Back to the Amazon future (sign me up)

Amazon.comAMZN +0.43% knows you so well it wants to ship your next package before you order it.

The Seattle retailer in December gained a patent for what it calls “anticipatory shipping,” a method to start delivering packages even before customers click “buy.”

There is more here.  And here are our previous posts on Amazon.

The pointer is from @MattYglesias.


The one-click patent was BS, but zero-click is true innovation.

Sure sign of a market top when Yglesias turns himself into a stock picker.

More innovation that will yield more convenience and more poverty. Come on, everybody, let's race to the bottom!

I eagerly await anticipatory billing.

Sign me up as well. This will work perfectly for ongoing manga series ... if Amazon can have those arive at my doorstep before I know they're even available, all the better.

You are already signed up.

From two years ago:

I think they actually made that video tomorrow to explain the patent but were able to upload it to youtube two years ago.

Amazon's P/E is 1447.86 as I type this.

Isn't that just an artifact of deducting off reinvested profits from earnings?

Anyone who did one of those "get 10 CDs for $0.99" deals in the 90s knows that deliverying products before the customer has bought them is not new.

Fun idea, but how is this patentable? Most of all, it points to the sorry state of our intellectual property law.

I am not a patent lawyer, but the only part of the claims that is not violated by a milkman or a Victorian London butcher's boy is the phrase "performing, by one or more computing devices."

Indeed how is this any different than having lots of little warehouses? What a stupid patent system we have

You anti-IP guys lack imagination if you don't see how small differences are patentable. For example, in software, they have various sorting algorithms that do the same thing (sort a list alphabetically for example) but perform radically different, based on subtle (and hard to even describe) changes. An example is bubble sort vs quicksort vs mergesort, see: (BTW Tony Hoare quicksort, but for every Tony Hoare who's a Good Samaritan who dedicates his IP to humanity for free, there's another who will keep the invention a trade secret: the recipe for Roman concrete was lost and reinvented in the Middle Ages)

Further the Amazon patent issued is 2 years old, as its successor issued already:

Publication number US8086506 B2
Publication type Grant
Application number US 12/542,346
Publication date Dec 27, 2011
Filing date Aug 17, 2009
Priority date Jul 20, 2001

I meant to say the Amazon predecessor issued already (the divisional) and you note the invention was filed in July 2001--a long time ago--so the US Patent Office has had a long time to critique it, no doubt anticipating public outcry, so probably the Amazon patent is sound. Also Hoare invented the quicksort algorithm and gave it away rather than, as AT&T has done with their traveling salesman algorithm, patent it, and rather than, as is common with software, keep the algorithm somewhat a trade secret by obfuscating the code and keeping the code closed source.

It's funny you mention algorithms, since they aren't patentable at all. Just nonsense like this Amazon patent.

Yes you can patent an algorithm if you wrap it in an interface that is hardware, like a PC. You just can't patent the raw formula in and of itself.

@Matthew, that is simply not true. A high profile example was the patent of LZW compression, used in .gif images.


Promising as a development but sucks that it is possible to get a patent for such stuff.

This makes possible an entirely new type of trolling. Competitors can launch massive DDoS attacks setting off false positives in Amazon's anticipatory shipping algorithm to intentionally cause Amazon losses.

Are all these consumer-information driven innovations bad signs for the future of free entry and perfect competition?

If you read the article, the real story is that the patent office appears to have issued another truly bogus patent that could disrupt other companies from doing the obvious. This concept seems to be nothing new, it is merely a specific obvious instantiation of the sort of "just in time" logistics methods that have been evolving for decades. Retail physical storefronts don't make you wait while they order something from the factory, the distribution chain relies on past demand to determine levels of stock to place in warehouses at hubs throughout the country(/world) and the level of stock for individual stores.

The given interpretation of the patent is doubtful and, most likely, won't hold in courts. The logistics system covered by the patent doesn't anticipate anything about a specific "you." Just like Walmart does it day in and day out, the proposed logistics system ships items to a geographical area in anticipation of consumer demand, e.g. based on earlier patterns of Christmas shopping. The only difference is that the shipped item can be re-routed to a specific customer address when a customer order is received by an e-commerce computer system.

In short, it's not a deep personalization technology as your post implies.

Amazon really doesn't need to patent this. They're the only ones who CAN do it.

A knocking at the door.

"Who's there?" I asked.
"Amazon drone," came a polite but firm bass voice.

I opened the door to find box on the step as the hexacopter retracted its delivery arm and spun up its rotors. An invoice icon appeared on my Amazon eyeglasses. What wonders had I been brought today and how much would they cost?

But then, I thought, which of the two of us was truly an Amazon drone?

No, no, no. You don't think, you just click "pay now".

Beautiful. Whenever you publish your volume of poetry, let me know. I'll buy it on Amazon for my Kindle.

It was sent to you already. Don't you check your in-box?

Curious. I'm sure Amazon is aware of 39 U.S. Code, section 3009:
"the mailing of unordered merchandise ... constitutes an unfair method of competition and an unfair trade practice... Any merchandise mailed in violation of subsection (a) of this section ... may be treated as a gift by the recipient..."

Also should add: "For the purposes of this section, 'unordered merchandise' means merchandise mailed without the prior expressed request or consent of the recipient."

Perhaps if you opt in to this service you could be deemed to have expressly requested whatever they send you? Can you expressly consent to a delivery if you don't know when it's coming, what it is, or how much it'll cost?

It doesn't seem like they are delivering it all the way to you. Right now I guess Amazon is taking orders and loading orders onto UPS/FedEx/USPS/etc. trucks (or freight trucks to those shippers) at the warehouses. I imagine they position items in the warehouses according to predicted local/seasonal demand - not so many gun accessories and camo outfits to the warehouses serving NYC, etc. They propose positioning items closer to the customer without waiting for the orders first - have items on deck at FedEx hubs, on UPS trucks, I would imagine at Amazon Locker locations, & even large apartments so they are closer when Amazon does receive an order. So nothing is "delivered" in the sense of Amazon relinquishing possession of an item to you, until an order is received, and then it's delivered, even if that delivery just means your apartment complex's management office or Amazon Locker is told to release pre-positioned box #FA73 to you.

They propose in some cases giving away extraneous pre-positioned items as more economical than returning them.

Think of Tyler Cowen's next book. Right now, a few days before release date they will start moving shipments to NYC for the pre-orders (I think they usually already get those in your hands on the release date). They propose moving shipments in addition for the estimated remaining NYC pre-orders and the estimated first-day orders. Then you would have a good chance of being able to pick up the new book from one of your regular Amazon lockers at midnight even if you didn't think to pre-order it.

Amazon really is building a different kind of Internet of Things, the ability to move Things that fit in a corrugated box with closer to the ease that we move network packets.

Amazon probably knows what my comment would be,

but for spite I will not say it.

And yet you did.

Anecdotal, to be sure, but I find that despite the huge amount of data Amazon has on my past purchases, the 'Recommended' list is of no interest to me. It also seems to only take into account the last three or four purchases I've made, as well as putting over-emphasis on 'one-off' purchases.

As an example, I've just bought a coffee machine; putting a dozen different models of coffee machine in my 'recommended' list isn't going to entice me to purchase a second.

Overall, I find the ability of organisations (companies, governments, etc) to actually use big data is highly exaggerated.

My own anecdotal Amazon experience is the same, especially as regards over-emphasis on one-offs.

@ Bill and Jimmy - Maybe they are being meta on this, people don't like the idea that corporations know too much about them. So Amazon is showing some bogus suggestions to re-assure people that their algorithms are not so clever. Meanwhile they slip in a suggestion or two that actually is a good one, slipping under your conscious. Like the people that say they are not affected by advertisements, who are wearing Nike sneakers, levi jeans and an Armani t-shirt instead of the same ensemble from Wal-mart..

Actually, for some strange reason, I really like the idea of opening a package that I don't know what's in it, even if I had to pay for it after all. Maybe childhood conditioning with Christmas and birthday presents? Actually, in some sense, you earn your birthday and christmas presents, as part of the family give and take, so maybe it is not too different afterall.

Another thought, I am actually not very good at predicting what will delight me, although I quite good at predicting (and satisfying) my needs. So I could put these purchase in a different mental category to say, coffee makers or a new iron. I could easily see how I could have a monthly budget of $100, for purchases under this arrangement for basically presents to myself, maybe they can monitor my smart phone as I open the package to see my heart rate and eye dilation for an automatic measure of my delight to update their algorithms.

If I was Amazon I would make the price of the goods sold under this arrangement at a discount, this way I have some justification to myself (and my wife) to go for it.

1. If they sent me something I did not ask for, then there is transaction costs (time) to return it, so it is not a truly costless transaction.

2. If they know so much about me or my wife, I would begin to worry if they sent me a box of diapers.

#2 is Target, not Amazon: []

Also, from that article, Target apparently also employs ChrisA's strategy of pretending to not know you *too* well to avoid creeping you out.

If only Target was as good at predicting hacker behavior as it is at predicting customer behavior...

I haven't looked at their recommendations in a while, but the ones I'm seeing right now are better than they used to be. A lot of them are complements to stuff I've bought or looked at recently.

Moreover by this point Amazon has surely gotten pretty good at identifying search patterns that indicate a high likelihood of ultimately buying something, probably with parameters customized to the individual shopper. How many tabs you open and how long, how much clicking around you do in the reviews, trial carts. I've bought hundreds of items from them over the years, so they've got the raw data for a pretty good model of me.

It's funny: I enabled "Google Now" recently, which is in the same game; serving up ads along with purportedly-useful reminders. Their edge is that they can read my e-mail and vacuum up data about my online purchases from all online sources. So Google can presumably reproduce much of Amazon's data, and more...

+1 for me. So true about the one-offs - even if you select "do not use for recommendations." From what I've heard, Netflix doesn't do a great job either despite the hype.

I've been buying everything on Amazon for 10+ years and that's possibly part of the problem. As I cycle through interests, hobbies and life events my purchases change. But Amazon doesn't really put it together that I've moved from teenager to college student to young adult with spare time and disposable income to parent.

The suggestions are basically books by authors whose books I've purchased in the past. That's the best that one of the four greatest tech companies on the planet can do, with 10 years of purchases, browsing, "likes" and reviews?

Same in Germany. Recommendations based on stuff I bought as a present get on my nerves particularly.

It also seems to only take into account the last three or four purchases I’ve made, as well as putting over-emphasis on ‘one-off’ purchases.

Yep, Amazon is pretty bad with that. I recently looked at Amazon's page for the upcoming McArdle book to see Tyler's blurb there, and they want to sell me several different books with that title now (and I didn't even buy the one I looked at). Also, my wife just bought my daughter a My Little Pony comic book, so it's decided I'm a brony.

Netflix is a heck of a lot better, or at least is when it suggests discs instead of streaming titles. Why doesn't Amazon have individual profiles inside family accounts, anyway?

I'd be pleased if they also paid tax as good citizens recognising the social infractructure that their business model relies on.

Most of the commenters haven't even read the abstract of the patent I see. Not a surprise.

What Amazon is actually patenting is that, through data analysis, they can figure out that X number of people are going to order something in a given metro area, and instead of sending everything from a big warehouse to the customer using regular shipping, they can send them to a distribution center nearby, which will then mail the items to the people that actually ordered them, thus saving in shipping, as it's far cheaper to send 10000 items in a truck and then use standard mail for the individual deliveries on much shorter routes.

Companies have been doing something like this in a larger scale forever: Nobody that does a significant amount of shipping has a single distribution center in the US. The heavier the merchandise, the more you gain by grouping your shipments, going as far as to just send containers.

The abstract doesn't make it very clear what is actually novel about this though. Maybe they claim that individual data analysis on order history is radically different from what everyone else is doing in the industry? At first glance, it doesn't seem novel to me.

I'm actually excited to read this one. I expect a lot from the previous books that I read.

Comments for this post are closed