Reviews and rankings…
We’ve all seen them—reviews for books, movies, restaurants, contractors, tourist spots, hotels, and so forth. To summarize the point I’ll be making here, they’re basically worthless. If you generally agree with that, you might not want to read any farther, but what I have to say about their oft-accompanying rankings might surprise you. I’ve considered this problem a lot, based on my previous experiences with statistics and sampling as a scientist. My conclusions are hard to put into words, but I’m going to try to do so. Smart, educated consumers demand better products, thus improving what we produce. That goes for authors as much as it does for Apple.
Does any reasonable person believe that a ranking system is useful? Amazon does, but they just want to accumulate votes to calculate stats on the products they sale, which aren’t mostly books anymore, so they treat books now just like any other product. And how they do all this is highly questionable. To be fair, how most retailers do it is highly questionable.
You can put the usual ranking systems in scholarly terms, A = 5, B = 4, C = 3, D = 2, and F = 1. In that case, 3 is an average performance, while 5 is excellent. While your professor or teacher might use a curve to assign grades to a large class, retailers do not. In fact, the author of a book, or the product manufacturer in general, has many “teachers” who subjectively assign an arbitrary grade based on an incorrect sampling process. Moreover, those teachers, while they might examine the performance of many “students,” pretend to have some absolute knowledge to define their ranking of many students. This is the first fatal flaw, but it only gets worse.
This subjectivity flaw is accompanied by natural biases. While some influences from cultural wars might come into play in these biases, let’s just concentrate on the mathematical biases. Who do you think writes product reviews? In particular, who reviews books? If you think that reviewers represent a cross-section of the buying public, I have a bridge in Brooklyn I’d like to sale to you. You see, reviewers tend to the extremes: they either generally like or dislike the product. The middle rankings are depleted. And, because retailers like Amazon often weight negative reviews more than negative, the statistics are completely skewed and NEVER can become a normal distribution, something scholastic curves are based on and many people wrongly assume is the case when dealing with large numbers.
These biases also affect political polling because ALL product reviews are just polls, and candidates are the products of enterprises that are designed to elect them. Same with issues. All polling is sampling, but the samples are often skewed in the way I’ve described. Take an innocent example: let’s consider ranking how much people like orange juice with one star for extremely dislike and five for extremely like. Now ask volunteers to express their opinion ASAP. Consider the first N who reply. That sample will be biased to the extremes. And don’t think the law of large numbers will save the day. Responses for the extremes dominate because people who are so-so or indifferent about the product or just can’t be bothered to consider it tend not to respond.
In other words, Amazon’s and every other retail ranking system have flaws that render them useless for would-be buyers. You’ve seen this in all aspects of life, in fact, even with professionals (lawyers, doctors, dentists, and so forth) whose rankings are biased down because those who “review” them have an axe to grind. Smart consumers base their buying decisions on other criteria, NOT reviews. Any retailer who claims their reviewing system is unbiased is full of it.
This is why I ignore Amazon’s ranking system when buying books. My “other criteria” are how interesting the book blurb is and whether my “peek inside” shows the author is a competent writer. I also use price (I won’t pay more than $6 for an ebook, for example, and only read print when some friend or relative gives me the book as a gift). Unlike the ranking of professionals, though, Amazon’s book rankings tend to be biased upwards, in spite of their weighing negative reviews more. There are bookbloggers, for example, who will only review a book if they “like it” (I do the same for my reviews of books I read for R&R, but not on Bookpleasures, both processes already somewhat biased by my selection criteria just described, yet I’ll never give a book a ranking if I can avoid it!). Most readers don’t bother to review, period, but when they do, it’s again because they usually like or dislike the book a lot.
Amazon and other retailers would have customers believe that their reviews and ranking systems are fair and accurate and actually mean something. They aren’t…and they don’t. The only thing they might do is set your expectations, which is bad. Don’t fall into the retailers’ traps. Develop your own personal criteria using your own past experiences. After all, you know what you like. Neither Amazon nor any other retailer does.
***
Rembrandt’s Angel (Penmore Press). Mystery, thrills, and romance await the reader of this novel. Scotland Yard Inspector Esther Brookstone is trying to decide whether Interpol Agent Bastiann van Coevorden will become husband #4 as she becomes obsessed with recovering a Rembrandt painting stolen by the Nazis in World War Two. As the case morphs into an international conspiracy, the two become closer as they battle neo-Nazis, a drug cartel, and ISIS. Available in ebook format from Amazon and Smashwords and all the latter’s affiliated retailers (iBooks, B&N, Kobo, etc.), as well as print format at Amazon or your local bookstore (if they don’t have it, ask for it). (To see excerpts from some reviews of this book, see the webpage “Books & Short Stories.” As is the case for most of my books, Amazon doesn’t show ALL the reviews—that’s another failure!)
In libris libertas!
