Amazon Go and the Future | Hacker News Reader

Amazon Go and the Future

stratechery.com

284 pointsmisiti3780172 comments

The impressive thing about Amazon to me is that it thinks/acts for the longer term. I guess it's Bezos that's the driving force behind this (he funds the Long Now 10,000 year clock project after all, so must have some interest in the long term).

Very many things become possible when you think on long timescales. Things like space travel.

I know Elon Musk is the poster boy for it at the moment, but it's comforting to me at least that In Bezos there's another 'moon-shotter' who thinks on the timescales required to actually achieve such long term goals.

I know that's a bit of a tangential leap from retail stores to becoming an interplanetary species but the long term mindset behind achieving the two the same.

There's good and bad to that. You want people thinking big and long term but you also don't want that to become more important than the short term and the small things.

That abstracted, encompassing way of thinking doesn't really care if Amazon's delivered a bad product that isn't worthy of consumers this quarter because there's always a next, and what's one customer and one sale when you've got 2 billion customers, a media empire, and and missions to the moon?

There's only so much focus any individual or company has. I guess in politics that would be political capital, there's probably something similar in business. If Bezos is the driving force behind Amazon, how does that scale as Amazon takes on more and more? And longer-term, does that carry over when he's not around anymore due to old age or a bus?

Yes, you are right, it's always good to keep a weather eye on the here and now lest you lose the source of energy powering your long term plans.

Actually, I feel we could use much more long-term focus. Because right now, pretty much every single company is solely focused on "the short term and the small things", and the result isn't delivering good products and making consumers happy - the result is optimizing products and services to minimum possible quality, while gaming consumer reactions, in order to extract as much profit as soon as possible.

Certainly agree with you in principle, but is Amazon really any different? I wouldn't want to give a family member or friend a Kindle Fire, because I Know that'd be locking them into a closed ecosystem with so so build quality. It's maybe pretentious and idealistic, but why is long-term thinking to lock people in?

This is the very first Q/A on Amazon's own Fire review page:

The first Gen Kindle fire had a faulty connection and will come loose overtime causing the device to no longer able to charge. It's a common problem but a hard fix, you have to take it apart and reattach the connection

Wouldn't long-term thinking not overlook details like that? That seems like the result of getting a product to market fast and then making consumers pay for the improvement in the next model. Which is fine if you are the entity that plans to be around for a hundred years selling those iterations, but not so much fun for the customer with a device that won't charge.

> Wouldn't long-term thinking not overlook details like that?

I suspect it depends on what the long term goals are. If quality is one of them, then over time, as they improve their processes in favor of quality, these kinds of quality problems should become less and less frequent.

Well, there's one form of long-term thinking like "let's build a device that lasts a long time," and another like "let's build a product line/customer base that lasts a long time."

Amazon's mostly interested, it seems, in just giving people more ways to interact with them, more ways to consume their content, more reasons to subscribe to Prime, more ways to get locked in as you say. Having a long view is not necessarily equal or even in the ballpark of being consumer-friendly.

I'm not sure that focusing mostly on making Fires cheap as dirt really means there's not long-term thinking going on. A buy-it-for-life tablet, of all things, probably isn't a great investment from anybody's perspective.

Tons of large companies think long term. It might seem unique in the context of the tech world, but it's far from it. Thinking long term doesn't always translate into great advancements. Particularly with large companies it can just turn them into a force that just eats up all competitors and stagnates (e.g. Walmart). Remember, investing in ensuring competitors can't enter the market is also long term, but it's not good for society.

I think what's important is investing money in moonshot projects, which is only one of the many ways that companies can think long term.

It's unfair to say that Walmart is stagnating. They aren't taking moonshots, but their eCommerce division is no joke.

It also doesn't hurt that Walmart effectively hit its moonshot. It murdered a million small businesses / local stores / mom & pops and in the process did make things a lot cheaper for many people.

They optimized the supply chain of the traditional retail model to its limit (and then over it in legally questionable ways) and changed how people buy a lot of things forever.

That being said, Walmart wasn't really innovating in doing it. It just did what everyone already knew worked but did it with less empathy and more scale. Its much the same way the current Amazon store site for regular things isn't innovative - they took something other people were doing and mixed in some scale and evil and came out on top in that regard.

To Amazons benefit, it was/is innovative in web services, book sales, warehousing, etc. But as a retail store? Until recent years they just took the Walmart model gone digital and ran with it, and somewhere along the way took the Ebay model and pasted it in as well.

It might also be to Amazons benefit that it was founded within months of the first ecommerce websites rather than some thousands of years after the first store chain or a hundred after the first integrated supply chain.

just out of curiosity, what drives you so strongly with themes of "lacking empathy" and "evil"?

He's probably aware of WalMart's history and impact.

https://en.wikipedia.org/wiki/Wal-Mart:_The_High_Cost_of_Low...

"The Arkansas-based company reported strong growth online, with e-commerce sales soaring 50 percent in the fiscal third quarter, a year after it acquired Jet.com. That's a slower pace than in the prior period when digital sales were up 60 percent."

https://www.cnbc.com/2017/11/16/wmt-q3-earnings-2017.html

Blue Origin matches that -- it's likely a Bezos-driven thing. I toured their facilities recently and what impressed me most is that they're building a base for repeatable, safe-as-airplanes space travel versus the get-to-Mars approach of SpaceX. Each stage of their development and planned development is along a clear, slow timeline to reach that end goal.

> that they're building a base for repeatable, safe-as-airplanes space travel versus the get-to-Mars approach of SpaceX.

Er, SpaceX is "building a base for repeatable, safe-as-airplanes space travel". That's the whole point of BFR and point-to-point transportation on Earth with it as well. Safer than airplane travel is a must-have for space travel, SpaceX included.

It's nice to know Blue Origin thinks like that too, but if you think that SpaceX doesn't have the same concern for safety, then I think you've drank too much BO kool-aid.

The smart way to become an interplanetary species would be to throw money at physics research and AI.

The technological payback from an automated grocery store is unlikely to help us get to Proxima Centauri.

That sounds harsh, but I think there's a fallacy in believing that technological progress is cumulative, and each small step - no matter how tangential - gets you closer to the big goal.

It's like arguing that if you keep developing steam engines you eventually get a modern airliner.

In fact you only get a modern airliner with completely new technologies, including GPS, composites, fly-by-wire, digital airflow modelling, and so on.

You can't build an A380 with Victorian sheet metal and rivets, no matter how many times you iterate.

And there's something Victorian about Amazon's need to dominate and mechanise every possible consumer market. The single-pointed focus makes it smarter than legacy corps when it comes to retail, but it absolutely does not follow that this proves Amazon is smart enough to invent its way out-of-system.

I don't know anything about physics research, but it would seem that there is a more nuanced balance between basic research and things that can be applied soon. Basic research requires a high risk tolerance and deep pockets, but a potentially huge return.

For example, I can't imagine how much money Google has sunk into self-driving. And, it's now looking like they'll make it back thousands of times over. But, years ago, it could have failed after a 9 or 10 figure investment. To sustain that (outside of government funding), you have to be doing something else that pays the bills.

At least with your given example you effectively cannot reach the scientific knowledge required to manufacture composites and especially turbines without already having a complete understanding of steam power. One is substantially less complicated both theoretically and implementation wise than the other.

And, from a business perspective, if you don't grab the low hanging fruit now to stay competitive your competition will and you will find yourself obsoleted and bankrupt on your quest for airliners in the age before steam trains.

I like reading stratechery time to time for tech related strategy writeups. But this writeup somehow loses plot. The focus is too much on what are fixed and variable costs and the repeat of Amazon's buyout of Kiva.

What would have been an interesting discussion is to look at how the retail market is actually unfolding:

https://www.bloomberg.com/graphics/2017-retail-debt/

And whether cutting out the variable costs of a cashier etc is really worth the trouble in R&D and systems.

Edit: After reading some of the responses, I wanted to add something. If we look at the linked article above, retailers are taking in loans to open more and more stores. The idea being more stores = more customers = more profit. But that is not materializing at all. The question becomes - where does the issue lie really? Are variable costs the reason? What does Amazon Go solve really?

Sure, some might say better software is the whole point of Amazon Go. But,we are not talking about a SaaS where the cost of delivery is nearly negligible - it takes nearly the same amount of real estate and fixed costs to deliver 1 or 1000 of a SaaS product. That is not true for B&M. If they want to ramp up then they will need to open more stores - that requires real estate and other fixed costs.

There's not just the reduced cashier costs, but the vastly increased revenue per foot of store that will come about when customers know they don't have to wait in line.

The store will also be less empty per dollar of sales; or, said another way, shoppers will spend less time in store, creating the appearance of an emptier store for shoppers walking by, making it less time-consuming to enter.

Oh, and don't forget those prime signups just for people who want to try the store (though the number of non-prime subs when these stores rollout will be not huge).

Indeed.

The article didn't touch on how this approach reduces friction for buyers, which is hugely important. For the short-to-mid term, I suspect that frictionless shopping is a bigger deal than change in cost structure that comes from booting cashiers.

Exactly this!

Stratechery has a giant blind spot about Amazon's strategy, completely missing their #1 objective: customer obsession.

Bezos was clear about this from the beginning [0] and he repeats it every chance he gets [1]. He's not just bullshitting, this is actually the strategy, and it fucking works!

Stratechery opined that the Whole Foods deal was about putting Amazon products in physical stores. Of course not, it's the opposite -- it's about bringing awesome Whole Foods products to far more customers than Whole Foods reaches today.

Now Stratechery says Amazon Go is about eliminating cashier labor and leveraging fixed costs. Ridiculous!

If a well-paid cashier costs $15 per hour and can scan 12 items per minute, the cost is only 2 cents per item.

In fact, the cashier savings is probably more than offset by additional labor on the sales floor and to maintain all the sensors, cameras, turnstiles and software.

This is all about Bezos' customer obsession. If I don't have to wait in line, I'll happily shop Amazon Go every time I can, multiple trips per week. As a customer I'd be thrilled to eliminate the dead time of waiting in a line.

[0] 1997 Letter to Shareholders https://www.sec.gov/Archives/edgar/data/1018724/000119312513...

[1] 2016 Letter to Shareholders https://www.amazon.com/p/feature/z6o9g6sysxur57t

I must be weird. At no time has the idea of a line ever influenced my decision to go into a convenience store.

Or people like me who don't go shopping in convenience stores at Saturday are weird. I only do it under duress if I forgot something the other day, because I know the line will be so damn long.

Happens all the time to me, I'd rather give my money to someone with more appropriate staffing than wait in line to buy something.

So where do you go instead?

Sometimes nowhere, sometimes down the street. I do the same thing for restaurants, if there's a wait or my reservation isn't ready on time I'll go somewhere else. There's tons of competition out there, no need to waste my time for the benefit of a business owner.

I agree with you, and go one step further. I refuse to use self-checkout. For two reasons.

1. I don't like the thought of contributing to a system that costs people (cashiers) their jobs. 2. The whole reason I go to a store is for customer service. If I wanted self check-out, I'd order online from one of the same-day-delivery outfits.

I don't like the thought of contributing to a system that costs people (cashiers) their jobs.

People in boring, low paid jobs shouldn't be there. There are far better things people can be doing than working a till. The idea that we're doing people a favour by 'enabling' them to have a job when that job is tedious menial work a tiny compute can do is nonsense. We need to automate those jobs away to free people to do more interesting, fulfilling things.

We also need to do things to work out how to enable those people to do more interesting work, like figuring out how to pay people if the free market fails, but that's a separate issue.

Not a separate issue to the person losing their job!

These have to be solved together, or big trouble will be coming.

> or big trouble will be coming

The implication here is that the trouble isn't already on it's way. Those jobs are operating on borrowed time at this point and I think they'll be gone sooner than most people think. This is a problem that we should have started addressing several years ago at least.

>We also need to do things to work out how to enable those people to do more interesting work, like figuring out how to pay people if the free market fails, but that's a separate issue.

Seeing as the free market has deeply failed many, MANY people (and stands to fail even more in the near future), I would hesitate to put more people out of work before addressing this "separate issue." Lots of people can have their lives ruined before those in power decide it's time to think about how to ameliorate the damage done.

In the history of economics and politics, that has never happened. Tumult creates unacceptable carnage, and the solutions fall out of the carnage. Furthermore, we don't always correctly predict the form that carnage will take, so it's not even as if we know what acceptable solutions might be until things are already in freefall.

So although it's a nice sentiment, I'm not so sure it's realistic.

When textiles automated in England, the carnage included a generation of weavers starving to death.

Consumerism and technology have coincidentally grown to provide a few more generations of employment. But now that's ending.

And sure its not happened in history, because we've never been exactly here before. But we're here now! And there's no sign that solutions will fall out of the sky to fix things. I for one don't want to wait for starvation or revolution.

then don't wait for it - ontario NDP adopted basic income as a platform point, basic income tests are happening in a number of countries. Advocate for it or something else that is attempting to solve that problem.

At the very least, we can push back against the notion that this upheaval is "for the best" simply because it's the outcome of the market. It's Panglossianism at its worst. Sometimes things that benefit capital are bad for the world and for the rest of society. Wealth doesn't trickle down except by political will. That political will can only organize when we, the very fortunate few who have better resources and education than most working class people, begin to give a shit about something other than what is technologically possible or efficient.

That's correct. But this is why you should shape your purchase habits in different ways than choosing lack of automation.

For instance, choose to shop at a coop grocer, where the workers are paid a livable wage and usually also become member-owners with voting rights and a share in the profits. (There are also consumer advantages to consumer coops, obviously.) In this case, it'd be to the workers' advantage to automate things which make sense to automate as they share in the higher profits and possibly a higher wage as well.

Personally, I don't use self-checkout usually because cashiers are almost always much faster when you have a lot of stuff. But generally, I'm much in favor of automation.

Just think of all the people that were displaced when the horseless carriage came about. Society will adapt. It may be painful for those directly affected, especially in the short term, but that won’t stop progress.

There are some people who are really only capable of doing these "boring, low paid jobs". Not everyone is capable of being an information worker or a hard physical laborer.

Once a basic income or solid safety net has been established, we can consider whether it makes sense to take actions in order to not obsolete low paid jobs.

Heck its not all about low paid jobs. That's just another domino. Folks who used to be office managers, technical writers, factory line managers find themselves automated out of a job. They turn to lower-paid jobs.

To compound it all, now the lower-paid jobs are going too. Leaving a 'bow wave' of steadily-lower-wage folks completely out of options.

Your second point doesn't apply here. No matter what you have to walk around and pick up the items you want (at least for now, if you're going to a B&M). The difference here is waiting in line vs. just walking out of the store when you're done.

And yet self-checkouts have not replaced cashiers. I often see more people waiting for cashiers than in the self-checkout line at the stores I visit. When I use the self-checkout I'm giving the cashiers an opportunity to have more time to serve those who choose them instead.

Lastly, every time I walk somewhere I don't fret about costing a taxi driver their job.

I'd rather just give someone money to live and let them find something productive they'd like to do with their time than shoehorn them into menial jobs that don't need to be done for the sake of having a job.

1. I don't like the thought of contributing to a system that costs people (cashiers) their jobs.

So only when you need customer service you care about their jobs?

Come on think about your statement for a moment and realize what the internet has done. You are part of that.

NYC Trader Joe's aren't convenience stores but my shopping times are informed almost entirely by the expected line length.

I avoid Safeway and Trader Joe's because the lines are so damn long...

You must be lucky to live where you do. I’ve definitely timed visits to Costco, Trader Joe’s, and Wal-Mart to avoid long lines at checkout.

The self checkout doesn’t help, it actually takes longer than a good cashier.

Agree; Costco is far more palatable on a Tuesday at 7 PM than on a weekend.

I think of Go as a better self-checkout. Those machines are awful, and Go’s tech basically does the same thing but in a smoother, more scalable way.

>I think of Go as a better self-checkout.

Yeah, I don't think this type of thing is actually that big a deal for a supermarket where stuff needs to be bagged etc.

But it seems like a qualitatively different experience than self-checkout at a convenience store/small grocery store.

I also expect a lot of US readers tend to think in terms of either supermarkets or fairly junky 7-Eleven style convenience stores. But there are a lot of places in Europe where there are more intermediate options like the Marks and Spencer markets.

Amazon Go is much more akin to a 7-11 than to a Trader Joes or Costco. They have yet to show that this model even works on this tiny scale. The capability of this tech to scale up effectively to an actual grocery store much less to a warehouse type store is completely unknown. (I personally wouldn't count on it.)

> The self checkout doesn’t help, it actually takes longer than a good cashier.

Yep. Someone who scans and packs thousands of items a day is going to be much better at it than I am, given that I'd only scan a few dozen items a week.

Unless the lines are really too long, I'll always take a cashier over using the self-checkouts.

Or maybe you've never experienced a convenience store with a significant line.

If there are 2 or 3 people in line that doesn't dissuade me from shopping there, but if I see a long line of a dozen people winding through the store, I almost always go somewhere else.

I've run into that at gas station convenience stores a lot -- I go in for a drink or snack while pumping gas and if there's a long line, then I just go back to my car and leave.

>I must be weird. At no time has the idea of a line ever influenced my decision to go into a convenience store.

I've walked into a store to buy something in the half hour before some large lottery is about to close (and everyone seems to be buying tickets), and walked right back out again.

Lines are also the primary reason I don't shop at certain stores (e.g. walmart).

In a way, what Amazon is really testing here is how to make brick and mortar shopping competitive with online shopping. But rather than undercutting their online business what they're really doing is finding ways to compete with other retailers in spaces that are hard online, such as groceries and immediate items.

A Prime subscription isn't required to use the store.

https://arstechnica.com/information-technology/2018/01/we-te...

It's per usual. Looks like Ben's strategy is to publish empty bait articles in his free blog, agitate the net, and than collect all that knowledge and give it to his subscribers. Smart strategy, but ,meh.

Are you a subscriber? The weekly articles are not bait. They are longer about bigger theories and trends. The emails are shorter and more focused on news. He just applies and develops the same theories to the daily news. Both are very good IMO. Who is better in this space?

I don’t agree this article is bait, but I listen to the podcast and I do like that it’s recorded long enough after the article to factor in some of the responses. It’s a valid strategy (and a good micro-example of his aggregation theory.)

"And whether cutting out the variable costs of a cashier etc is really worth the trouble in R&D and systems."

No, it is not worth it... for a retailer. This is what makes it an effective moat. Fast forward 10 years and Amazon will have a cost advantage over all the other retailers. Amazon store's can be open 24 hours without any extra costs. They can be much smaller or hold items that are less profitable.This will eventually pay for the moat but it will take a very long time. Longer than most companies are willing to wait for it to pay off. Amazon will also you the tech in more ways than just these stores.

"Fast forward 10 years and Amazon will have a cost advantage over all the other retailers."

Once this happens, then Amazon can sell every retailer Amazon Retail Services the same way that it sells AWS. Maybe they can even be deployed via shipping containers on demand. State fairs, festivals, summertime around lakes, disaster recovery areas all have a spike of consumers that is limited by where retail storefronts are already in place and mobile facilities like food trucks.

I don't think so. They want to be the only middleman. Suppliers can buy shelf space in Amazon stores but I don't think they will help retailers. They want to make a small percentage of every sale.

It's a long term play, I think. All of the tech becomes a sunk cost, and over time the removal of variable costs creates the ability to scale with costs still trending to 0, or, less upwardly.

I wish there was a mention that 5 people can buy the same milk and all can each pay a different price (and none know what the other paid either)

I always thought it would be something to charge a wealthy person more since they can afford it. With GO franchise i could charge $20 for milk to the teacher pensioner backed by my tax dollars; $0.50 to the poor mother of 3 kids, and $10k to jeff bezos. Then I could stop subsidising farmers.

Of course my points are extreme but we never had a way to charge different people this easily

I wonder if only poor people will visit such stores in the future, once the word of this gets around.

Yeah, all the people that this would affect negatively (as OC described) have the power to prevent it. It likely would work the other way around where the rich automatically get it deducted from their taxes, somehow.

> all the people that this would affect negatively ... have the power to prevent it

This is based on the incorrect assumption that the people being gouged by variable pricing are rich (or at least "not poor").

> (as OC described)

The OC's pricing - like his patently incorrect view of teacher income - isn't based on reality. Being poor is very expensive[1][2]. The entire point of this new wave of price discrimination is to use modern data analysis to minimax how much each individual can be gouged. This is always regressive, because the poor do not have the same ability to shop around for better deals, buy in bulk, and other money-saving techniques that require up-front investment of time and/or money.

[1] https://www.huffingtonpost.com/lauren-greutman/why-being-poo...

[2] https://www.economist.com/news/united-states/21663262-why-lo...

I completely agree with everything you've said.

My point was that the OC's scenario couldn't happen because, if it were even on the table, they would prevent it from happening.

Thereby allowing poor mothers with 3 kids to start a business as supermarket shoppers for Jeff Bezos types of people?

But then their prices would go up because their income goes up, pretty much crippling their business model.

So they are encouraged to recruit other poor mothers to become part of their downline and receive a percentage of what those mothers bring in.

Basically, a MLM but with milk and Jeff Bezos at the core.

10/10, would invest

This is quite common in third world countries. Essentially, foreigners, whether expats or tourists, will pay a different price to locals for everything from street food to rent to government run tours. Local prices are displayed in the local language, foreigner prices in English.

One of my go-to lunch spots offers a discount if you have an in-state Drivers License.

> charge a wealthy person more since they can afford it. With GO franchise i could charge $20 for milk to the teacher pensioner backed by my tax dollars

You think teachers are wealthy? Dude,... get some experience with reality instead of believing b.s. you read online.

Yeah, I don't know what type of world they live in. I know a lot of teachers, most of which work for 30-50k their entire lives.

After an entire life of teaching/putting up with kids, I don't think it's unreasonable for them to be rewarded with an actual retirement. It's not like their pension allows for them to buy a new car, even.

It's strange seeing this sentiment from an industry that routinely makes over six figures, many while still in their 20's or 30's.

Teachers never come anywhere close to that in their entire career.

My example is strictly NJ based. A different place 4 sure.

The checkout free experience doesn't magically exempt them from price display laws.

Here a store has to display prices for a high percentage of the items on the shelves and if there is a discrepancy at the register the displayed price is the decider.

It'd also be perfectly possible to implement different prices with a traditional barcode scanner. Just have a button on the register (or say, give a discount if the person signs up for a store card...).

So they display the lowest price they'd accept (could even be automated) and have the checkout try to change the price to higher values based on the customer.

If the customer notices a discrepancy, they give them the lower price. They win out in the long run since it depends on the customer to notice the discrepancy and point it out.

Any minor advantage at their scale wins out as a pretty big deal.

And then they get sued for breaking the law?

What you described is called fraud.

> If the customer notices a discrepancy, they give them the lower price. They win out in the long run since it depends on the customer to notice the discrepancy and point it out.

That kind of fraud is what the corner bodega does, not what Amazon does.

Amazon offers free bodegas to anyone as long as they get a cut.

That's called fraud.

I think a better approach would be to display an initially high price, and mention automatically applied coupons at checkout - or require the user to use their smartphone's app to apply the coupon.

That makes calculating the final price far too hard. If the customer has to use a smartphone app to shop anyways, why not just have the app display the personalized price?

This is how Amazon Books stores work. Prices aren't displayed on the shelf, but using an app. The price is different for Prime members vs. non-Prime members.

Last time I was in an Amazon Book store, BOTH prices were displayed on the shelf - Prime and regular. Has this changed?

No prices on the shelf in this article: https://www.recode.net/2017/5/24/15683852/amazon-books-nyc-t...

Maybe it depends on the state, and its laws on price marking.

Grocery stores already obscure prices with specials.

$3.75 each Or buy two and get the second one at 25% off with bonus card.

I suspect that few customers do the mental arithmetic to figure out what their final cost will be.

Look at Recode's pictures. There are prices for the items on the shelves.

https://www.recode.net/2018/1/21/16913984/what-does-photos-a...

Btw. is our current system of farm subsidizing that different?

Subsidiaries by and large are paid from money collected through taxation. Anyone's level of taxation is ~based~ should be based on the personal income.

So we actually have this, somehow, today. Without the need for every shop to actually know about the customers wealth level.

Are there no prices on the shelves in an Amazon Go store? Surely people want to know the price before they buy an item? Not having a checkout doesn't change that does it?

I wonder too. I assume they are electronic. So snowstorm- raise milk and bread.

So the prices change based on who's looking at them? What if two people are looking at them at the same time? How does this work?

How about the one who happens to be looking at it the "closest" (according to the algorithms)? That would pretty much immediately turn into a "I see you can buy it for a lower price, how about I pay you to buy it for me instead" situation...

I suppose you could use lenticular screens like those used in some cars to avoid driver distraction[0 - old article but you get the point] or for the 3D effect in the Nintendo 3DS. This might be getting a bit silly though...

[0] - https://www.popsci.com/cars/article/2008-12/mercedes-benz-sp...

Yes there are prices on the shelves.

https://www.recode.net/2018/1/21/16913984/what-does-photos-a...

So what is losteverything talking about? How do they charge people different prices if the prices are printed right there on paper?

Not this specifically, but with a totally electronic system and no point of sale to anchor there is large opportunity to easily and programatically have dynamic pricing at the known individual level. Supply, time of day, whatever can be instantly baked into a pricing system.

And why not? No humans needed to change pricing. Plus i would think you could scan to get "your" price. Prime member $x. Etc..

I hope i did not imply this was happening now..

Aaaah... you said

> I wish there was a mention that 5 people can buy the same milk and all can each pay a different price (and none know what the other paid either)

And the answer why it isn't mentioned in the article is... because it's not a thing that's happening.

Yep. The author gives reasons for Amazon's decisions. I wanted a pricing discussion. My bad

It's hard to do that in a physical store since everyone can see the same price display. I doubt many people would be happy in a store where they need to look up every single price on their mobile device.

Kohl's seems to be doing pretty well with that system. They print a high price on the tags, then discount it by rack, and mail out 30% off coupons with short expiration dates to some of their rewards card holders every month.

Wealthy people buy the cow.

Wealthy people invest in a fund that has tax-sheltered elements operating in the global cow sector.

So how much of my personal life do I need to hand over to a faceless corporation so it can decide how much I ought to have to pay for milk?

Just my annual income, or should it also take into account if I've had an expensive medical procedure recently? Or my number of dependents? The value of my stocks? Or if my breadwinning spouse just threw me out of the house?

What a treat this info would be for the ad department.

This would send me back to the traditional grocery store.

Not going to happen as long as GO has to compete with traditional grocery stores.

If that system was ever popularized, rich people would just have poor people shop for them. More so than they already do I guess.

I have such a visceral dislike of this. I don't trust algorithms to be fair (see multitudes of examples in Weapons of Math Destruction), or for algorithms to consistently work without interference from malicious actors.

What something should cost should be it's "worth", however two parties entering in a transaction deem that.

Or, worse to worst, I think there's a case for "vintage stores" if dynamic pricing goes live at brick/mortar stores--come one, come all, pay the same.

Poor people are not allowed in the store. A login to your Amazon account with the app on a smart phone is required to even enter the store.

When Amazon Go was first announced, my guess was that they were using RFID tags on every item in the store.

I now understand why that was misguided: the cost of RFID tags would scale linearly with the amount of goods sold. With Amazon's camera/ML technology, they can reproduce this store wherever they want for just the cost of the hardware and maintenance.

I really appreciated the article for explaining that.

> Well, I would ask, what about the labour of Marx’s day, the factory workers borne of the industrial revolution that he thought should overthrow the bourgeoisie?

From what I understand, Marx's contribution to political thought is his philosophy of history. it's not that the proletariat would overthrow the bourgeoisie in his day— but rather that it was simply an inevitability. History was marching towards it.

I know Ben was using it as a way to tie-together political implications— I'm just clarifying.

The last section of this piece is funny because the first 3/4ths of the essay is fundamentally Marxist in its descriptive characterization of the role of technology in capitalist economies. In particular, this part of the last section is straight-up wrong:

> Marx saw a world where capital subjugated labour for its own return; technologies like Amazon Go have increasingly no need for labor at all.

First, because even the author's idealized view of Amazon Go does require labor (the author even discussed this at length; that labor takes the form of R&D). Second, because Marx did see this possible world, and even wrote about it extensively in chapter 15 of Capital.

I find that happens a lot in articles like this one -- ideas that originated with Marx are parroted by an author who doesn't realize that Marx wrote extensively on the role of technology in capitalism.

General advice on articles like this one: If you're going to write about labor and technology in capitalist economies, and don't want to unknowingly parrot lots of Marxist ideas, you should read capital. Capital certainly contains some prescriptions you may disagree with, but it also contains a lot of descriptions. If you're thinking about labor and technology, there are better than even odds that you're re-inventing one of Marx's descriptions. He was the first one to say a lot of the obvious stuff.

Even more general advice: western stigmas aside, familiarity with Marx's writing is as indispensable to understanding modern economic thought as familiarity with Smith, Hayek, Keynes, et al.

At Amazon, the debate to set up brick and mortar stores used to come up at every year's planning session. For Amazon it was never simply about doing brick and mortar stores for the sake of having brick and mortar presence, but categorically how would our stores be fundamentally different (experience, value to customers). If there was not a clear answer to this, the discussion was punted to next year.

With their bookstores they have fundamentally altered the buying experience. And now with their grocery store too, they have fundamentally altered the experience. They only open it up to public once the "experiment" has worked. Expect this to be just the beginning. But over the "long-term" this will become the new normal. There is no question about that.

It is the long-term thinking that leads to these experiments and initiatives in the first place. Not all long-term thinking is alike (comparing Amazon with other companies). At Amazon, it manifests into every aspect of business, strategy, and operations.

I wish the author would skip the accounting and econ 101 segments that eat half of this article. The author's writing is usually kind of fun, but it is unfortunate he feels his readership doesn't understand the difference between fixed and variable costs.

Personally I appreciated it

I liked these sections.

It’s always amazing to me how many developers don’t understand marginal cost and opportunity cost and therefore misunderstand the true economic power and forces behind software.

I like how Microsoft, Google, and Facebook were framed and it was essential for the author’s point on how Amazon is trying to do for reality and real stuff what others did for digital.

This is also a key piece to understanding why developer salaries can be so high. Revenues can grow much faster than costs, which makes the high salaries economically possible.

Don't have mba or ever studied economics, so that helps.

The line to get in reminds me of how the line for self checkout is sometimes longer than the staffed lines.

Not to mention the self checkouts, at least from my experience, need to be staffed as well. One person can maintain maybe 3-4, so it's not unusual to see half the self-checkouts closed.

Self checkouts can just appear to be closed for no reason because of things like it's rejecting coins, or keeps getting coins stuck, or the weighing scale isn't correct. Faults with over complicated machines rather than a personel problem

Sure, problems with individual machines happen. But in my experience, it's almost always banks of machines that are closed due to staff limitations.

It's quite common for a single staff member to manage 10–12 self checkouts in Australia. This will improve given time for the general population to warm to the SCO process combined with improvements in the machines themselves.

It's new, everyone wants to try it. If this becomes the norm, the line won't be an issue.

Self-checkout lines usually move much faster though.

That seems to be a function of the number of items that people bring to those checkouts rather than anything else. All things being equal, it's always been my experience that the self-checkouts are slower, not faster.

Sure, but there's one line for 4-8 self checkouts which makes the line move faster even if everyone is a technophobe.

Usually because they do the correct thing and have one line for 6 machines, while every other line is for one particular cash register.

I always go for the staffed lines because they do the job faster than I can do it myself. Especially for things where someone needs to look at your id, like buying alcohol - a cashier is faster than self-checkout.

Funny thing- when self checkout first opened in California you could buy alcohol from them. Unfortunately, the unions pushed hard to get that banned.

It's tiresome to read articles like this that just accept these loss-leader vanity projects as real, impending culture shifts. This is a clever convenience store built to test and more importantly showcase Amazon's ideas for automation and machine learning. But it's not a grocery store, and it's not yet anything close to a scalable business. The idea the author floats that this type of place could ever be open with no staff present is ludicrous.

To get a bit more specific, this Recode article has actual pictures from inside: https://www.recode.net/2018/1/21/16913984/what-does-photos-a...

The store has a bunch of ready-to-eat and semi-prepared bundles that hit a very narrow niche in the grocery market. There are a few pre-packaged name-brand products, but everything is arranged in a highly regimented manner, and the store is very small. It's revealing to note what's missing: any fresh produce whatsoever.

What I'd rather see from these articles is a bit more of a critical eye on the claims. Does the company actually think they could leave the store open with no employees present? Do they even have a reasonable premise that fewer cashiers even means fewer employees? (Someone needs to be there to help people who don't have the app, or to watch that no one jumps the turnstiles, or to keep the place clean, or to restock the shelves.)

And what about the software itself? Do the cameras keep track of the inventory correctly? What if a customer puts a product back down in the wrong place? How much awareness does the software actually have? Is it tracking your phone or you? Does it attempt to deal with scams (can you bring an empty boxed meal kit into the store, replace a new one from the shelf with the empty, and then walk out)?

Maybe these aren't big concerns for this single store, but they are huge concerns if you want to scale this to compete with 7-11 or Pret a Manger, both of which, I will note, have more fresh fruit available for sale.

Now, I don't think Amazon is really thinking that sort of thing is imminent, or that they are actually going to make money from this store. This is a vanity experiment, nothing more. They are getting lots of free advertising from a credulous press, and that's plenty sufficient for them.

True, even as a vanity project it could increase Amazon's share price by enough to justify the effort, even if it fails.

Have they ever said that these stores would be 100% free of staff? Someone has to stock the shelves and prepare the food.

Maybe they'll offer produce in individually wrapped portions, or a tiny Kiva robot will bring it to you, or a robot arm, or edible NFC tags... I'm sure they are thinking about it. To just write it off because you thought of a few corner cases is premature IMO.

>Does the company actually think they could leave the store open with no employees present?

>Do they even have a reasonable premise that fewer cashiers even means fewer employees? (Someone needs to be there to help people who don't have the app, or to watch that no one jumps the turnstiles, or to keep the place clean, or to restock the shelves.)

Yes, it seems pretty obvious you reduce the labor that needs to be done per customer, you reduce your marginal cost of labor per customer (and replace with fixed and RD costs).

>And what about the software itself? Do the cameras keep track of the inventory correctly? What if a customer puts a product back down in the wrong place? How much awareness does the software actually have? Is it tracking your phone or you? Does it attempt to deal with scams (can you bring an empty boxed meal kit into the store, replace a new one from the shelf with the empty, and then walk out)?

Amazon isn't incompetent at software, and finding any edgecases they hadn't considered is precisely why they need to run this experiment on a small scale.

> This is a vanity experiment, nothing more.

That is pretty obtuse. This is far more than a 'vanity' experiment. Even if Amazon never launches this beyond one store, Amazon is experimenting to gain a unique understanding of the potential future of offline-retail (which is important as offline-retail based companies like Walmart continue to directly compete with them by moving online.)

Amazon's purchase of Whole Foods seems to indicate that they are serious about exploring offline-retail and are not simply looking for free advertising from a credulous press. (Although I'm sure that doesn't hurt.)

You're of course right that this first store isn't turning a profit in itself.

But I'm sure it's far more than a vanity experiment. If it's possible to build a more effective store this way, Amazon will figure it out, and in 5-10 years there will be thousands of them across the world.

Not to mention what happens when this store gets robbed blind and the PR fallout form that event. I know it’s unlikely, but something like that could have very damaging effects on whether this will see widespread adoption.

I’m only saying this because I’ve lived in bad enough neighborhoods where I don’t think just having a single person at the door is going to stop a robbery. Unless it has some sort of auto locking, store lockdown feature.

You are correct, if Amazon Go stores end up being targets for robbery that would damage roll out of additional stores the problem is that I can't think of how any one could rob them blind. The primary problem being that there is no cash to steal in the store and the goods all have low price to bulk ratios making them just plain difficult to steal. They just aren't worth robbing. Not that someone won't try anyway.

That isn't to say they couldn't be targeted by shoplifters, but that is a huge difference from being robbed blind. And even here the store has two advantages in that login with the app is required to even enter and the whole store is covered with sensors and cameras in addition to the staff.

As for the PR fallout I'd actually be interested to see how that goes, how much biometric data Amazon was able to gather on them even with a mask and/or bulky clothes, how the security features of the store actually work.

Are would be robbers really that interested in stealing tuna nicioise salad boxes and quinoa wraps? Robbers want cash from the till and things they can fence quickly, seeing as this store doesn't deal with cash, I suspect the only petty thefts will be people circumventing the camera/sensor technology to steal their protein enriched breakfasts shakes

They sell alcohol, I would imagine that would be worth taking to some.

A normal store doesn’t even have the turnstiles. Just grab your stuff and walk out casually. How do they keep from getting robbed? Why is this different? At least in the Go store, the system can presumably detect people who haven’t badged in / jumped the turnstile, before they actually do anything.

To answer some easy questions:

>> The idea the author floats that this type of place could ever be open with no staff present is ludicrous.

No on has ever said the store will have no staff present. At the very, very basics: it sells alcohol and is in the United States. It has staff present. Not to mention security, stocking, etc.

The value proposition is a lower margin cost and better customer experience, not zero staff.

>> The store has a bunch of ready-to-eat and semi-prepared bundles that hit a very narrow niche in the grocery market. There are a few pre-packaged name-brand products, but everything is arranged in a highly regimented manner, and the store is very small. It's revealing to note what's missing: any fresh produce whatsoever.

What does this have to do with the store's ability to scale? How does pre-packaged food (perishable) differ from fresh produce (perishable) if they decide to stock produce? Are you worried about Amazon's ability to handle logistics?

An initial test location in a limited space (this is downtown Seattle) has to have customers and they identified pre-packaged food and consumable items like wine as what the customers in the area wanted, not a scaled down Whole Foods.

>> And what about the software itself? Do the cameras keep track of the inventory correctly? What if a customer puts a product back down in the wrong place? How much awareness does the software actually have? Is it tracking your phone or you? Does it attempt to deal with scams (can you bring an empty boxed meal kit into the store, replace a new one from the shelf with the empty, and then walk out)?

The store has been in private beta to thousands of Amazon employees for over a year. It's safe to assume they're confident in their technology. Other tech sites have been given permission to try and trick the software. They failed.

>> Now, I don't think Amazon is really thinking that sort of thing is imminent, or that they are actually going to make money from this store. This is a vanity experiment, nothing more. They are getting lots of free advertising from a credulous press, and that's plenty sufficient for them.

Even if this were true, you fail to make the case here.

>What does this have to do with the store's ability to scale? How does pre-packaged food (perishable) differ from fresh produce (perishable) if they decide to stock produce?

Unless it's all packaged, produce needs to be weighed. Produce is one of the main ways that self-checkout, even for small purchases, really breaks down in grocery stores today. One of my local supermarkets has switched to mostly self-service and I've largely switched to a different chain specifically for that reason.

The local news here says the stores will be staffed. There's just no checkouts.

More to your point, when you have as much money as Amazon, you can afford to throw every idea at the wall and see what works. If you generate a crap ton of publicity along the way, that's just gravy.

This same article could have been written 120 years ago about Automats https://en.wikipedia.org/wiki/Automat

Or the tech reinvention of automats: https://www.eatsa.com/.

One interesting question about this is that of strategy. How will Amazon use this to strengthen it's power?

One possibility is of course building convenience stores. but the task is immense - there are 150K conevinence stores in the US, more globally. and they have only a limited window until competitors copy the tech.

So they will probably want to get the maximum strategic benefits in that time window.

And while licensing money is nice, that's not the best asset convenience stores has got. Their best asset is their location and their space. If Amazon get access to some of that, they could build a distribution network, maybe for grocery pickups, maybe at a global scale, and at a very fast growth rate.

And somewhere along this process, i suspect Amazon could convince or get enough power so that convenience store owners will partner with them, for example on selling Amazon private label.

How "limited" is that window? Let them copy it.

Surely Google could develop this in less than 5 years - and that's not enough time to own all the stores in the world.

Google won't do it. They are incapable of anything that doesn't involve selling ads. I think that's a major point in the article. No other company capable of executing this seems to be interested in doing it! Only Bezos!

When I got into computers eons ago, I really believed that they would be a force for social good: relieve tedium, improve the lot of most humans, and so on. But I really have trouble seeing a huge benefit to the inevitable elimination of many checkout jobs across the world. Sure, lining up takes a bit longer, but there is a social aspect to grocery shopping that is disappearing, and that is not good. I just extrapolate to a world where 3 or 4 companies own everything, and I wonder where that leaves all the other people on the planet.

> but there is a social aspect to grocery shopping

What are you talking about? I have never in my life had any meaningful social interactions with grocery store employees.

If you shop in a neighborhood store, and your neighbors also shop there, you run into people you know at the grocery store. Not employees, but customers.

Will removing a cashier significantly change this?

Yes; there is no line to chat in.

I shop at a neighbourhood grocery store, and I regularly chat with the checkout staff. One of them brought her daughter to our house on Halloween. I know many of them by name. Sure its not a deep connection, but it is a friendly "Hi, how are you doing".

Retail with no employees was done in China last year, fwiw:

https://www.technologyreview.com/s/608104/in-china-a-store-o...

"Purchases are made using an app to scan a barcode and paying over the phone."

That's like the difference between a fully autonomous car and an automatic train.

The economics speak in this is just wrong:

    A cashier is a marginal cost... for a convenience
    store to sell one more item requires some amount of
    time on the part of a cashier, and that time costs
    the convenience store operator money.

Since the cashier is paid regardless of whether the item is sold they are not a marginal cost in this context. If anything other than a publicity stunt Amazon Go is instead an effort to increase employee productivity, similar to grocery stores putting in self-checkout kiosks where two employees can supervise 10 to 20 kiosks instead of staffing one traditional checkout lane.

>Since the cashier is paid regardless of whether the item is sold they are not a marginal cost in this context.

They're not a marginal cost but they are mostly a variable cost. Cost accounting is tricky. Most things aren't pure fixed costs or pure variable costs but cashiers are pretty close to the latter.

The points about marginal costs are true, but I would argue the thinking driving the Amazon Go store is the customer experience:

How do we improve the retail experience for customers and differentiate?

Then they worked backwards to make the economics work.

These stores, if they open at scale, will most likely lose money, but the customer experience will be so far beyond current brick / mortar stores that going to a Walmart becomes a jarring experience. Just like Amazon.com vs. other e-commerce sites. That's the moat.

I wonder if Amazon tried other concepts like putting cameras at the checkout instead so putting all your food on a conveyor and let cameras scan and weight everything. Similar to that at the airport scanning your carry on items.

Scaling the store size would seem extremely expensive, so much infrastructure.

I used this store every day over the summer when I worked in this building. Don't underestimate the reduced friction of cashier-less shopping: I went to that store way more often than the alternatives. I was able to get in and out of the store in around 30 seconds if I knew what I wanted.

First Amazon kills retail stores. Then it opens its own retail stores. Sounds like a winning strategy.

They disrupted the space, killing traditional stores. Theirs will not be a traditional store.

Once upon a time our local small town had a general store. The freeway came thru, and it went out of business. Then a gas station came in by the freeway. Then it started offering some groceries, clothing, amenities. Hey Presto! and we have a general store again.

Once other super markets adopt automatic check out, shopping at supermarkets would be a lot less bad~.

The article doesn't explain why Amazon wants to create a future straddling both sides of online and offline retail. If drone deliveries are the future, and VR is going to be as good as real, then, isn’t offline retail headed for a decline? Isn't that what Amazon wants us to believe?

Amazon wants to be part of every financial transaction you make, and they realize that not everything can be done online.

> If drone deliveries are the future

Noone thinks this for 100% of deliveries. I think about 98% of my Amazon orders today would be too heavy for the first 10-20 generations of drones. Not to mention, many people live in places where drone delivery will never be allowed by law.

> and VR is going to be as good as real

Noone thinks this either, not for all people and not for all time. Amazon isn't making financial decisions based on thinking that 100% of people will eventually be in VR 100% of the time. Even if the technology is massively successful, that would not happen.

> isn’t offline retail headed for a decline?

Yes of course. It's declining from about 90% of money spent to about 80% over the next few years or so (estimates)

> Isn't that what Amazon wants us to believe?

What? Amazon wants your money. They don't care what you think the future will be like.

You're right. I wasn't suggesting that VR/AR and drones would take over the retail landscape in a flash but these developments, combined with the general convenience of online retail, might make offline buying a highly unfavorable option for plenty of everyday things. If groceries being sold at Amazon Go could be bought just as easily through an app and delivered reliably in a short while with no additional cost, would the sophisticated system remain just as useful?

I have a lot of trouble with hot takes like the quoted tweet. It's not funny. It's so dumb. It's pointlessly negative. Sometimes I feel like this kind of midbrow objection is sucking out all the headspace that should be reserved for thoughtful commentary.

I don't see the quoted tweet as an pointless, negative objection at all. It's just pointing out the irony that a supposedly lineless store is so popular that people are lining up to get in.

>this kind of midbrow objection

Pot. Kettle. Black.

Reply on news.ycombinator.com