Slashdot Mirror
Amazon: We Can Ship Items Before Customers Order
An anonymous reader writes "The WSJ is reporting that Amazon has obtained a patent for 'anticipatory shipping,'' and claims it knows its customers so well it can start shipping even before orders are placed. The technique could cut delivery time and discourage consumers from visiting physical stores. In the patent document, Amazon says delays between ordering and receiving purchases 'may dissuade customers from buying items from online merchants.' Of course, Amazon's algorithms might sometimes err, prompting costly returns. To minimize those costs, Amazon said it might consider giving customers discounts, or convert the unwanted delivery into a gift. 'Delivering the package to the given customer as a promotional gift may be used to build goodwill,' the patent said. Considering the problems that can arise when shipping something a customer did not order anticipatory shipping has the potential to backfire faster than an Amazon drone can deliver."
Nope they won't charge you, the article says the items are held at a local level waiting for a matching order to show up before it knows where/who to deliver to so the billing process isn't predictive, just the inventory/distribution/shipping is.
Dood! Like wouldn't it be like awesome if we could like invent this like really cool time machine and like go to the future to see like what people have bought, and then like go back to when these people like sat down at their computers to think about buying something, and like it shows up right then! They'll like think its like magic!
Dood! that's frikken awesome! Now where's my goddamn Fritos!
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Nope they won't charge you, the article says the items are held at a local level waiting for a matching order to show up before it knows where/who to deliver to so the billing process isn't predictive, just the inventory/distribution/shipping is.
Yes, to the surprise of nobody, another badly written headline is a terrible summary. All distribution chains do this already -- What Amazon has patented is a particular set of data mining methods in the hope that it will result in a slight increase in efficiency in this process.
Of course, to anyone who's studied caching problems in CSci... this patent would be almost painfully obvious. It's the same thing we've been doing in computers since, erm... the 80286 days. But when you're a large company in America, the rules don't really apply to you.
#fuckbeta #iamslashdot #dicemustdie
So lets put some bulky, heavy things in our shopping carts and on our wish lists and really mess with Bezos for a few weeks.
People say that, but they never cite anything.
39 U.S.C. Section 3009
Or, an explanation in layman's terms by the USPS.
Is The FTC a credible enough source for you?
Q. Am I obligated to return or pay for merchandise I never ordered?
A. No. If you receive merchandise that you didn’t order, you have a legal right to keep it as a free gift.
It's pretty different from a standard caching operation.
It's more like a massively parallel distributed caching operation where the act of caching something removes it from the original data source until it is uncached, and where latency is at least a day or two and cost is very high.
The real innovation is knowing what to cache with enough confidence to act on it...with a granularity of a single customer.
My God can beat up your God. Just kidding...don't take offense. I know there's no God.
Amazon is merely pushing the tendrils of predictive modeling down a level in their supply-chain. No, they're not going to actually deliver something to you before you order it. But experience tells them, through predictive modeling, that someone in your immediate neighborhood is likely to order more boiled peanuts in the next day or so, so they simply box them up, put them on a truck and once that truck gets to your neighborhood, they lie in wait. Sure enough, Bubba Hatfield, your neighborhood transplant from the land of dixie, gets him a hankering for some more boiled peanuts which, for some reason, they never have on the shelves in the local grocery store. He'd really rather buy some off the shelf at a local store, on account of how bad his craving is, but knowing there's some boiled peanuts on the way will help salve his itch a little, so he fires up his browser and finds him some of that bliss in a can. Now, what to his wondering eyes does he see? Under delivery options, there's a new 'IMMEDIATE DELIVERY' option for just $5. What? Are they going to use a rocket to send a can of boiled peanuts all the way from wherever the hell Amazon is all the way out here? He skeptically reads the 'more information' link about this new delivery option. All it says is they guarantee delivery in 30 minutes or less, or his peanuts are free. What the hell? Yeah, an extra $5 for a can of peanuts is ridiculous, but the thought of being able to eat some of those heavenly morsels within just a few minutes is too much. He selects IMMEDIATE DELIVERY and punches the buy button. The friendly Amazon truck, which just happens to have boiled peanuts among its cargo, adds Bubba's address to its current route. In 27 minutes, 30 seconds, an incredulous Mr. Hatfield is gazing, teary-eyed, at a can of purest dixie delight right there in his hands.
It's the same thing we've been doing in computers since, erm... the 80286 days.
Refreshingly, at this time, the novelty will be in the fact that it is NOT on the computer!
...with a granularity of a single customer.
They don't need granularity to a single customer. If Amazon can find 100 people in a city that have a 50% chance of ordering a product, then they can pre-ship 50 to that city's local distribution center. Then when approx. half of them actually place their orders, most of them will get it quickly, even though Amazon didn't know precisely which people would actually order. This will work better with more popular items, where the hits and misses are more likely to even out.
It's pretty different from a standard caching operation.
Okay... and your argument for this is...
It's more like a massively parallel distributed caching operation where the act of caching something removes it from the original data source until it is uncached, and where latency is at least a day or two and cost is very high.
So it's the same predictive logic used for caching, except it takes longer, and it has a queue hung on the side. I don't call that "pretty different" from a structural standpoint. "Pretty different" for me would be the difference between a predictive caching algorithm and, say, TCP/IP flow control algorithms, which also try to be predictive, but have very different constraints.
Either way, this is neither an unusual, innovative, or in any way exceptional application of decades-old algorithms and information processing engineering. It should not be patentable, and that was my point... not quibbling over whether it's "slightly" different or "pretty" different... to qualify for a patent, it must be truly groundbreaking, not merely taking existing formulas and process and adapting it.
#fuckbeta #iamslashdot #dicemustdie
There is still room for novelty in solving a traditional, well-explored CS problem in the physical space, largely because the cost of operations is different. In a computer, quicksort is the accepted way to sort data without foreknowledge of how it is mixed. Sorting railcars using quicksort would be a terrible idea because you can't swap arbitrary cars in constant time (https://www.americanscientist.org/issues/issue.aspx?id=369&y=0&no=&content=true&page=5&css=print). In this case, Amazon may well have developed a novel caching scheme that is efficient in the space of their distribution network, which likely has a different topography than the memory of a 286.
"Because Science" is one step from "Because old book". Try "Because of my experiment testing my falsifiable assertion".