The Amazon order test as an algorithm for evaluating books

If you read a book, how many other related or similar books does it make you order?  (Of wish to order, if you are budget constrained.)  If the number is at least three or four, the book you read is almost certainly very interesting and worthwhile, if not always accurate.

Andrew Roberts’s biography of Napoleon made me want to read an additional biography of Napoleon, because it made his life to me more interesting.  It made Napoleon’s period more interesting too.  I might read a book on cavalry tactics as well, a topic I have never read on before.

Some books pretend to be the final word on a topic, but it is unlikely they succeed.  If you don’t end your read with some additional book orders, maybe you need to ask yourself what exactly went wrong.

At times it is not a book order which is the appropriate follow-up.  Say you read a book on Sri Lanka and you respond by going to Sri Lanka, well that counts too.  Or a biography of Beethoven may lead you to more of his music, rather than to another book on his life.

If I apply the Amazon order test, the best book for me this last year was Michael Hoffman’s Where Have You Been?: Selected Essays.

Hofmann’s book wins additional points for chain effects, namely the books I ordered, as a result of reading Hofmann, in turn made me want to order further books.  But chain effects are tricky.  Following my read of Andrew Roberts, and then a follow-up Napoleon biography, will I read yet another life of Napoleon?  That may depend on how good the follow-up is, and Roberts should not be held liable for that.  Or should he?  What should you think of a book which leads you to so-so follow-ups rather than to excellent follow-ups?  A blog post which does the same?

What percentage of the value of a book is derived from the quality of the follow-ups it induces?  Under plausible rates of discounting, for serial readers this could easily by eighty or ninety percent or more.  (Could it be that actual book reviews are not consequentialist? Horrors.)  How about a book review outlet which refuses to consider the books under consideration, but rather considers and evaluates what they will induce you to read next?

I would subscribe.


I love history and I’m deeply interested in books covering the Atlantic slave trade because I’m from Angola and I struggle to understand why so little is known/said about the contributions of Angolan/Central Africans to the formation of the Creole culture of the Americas we know today. When I first read a book on this theme by John Thornton I immediately ordered a second one from the same author (and his wife Linda Heywood) and I can count about 5 books I ordered after the first one.

Not sure of the answer to TC's question, but I have --as many chess fans do--way more chess books in print and e-book format than I will ever have time to read.

would this be a better test to evaluate book reviews?

Roberts' The History of English Speaking Peoples is good too.

And this take on the battle of Waterloo by Alessandro Barbero has lots of interesting details on the battle:

Of the top of my head there is the almost completely psychological test between a cavalry charge and the infantry in a box formation. Hold the box and the cavalry's horses will balk and likely be spent - their riders ripe for slaughter. Break from the box and run - and you die.

Another bit refers to a French officer about to be executed in the field when he makes a sign of the Freemasons, which a British officer recognizes and leads to his rescue.

Yes, truly amazing battle - Very much a coalition battle with over 3/4 of Wellingtons army comprised of other nations soldiers; Wellington spends all day riding up & down the front lines, without a single injury! Napoleon disappears for a portion of the battle; no-one recognizes the tactical significance of Le Hay Sainte, until almost the last moment; the French Cavalry charge & take the English guns but leave the nails behind (smash the sponge staves, boys..); Blucher's Prussians arrive on the field late and are initially mistaken by Napoleon as being friendly forces. And of course, the battle at which the Rothschild's were able to assure their fortune.

Well most battles have around 10% casualty rates, even the Charge of the Light Brigade, a disaster for the British, had a mere 33% casualty rate, so I'm not too surprised Wellington did not suffer a scratch (Churchill in Afghanistan and Africa also was very brave/ reckless and did not suffer any injuries). Compare to the transportation of Turkish troops during WWI, which was done in such an unsanitary fashion and in the dead of winter that up to a third of them died.

As for Rothschild making a fortune, I heard it's more accurate to say they *kept* their fortune, since they were assuming a anti-Napoleon defeat, so advance warning of a victory allowed them to cut short their losses.

In the spirit of Tyler's post, I recommend John Keegan's The Face of War, which discusses three historical battles (Agincourt, Waterloo, the Somme) that took place in the same area.

"I might read a book on cavalry tactics": oh that British cavalry generals ever had.

I'm reading the Napoleon bio as well and it's made me want now read War and Peace next.

Yes! This is also my favorite thing about any great book - when it leads me to more great books. Then the pleasure is multiplied.

Usually I'm led to more books by the same author. Unfortunately I usually run up against a Peter Principle - I keep reading until I find a clunker, and then I'm like, "Well foo, he's not so great after all."

I loved the Roberts book - learned a lot - and came to a much better understanding of the time and the man...

Looking for a good Wellington book - and Alexander - if they exist - any recos?

Also known as, the "comfort zone" approach to book selection.

If the number is at least three or four, the book you read is almost certainly very interesting and worthwhile, if not always accurate.

Fascinating geeky presupposition that the value of books is their informative value.

Seems geekier to suppose books are for something other than info content.

The Napoleonic era appear to be less interesting than it really is because Napoleon dominates it.

Napoleon was a pretty interesting guy himself but there is definitely a crowding out effect.


Before I read the Roberts book I might have agreed with you...

After reading it I would urge you do the same...

I found both Napoleon and the times very fascinating - and much different than my limited learning about Europe and the period following the French Revolution.

If I apply the Amazon order test, the best book for me this last year was by far Jules Evans's Philosophy For Life. It introduced me to the ancient idea of philosophy as a practice ("spiritual exercises") which I found to be very powerful and which in turn made me want to read a lot of ancient philosophers (Aurelius, Plato, Seneca, Epictetus, Rufus, etc.), a couple modern philosophers (Nietzche, Schopenhauer), and even some academic-tainted philosophy (so long as it remained true to the ancients) (e.g., Pierre Hadot, Martha Nussbaum).

Classical Philosophy: A history of philosophy without any gaps, (OUP 2014) by Peter Adamson -- it is really good.

I have rated about 1500 Netflix films on their five-point scale -- over the last ten years.

What they come back with and recommend for me is usually not useful.

I have been very disappointed by the Amazon recommendation system. It seems to work by taking the most unusual book you have ever rated highly and recommending 20 books which are virtually identical to it. So if you rate, say, a physics textbook highly, your next ten recommendations are for other textbooks on the same subject. Or if you like something by Shakespeare or Mark Twain, you get a dozen editions of the same book.

Recommending other books by the same author or on the same subject is both useful and trivial. In a public library, those books would literally be sitting on the shelf next to the book I just read! I figured that little trick on my own, thank you very much.

I have a strong tendency to pick up books that are turned outward on the library shelves, also, although I'm pretty sure the librarians do that randomly.

There are just a lot, lot, lot of good and interesting books in the world now. I don't think it's hard to find one. Serendipity is useful, if you've got the time.

That is an intentional decision Amazon made to make it easier to serve up book recommendations. They make a list of distinctive books bought by people who bought a certain book then combine the lists for all the books you bought to get your recommendations.

I don't think I've ever bought a book on Amazon's "recommended for you". But I buy used books, so maybe they just don't have a useful set for me yet? I will Google books my kids (in particular) like and see what different paths that takes me on, and I can often find comments that include recommendations or find discussions of the book on sites that discuss other books of interest.

I mine comment threads like this one, for example, for recommendations. Newsletters like Prufrock also have been super handy.

It seems to a layperson like myself that the point of anything Amazon is doing there is not going to be to primarily be to dig out books that are useful to a person or to put forth good books, but to sell books. Since the other uses are secondary, it just seems logical that there are probably better ways for an individual to find good recommendations. But it may not be worth a busy person's time to find those ways, so maybe the Amazon route is good enough.

I also suspect a lot of Amazon buyers have a buy/read ratio of 10/1 or more, so the algorithms, if they are any good, should be geared to favor those guys. Amazon doesn't care if you like the book or read the book, just if you buy it and then buy more.

Comments for this post are closed