding them out.

[tialys]

What happens when the price changes then? Suppose it's on sale in a month for $1.99 and still is loaded up with 1-star reviews that say "SO OVERPRICED!" -- not very useful then is it? Ideally, I'd love to see a completely independent review source that has common 'tags' that are easily understood. Rather than complain about the price, just check a box that says you don't believe it was a good value at the price, so when someone looks for reviews, they see "Customers don't think this product is a good value at $14.99" (Which is relevant, and useful even if the price changes).

[jonnathanson]

I don't have as deep of a problem with the way reviewers use the star system, but I do take issue with how limited and, consequently, limiting it can be as a review metric.

This is especially apparent in certain categories, like Blu-rays, where people will arbitrarily use the star ratings to review either the content of the movie, the tech specs of the film print, the transfer from original stock to Blu-ray, the number of extras available, and so forth. It's gotten to the point where a star-based review of a movie, at a glance, is essentially meaningless. (Conversely, at least Netflix removes some of this problem due to its contextual purity).

The same thing applies, albeit to a different and lesser degree, with books. People use star ratings as referendums on the publisher, the author, the price, the cover design, the time of day, the weather, how many people showed up to their party last weekend, etc. There is no consistency in measurement. That's the nature of UGC, I suppose, for better or for worse.

[larsberg]

I haven't gotten any direct numbers from them, but the authors of technical books that I've talked to have told me that your Amazon sales are proportional to the number of reviews, not the rating associated with the product. They claim that just getting more reviews is important and always worth comping review copies to people you know will write a review, even if they are not all 4/5-star reviews.

If I had to guess, most people only click on items with high numbers of reviews, read the "top 3 most useful," and then make their purchase decision from those, rather than the long tail of rants from either the easily-bribed or disillusioned masses. These voters may inadvertently be steering people to the Kindle version by inflating the number of reviews associated with it over the hardcover.

[Tichy]

Hm, sounds more likely to me that they got it backwards - if a book is "popular", not as in liked but as in featured in many pop contexts like TV and newspaper reviews, it will also receive more reviews, because more people will have bought it.

[alex_c]

Yes and no... just as a single data point, my behavior tends to be to look for the "popular" technical references and choose from those. If a book has a small number of excellent reviews, I probably won't choose that over a book with a large number of average-to-good reviews.

I'm not arguing this is rational, just something I noticed myself doing - if others are the same, I can see a feedback loop where more reviews lead to more purchases.

[brendino]

Amazon would be much better-served with a review system that enables users to choose a 5-start rating, AND an overall thumbs up/down recommendation. Then, users will be able to see how good a book is based on how people rate it, and what percentage of readers recommends it. Unfortunately, this system would not solve the problem of eliminating garbage reviews.