From the Editor: Shall We Study Amazon's Pricing Together?

Price Gun

Is it possible to figure out how we're being profiled online? By Doc Searls

This past July, I spent a quality week getting rained out in a series of brainstorms by alpha data geeks at the Pacific Northwest BI & Analytics Summit in Rogue River, Oregon. Among the many things I failed to understand fully there was how much, or how well, we could know about how the commercial sites and services of the online world deal with us, based on what they gather about us, on the fly or over time, as we interact with them.

The short answer was "not much". But none of the experts I talked to said "Don't bother trying." On the contrary, the consensus was that the sums of data gathered by most companies are (in the words of one expert) "spaghetti balls" that are hard, if not possible, to unravel completely. More to my mission in life and work, they said it wouldn't hurt to have humans take some interest in the subject.

In fact, that was pretty much why I was invited there, as a Special Guest. My topic was "When customers are in full command of what companies do with their data—and data about them". As it says at that link, "The end of this story...is a new beginning for business, in a world where customers are fully in charge of their lives in the marketplace—both online and off: a world that was implicit in both the peer-to-peer design of the Internet and the nature of public markets in the pre-industrial world."

Obviously, this hasn't happened yet.

This became even more obvious during a break when I drove to our AirBnB nearby. By chance, my rental car radio was tuned to a program called From Scurvy to Surgery: The History Of Randomized Trials. It was an Innovation Hub interview with Andrew Leigh, Ph.D. (@ALeighMP), economist and member of the Australian Parliament, discussing his new book, Randomistas: How Radical Researchers Are Changing Our World (Yale University Press, 2018). At one point, Leigh reported that "One expert says, 'Every pixel on Amazon's home page has had to justify its existence through a randomized trial.'"

I thought, Wow. How much of my own experience of Amazon has been as a randomized test subject? And can I possibly be in anything even remotely close to full charge of my own life inside Amazon's vast silo?

After I got back to the meeting, I looked up Dr. Leigh's book on Amazon. (Here's the link.) I was kind of surprised to find the Kindle price was $26.12, while the hardcover price was the publisher's listed one: $27.50. Shouldn't the Kindle be a lot less? Then I found myself wondering: Are these prices part of a randomized trial?

So I conducted a quick trial of my own on two different browsers. One was Chrome, the other Brave. I wasn't logged in to Amazon on either, but on Brave, I did my best to mask where I was coming from by searching in a private tab over Tor. (Far as I know, Brave is the only browser to offer Tor on a start page.) I took screenshots of the results and posted them on Flickr, here. So you don't have to fire up a browser of your own, here's a table of the same results:

Book Kindle Hardcover Paperback
Randomistas on Chrome $26.12 $27.50
Randomistas on Brave $20.02 $27.50
The Intention Economy on Chrome $14.57 $14.50
The Intention Economy on Brave $19.65 $26.21
The Cluetrain Manifesto (10th Anniversary Edition paperback) on Chrome $11.99 $9.97 $13.11
The Cluetrain Manifesto (10th Anniversary Edition paperback) on Brave $9.34 $9.97 $13.11
Biblio Tech on Chrome $17.99 $13.93
Biblio Tech on Brave $14.04 $13.93
Data and Goliath (hardcover) on Chrome $21.30
Data and Goliath (hardcover) on Brave $21.26
The Undoing Project on Chrome $10.83 $9.46–$11.40 $13.75–$14.48
The Undoing Project on Brave $9.90 $9.46–$11.40 $13.75–$14.48

That was on July 27, 2018, in Rogue River.

Here's August 3, 2018, in New York City:

Book Kindle Hardcover Paperback
Randomistas on Chrome $26.12 $27.50
Randomistas on Brave $32.66* $27.50
The Intention Economy on Chrome $14.57 $14.50
The Intention Economy on Brave $18.21 $14.50
The Cluetrain Manifesto (10th Anniversary Edition paperback) on Chrome $11.99 $10.38 $9.89–$12.93***
The Cluetrain Manifesto (10th Anniversary Edition paperback) on Brave $9.34 $10.38** $13.11
Biblio Tech on Chrome $17.99 $13.82
Biblio Tech on Brave $14.04 $13.82
Data and Goliath (hardcover) on Chrome $22.00****
Data and Goliath (hardcover) on Brave $22.00****
The Undoing Project on Chrome $10.39 $14.06 $14.48
The Undoing Project on Brave $9.34 $14.06 $14.48

* Another price appeared and disappeared before I could read it. I also needed to answer a CAPTCHA to prove I wasn't a robot.

** Earlier it said "From $1.14".

*** Earlier it said "$9.89–$13.11".

**** Earlier it said "$21.26".

Since then, I've been checking Amazon prices often on different browsers, different devices, both logged in and not, and again using Tor on Brave. Prices for lots of stuff (not just books) have been all over the place. And information about products changes too.

For example, today (as I write this, on deadline) is August 19th. On Chrome, Randomistas is now $18.24 on Kindle, $19.20 in hardcover and available in paperback for $32.39. (It wasn't in paperback before. See here.) On Brave through Tor, it's the same for hardcover and paperback but $20.09 on Kindle.

To my amateur mind, Amazon's pricing calls to mind Dave Barry's classic column, "The Unfriendly Skies", published in 2010 and more relevant than ever today. Under "Answers to Common Airline Questions", he begins:

Q. Airline fares are very confusing. How, exactly, does the airline determine the price of my ticket?

A. Many cost factors are involved in flying an airplane from Point A to Point B, including distance, passenger load, whether each pilot will get his own pilot hat or they're going to share, and whether Point B has a runway.

Q. So the airlines use these cost factors to calculate a rational price for my ticket?

A. No. That is determined by Rudy the Fare Chicken, who decides the price of each ticket individually by pecking on a computer keyboard sprinkled with corn. If an airline agent tells you that they're having "computer problems," this means that Rudy is sick, and technicians are trying to activate the backup system, Conrad the Fare Hamster.

I now know, after doing some digging, that what's really going on here is "dynamic pricing", and there is a lot of jive about it on the web. (Here's a search: https://www.google.com/search?q=amazon+dynamic+pricing.) And I get that it's about lots of variables other than personal ones: A/B and randomized testing across populations, competitors' prices (again, viewed through different browsers or whatever Amazon's robots might use to simulate human queries), short- and long-term trends, inventory available now or back-up supply chains, scenarios, choice presentation and so on.

So here are a few serious questions: How might we best research this from our side—the one where humans use browsers and actually buy stuff? Is it possible to figure out how we're being profiled, if at all—and how might we do that? Are there shortcuts to finding the cheapest Amazon price for a given product, among all the different prices it presents at different times and ways on different browsers, to persons logged in or not? Is this whole thing so opaque that we'll never know much more than a damn thing, and we're simply at the mercy of machines probing and manipulating us constantly?

I'm hoping some of you have answers. I also think it would be fun to put together an Amazon Pricing System Research Project using Linux Journal readers. (Or maybe, hmm, Amazon's own Mechanical Turk.)

Possible? If so, write to me and let's see what we can do.

Note: pricing gun image used with this article courtesy of PriceGun.com.

About the Author

Doc Searls is a veteran journalist, author and part-time academic who spent more than two decades elsewhere on the Linux Journal masthead before becoming Editor in Chief when the magazine was reborn in January 2018. His two books are The Cluetrain Manifesto, which he co-wrote for Basic Books in 2000 and updated in 2010, and The Intention Economy: When Customers Take Charge, which he wrote for Harvard Business Review Press in 2012. On the academic front, Doc runs ProjectVRM, hosted at Harvard's Berkman Klein Center for Internet and Society, where he served as a fellow from 2006–2010. He was also a visiting scholar at NYU's graduate school of journalism from 2012–2014, and he has been a fellow at UC Santa Barbara's Center for Information Technology and Society since 2006, studying the internet as a form of infrastructure.

Doc Searls