The 800 Pound Gorilla Amazon Tried to Kill is Google's Trojan Horse
blog.jasonshah.org
blog.jasonshah.org
The author compares buying a shoe rack on Google Shopping Express to buying on Amazon. He is impressed that a plywood laminated shoe rack sourced by Target is only $12 while a bamboo shoe rack on Amazon is $20, and makes much of imagining, zomg-like, 50% savings on his Amazon purchases over the year.
If you look at the same Amazon search results page where the author finds the bamboo rack, there is a serviceable resin rack for $13 (the author is obviously not going high-end with his plywood one). Four results down from his bamboo rack is a natural wood rack for under $16.
The real gorilla horse in the room is that the author goes on to lament $5 shipping for Google, which swamps any savings he was getting by going with inferior product. "Now, Google has to make sure its free delivery thing becomes well adopted the way Amazon Prime was, because the $4.99 delivery was a non-starter for me, and ruined the economics."
Twist ending?
Personally I think fretting over three dollars on a shoerack is a bit silly, since the shoes it's holding will each be at least an order of magnitude more expensive than the rack.
Also relevant: http://wiki.lspace.org/mediawiki/index.php/Sam_Vimes_Theory_...
Self-driving cars will make natural delivery vehicles. If Google partnered with local retailers to provide check-out via wallet and delivery via automated cars, it would disrupt more than just Amazon.
Google has all of the nascent building blocks in place: Google Wallet, Google Shopping Express, Google Shopping, Google Maps with StreetView for routing, automated vehicles, and, of course, eyeballs.
"You have 20 seconds to comply." ;-) http://www.youtube.com/watch?v=Hzlt7IbTp6M
Are you sure about that? Self-Driving cars are currently a very limited use case. They drive, but they haven't shown to be very adept at more anomalous behavior that humans can do, such as following detour signs, following traffic officer signaling at intersections, or finding parking spots in locations where they are unlikely to be towed.
Furthermore, when you want to actually deliver something, you have to do far more than a car is even capable of, such as unlocking doors (for apartment buildings with controlled access), opening the million different types of latches for front yard fence gates, obtaining legitimate signatures from humans, determining access requirements for new locations, finding and successfully navigating staircases, escalators, and elevators, and a dozens of other tasks that are years away and not prioritized in current research.
The other parts of delivery are more of a problem. But remember at one time we thought it was necessary to have humans pump gas, operate elevators, and dispense cash. Those tasks were only partially automated, but sufficiently enough that the rest of the job was dumped onto the customer.
The cost for any private party to send up their own monitor drones is coming down very quickly and will likely be insignificant by the time unmanned delivery vehicles are common. If one unmanned delivery car sends out a distress signal, other unmanned delivery cars in the area could release drones monitored by a local security firm to go see what's happening, follow the bad guys, and tell the cops exactly where they went.
The sort of drones I'm talking about might look like this: http://www.youtube.com/watch?v=YQIMGV5vtd4
(You can already get copters of that sort with at least a 10 minute battery range for a mere few hundred bucks.)
I'm sorry, but anybody building an unmanned delivery system will have thought of that right away and dealt with it any number of ways. For instance, consider this scenario:
You order a pizza using an app on your phone. You get a text message "the car is 5 minutes away from your address - please meet it outside. Your unlock code is 537." You go outside. You get another text message "the car is pulling up to your address now; your unlock code is 537." You see a delivery van pull up and pause in front of your address. There is a well-marked door on the right side of the vehicle. The door is locked, but there is a keypad. You enter the code 537 (or alternately just wave your phone or credit card near the reader, authenticating with RFID or bluetooth. Or insert the card you used into a card reader.) and the door opens, your pizza boxes are revealed - one or more pizzas are dispensed like a stack of money showing up from a slot at an ATM. You take your pizza(s). The door automatically closes and the vehicle drives to the next delivery location.
If you don't show up on time, the car texts you once or twice more, waits for another 5 minutes or so, but eventually gives up and goes off to make the next delivery.
If some hoodlums show up first, they don't know the code and don't have the phone/credit card that was used to make the order, because they are not you and weren't told by you to pick up the pizza from the car. They cannot open the locked door so they cannot "grab and go" without defeating the physical security of the car, which would in all likelihood be stronger than would the physical security of a MANNED delivery vehicle. They could of course mug the recipient to get the pizza, but that's no different from mugging a delivery guy now.
Before they move on, existing pizza delivery drivers delivering to an apartment have to: park the car, walk to the door, get buzzed in, wait for an elevator or climb the stairs, walk down the hall to find the right apartment, knock on the door, wait for someone to come to the door, drop off the pizza, walk back to the elevator/stairs, go back down to the ground floor, go back out to the street, get back in the car. In short, current pizza deliveries today often spend more than 5 minutes at each site just making a normal delivery. So yes, an automated truck that just stops at the curb at street level and expects the customer to pick up their pizza can afford to wait several minutes and still be ahead on time compared to the old system.
> You honestly fucking think they will spend thousands of dollars on a vehicle to equip with all this automated bullshit? The costs would be astronomical
Automated driverless delivery vehicles will ultimately cost less than vehicles that have drivers. You don't have to waste space on the driver area so the vehicle can be smaller and lighter or carry more cargo per unit size. They can skimp on safety equipment too - no need for seat belts and airbags or even an internal dash display. Fewer parts = more reliable. There's also the difficulty of hiring reliable people who can work exactly the shifts you need and the cost of paying them. It might be cheaper to hire people now but that won't always be the case.
So I do expect future companies that need to deliver stuff will want to buy or rent the use of a standard off-the-shelf driverless vehicle, possibly adding standardized accessories like a selective-opening side door. I don't expect they will make one themselves. They'll buy the finished version from some other company that specializes in making them. That way the R&D expense is spread across millions of customers and thousands of firms, not just one.
>This is the dumbest shit I have read on HN in a long fucking time....
You seem very angry. Is there something else upsetting you that you want to talk about?
That's still hugely inefficient, though. The optimal solution is what Amazon does for most things- a giant, central warehouse with stuff that you deliver.
As seamless as the front-ends of eCommerce have become, it's easy to forget how much brute force is still required on the operations side of the business.
Automated cars can make deliveries pretty easily, but they can't move merchandise in quantity between locations. The latter is the bigger barrier to scaling the system.
But nonetheless, a major distribution network needs to be in place before this can happen.. If you only want same day deliveries, you could do with maybe 20 centres in the USA and automated delivery vehicles, if you want 1-2 hour deliveries you have to be much closer. And that's where the battle for omnipresence will be won.
Automated trucks sure as hell can...
It's hard for me to see it unfold this way.
Sheesh. Not that Google's tool doesn't have merit, it might.. but we'll have to wait for someone to write an article worth reading. This is silly.
Uninformative blog post, OP.
Internet shopping has a good historical analog. Mail order catalogs. It took a long time for things to settle down, but eventually everyone wound up with both mailorder and retail outlets.
Even the 800 pound gorilla of mailorder, Sears Roebuck, became a hybrid. It took decades before they opened their first store, but today it is mostly known as a chain of department stores.
I am sure that the same will eventually happen to Amazon. However not for a long time.
Digital experiences do theoretically offer that possibility. One example is using virtual augmented reality to see how that shoe rack will look in you're home, how does it combines with your wall and floor colors, etc.
And this "digital better than real" has already started todat. Some high quality furniture retailers or hotels(forgot which) ,use computer generated models for their catalogs, not carpentry and photography.
Also I remember my mother taking out a catalog, putting the picture of the thing she wanted in the place she wanted it, then standing back to better imagine how its colors would fit in. That is a valuable feature that you can't get in a store. (Things do look different in different lighting.)
2. don't get to physically interact - most products don't need physical interaction , just visual one. And there's haptic feedback tech if needed.
3. slow to arrive - online is improving(1 day shipping, maybe better for a fee) and will be good enough for most cases.
And remember:store depend on volumes to pay the rent and keep the lights on. Once online takes enough market share from them, they'll might collapse under their own weight. Some even start to see this happening in financial statements of big retailers.
Mail-order catalogs historically played a very important part in the life of rural Americans. You could even buy a house via mail order.
Even when it became possible to travel to a store easily, mail order catalogs still offered one thing that stores didn't: time.
You could spend hours, even days poring over your catalog, analyzing and obsessing. Every kid assembled his Christmas wish list from the Sears Christmas catalog, even urban kids.
Did you know Microsoft was going to own ecommerce too? Yep, throughout much of the early to late 1990s, it was a common statement. Magazines ran stupid cover stories proclaiming Microsoft would dominate ecommerce.
And a modern interface (re: Google)? What does that even mean. Like how Craiglist has suffered immensely because it doesn't have a supposed modern interface? It's a laughable statement. Just another indication the writer doesn't know what he's talking about. Consumers don't care about what some designer may think is a so-called modern interface. Consumers care about an interface being simple, fast loading, easy to use, etc. Amazon does a great job at that; they've compacted millions of products down to a single search bar and a drop down nav that is very easy to use.
The extrapolation the author makes about saving $2k to $5k per year is silly in nature. Complete nonsense. It ignores whether one product is superior to the other, such that one is more expensive / cheaper. It ignores consumer demand, dictating higher or lower prices. It ignores supply of the product influencing pricing. Personally I liked the bamboo shoe shelf and would rather pay $20 for it than $12 for the other one that I didn't like. The author didn't even bother to actually compare products. The caveat at the bottom nullifies the entire point of the post.
i stopped reading after that.
1) Selection - It’s easy to sell the popular stuff. It’s hard to sell the tail of demand. Aggregating together big box retailers can never compete on selection. People want to order things together, from the same place. Selection is crucial.
2) Supply Chain - It’s hard enough to have an effective supply chain when all the inventory is under your control. It will be nearly impossible to deal with N retailers supply chain systems - inventory availability, location, cross-company orders, etc and be efficient enough to make a buck.
Unless Google is willing to physically get in the game and manage some warehouses I don’t see this being much of a threat although it might provide some nice competition for next and same day orders.
And hurting the stock is a good way as any to hurt a company.
That being said, most of the time when comparing amazon to local retailers, I notice the price is almost identical, except for tax.
[1] Google https://www.dropbox.com/s/vbnpel7w2mscpv2/google_express_pro... [2] Amazon https://www.dropbox.com/s/gh0rcveo1sqtmgw/amazon_prime_produ...
> Ouch, this is a really sad comment on Harvard these days; I'd be ashamed if a 14 year old made this qualitative vrs quantitative category mistake. (Which you since fixed, but please - it should never have been written).
Try looking into GOOGs shipping network, and compare it against AMZNs. Since we have long term data on AMZNs model (including $775 mil on a robot production facility):
1) Is GOOG outsourcing or doing this internally?
1a) If not, who are they using?
1b) If they are, where is the spend in their yearly budget reports? (Not seeing it)
2) Can they match the coverage that AMZN has?
3) Can they match the volume that AMZN has?
4) Can they hope to catch up before AMZN makes the robot jump it's planning?
And so on.
You're almost an adult - try to analyze the world a little closer, you're allegedly the elite of America.
Is HN really this naive? This is terrible, if this was from an Oxford or Cambridge student - well, sorry, I'd hope they couldn't produce this level of immature nonsense.
And yes, I suspect you'll "down vote" - but sometimes a short sharp slap should be required.
Oh, and the heavy use of trackers is both intrusive and unnecessary for a private site. You're not providing any service, barring selling yourself, which should again make you pause at the "whys" of using the trackers you are.
Google Shopping Express: schedule a window for delivery, during which you have to be at home.
Amazon Prime: I pick it up at my convenience from my mailbox.
When you place an order, there's a checkbox to indicate whether you want to require a signature, and a blank field for delivery instructions. It's up to you: if you don't request a signature, they'll happily leave your order wherever you ask them to.
(UPS/FedEx, on the other hand, will just leave a delivery note on my door. Depending on how risk-averse the sender and driver are feeling that day, I can sometimes leave a signature for the next delivery attempt.)
This guy is clearly an outlier and not suitable to use for judging ecommerce decisions.