Amazon Reveals the Robots at the Heart of Its Cyber Monday Operation | Hacker News Reader

Amazon Reveals the Robots at the Heart of Its Cyber Monday Operation

wired.com

105 pointsprostoalex61 comments

> pre-Kiva, pickers would walk the aisles to the bins storing the needed items. Now the shelves come to the pickers instead.

There's a great video of the things in action @ http://www.youtube.com/watch?v=6KRjuuEVEZs

It's like a preview of when our highways will be full of self-driving cars - very cool.

And I guess the next step is robots actually pulling the merchandise off of the shelves? Or maybe they'll just skip moving the shelves all together and just have collection robots roaming around, pulling product, and then packing it somewhere off to the side?

Makes me a little sad.. It's been almost 3 years since Amazon bought Kiva.. no improvements to the tech, and they seem to have just locked up the tech for themselves.

Before Amazon, Kiva was growing at 130%/year and had over $100m/year in revenue.

Hope Amazon eventually starts selling these again.. could really change fulfillment. But not so optimistic.. considering that would be walmart.com and other amazon competitors.

I interviewed at Kiva a few years ago, and asked about this kind of. The reply I got was that they will continue to support existing customers (besides Amazon) but don't have plans to bring anyone new on board.

Amazon doesn't want to sell them - Amazon offers fulfillment as a service. It is a definitive competitive advantage. That said, it may also be a great startup (tech companies certainly want, most technology issues solved, etc).

How difficult would it be for another robot company to build similar technology? Patents are also most likely an issue.

Not sure how hard it would be to copy Kiva.. but there's one other competitor: swisslog autostore

http://www.swisslog.com/en/Products/WDS/Storage-Systems/Auto...

I completely overlooked this.. they have a product that is nearly identical to Kiva:

http://www.swisslog.com/Products/WDS/Storage-Systems/CarryPi...

Starbucks did something similar with Clover.

    Won’t Amazon try to find ways to replace as many of those human workers
    as possible? Clark says that won’t happen, because the rise in
    productivity will give Amazon the means to grow. And growth means
    Amazon will need to hire more people.

I find this idea pretty interesting. That the amount of jobs stays the same even though robots do more and more of the stuff humans did in the past. So far it seems to hold true. We came a long way since the stone age and there still is useful stuff to do.

Maybe this will not change until robots are better at everything. Strong AI etc. But then we will probably somehow merge with the robots. Enhance ourselves with technology. And stay competitive.

Maybe jobs are forever.

For me, "The amount of jobs" is not the only factor, the quality of the job matters as well.

That paragraph suggests that the productivity gain benefits workers (labour) as well as the owners, (of capital) because it creates jobs. But each of those jobs could be a low paid, low quality job; as is suggested by the following:

http://www.motherjones.com/politics/2012/02/mac-mcclelland-f...

http://www.theguardian.com/technology/2013/dec/01/week-amazo...

http://www.salon.com/2014/02/23/worse_than_wal_mart_amazons_...

Yes there may be more (or the same) number of jobs; as suggested by the paragraph. But each of those jobs could be lower paid and lower quality in terms of job satisfaction. So it may still be the case that a greater share of the productivity gain accrues to capital; instead of labour.

Rather than freeing us; our existing societal structures may mean that technology instead helps to create a story of inequality - a hollowing out of today's society with "barristers and baristas and no-one in the middle". Whether that society is a moral one; or even a functional one, is a separate question I suppose!

What I liked about the article on the Kiva Robots at OAK4, is that it emphasized is that the part of the work that's being optimized is the grunt work of moving stuff back and worth - the (relatively) more sophisticated work of identifying the item, verifying the quality, and confirm it's the right element for the order, is done by humans. While it might not be very high on the feeding chain, it's at least a little higher than it was before - and that's a good thing.

I'm all for humans being automated out of the low-order activities.

http://en.wikipedia.org/wiki/Barge_Haulers_on_the_Volga#medi...

This is a positive interpretation of the scenario that may well be valid. Perhaps the question of whether job satisfaction at Amazon has improved or not is a little subjective.

I love your linked picture. I think that with it you imply that in the past, technological progress has been linked with higher standards of living for most. I'd agree with that, but I am wary of extrapolating that trend into the future; or maybe just concious that significant social turmoil came along with that progress.

Perhaps the situation is not so different, but we are merely doomed to repeat the mistakes of the industrial revolution; during the robotics/knowledge revolution!

Hasn't the average global standard of living more or less skyrocketed since 1950, even as the population has more than doubled?

The only substantial, technology related decline in standard of living over that period is of relatively well off Americans, primarily the blue collar middle class (and it's messy to actually try to compare across time periods like that, there are things that cost very little today that didn't exist 30 years ago).

I think the amount one gets paid is equal to the amount of value one creates for others.

After all it's an auction, right? If somebody offers you to pay $8/h you will not work for $7/h. And if your work would create an additional $9 for someone, they would happily offer you $8.

Minimum wage jobs are an exception to this rule. Because the amount paid is higher then it would be in a free market.

We could look at how we can increase the amount paid for minimum wage jobs. But I find it more fascinating to think about how we can make people create more value. So they do not need to compete for minimum wage jobs.

The employer will want to pay the least they can to get the job done. This means you will likely get paid an amount less than the value you create for others. There are a lot of cases where you may choose to work for 7/h even though 8/h is on offer elsewhere such as a shorter commute, better hours, or nicer work environment.

"There is one rule for the industrialist and that is: Make the best quality of goods possible at the lowest cost possible, paying the highest wages possible."

-- Henry Ford

http://www.brainyquote.com/quotes/quotes/h/henryford151869.h...

How do you explain Walmart (low quality goods, at the lowest cost possible, while paying the lowest wages possible, going so far as to show your employees how to use government benefits programs because you don't pay enough)?

Why would an employee work for less than the value they are creating? Seems like irrationally altruistic behavior on their part, or, alternatively, a marketplace in which the employers are not competing for employees (which has been known to happen in Silicon Valley).

In general, though - employees who are making less than the value they are creating, will simply be hired by another employer who can profit on that gap. This process repeats itself until the employee has reached the level at which they are making roughly what their value contribution is.

e.g. If you have a lawyer who bills out at $500K/year, and is only making $50K/year at Law Firm #1, they will quickly go to a law firm #2 that will hire them for $100K a year, and which will then profit $400K/year from their activity. That lawyer will likely find another job soon at Law Firm #3, where they will make $150K/year, and so on.

There are lots of reasonable reasons to work for less than the value you create.

- First, you can never actually reach 100% while working for someone else. If you reached 100%, then the employer would look at you and say "I can make exactly the same amount of money if I employ this person or not. Why should I do it? There is absolutely no benefit to me."

- You might want to work on a certain problem with a certain team. For example, let's say I want to work on the next Pixar movie, with the Pixar team. There's no employer other than Pixar that can offer that to me. They can pay me less than I contribute, the only thing they need to guard against is mass defection.

- Switching costs. If I am earning 98% of what I would like, switching companies might be not be worth the extra effort.

- Other intangibles. Unlike what some economists would like you to believe, not everything of value can be reduced to a single unit (money) and rational decisions derived from that. Things might be of value to me or my employer that aren't included in the value that I create or my salary. These things don't necessarily behave linearly.

Look at the profit per employee at firms such as Google, Apple, Facebook, and Microsoft. There are a whole lot of people being paid less than the value they create. (Remember, this is profit after expenses and salaries and such)

http://www.buzzfeed.com/peterlauria/what-18-of-the-biggest-t...

That's based on the assumption that only employee labor creates value.

The primary reason in a neoclassical model would be a lack of perfect information, wouldn't it.

An employee may work for less than the value they are creating when - not being omniscient - they are unable to correctly value their work.

That's the best response in this thread, and I think the most correct.

  Why would an employee work for less than the value they are creating?

When two parties are negotiating a deal which creates value, both parties want to capture as big a share of that value as possible; the outcome will depend on the relative strength of their negotiating positions.

Let's say you have a machine that can make hot dogs for $3, and a concession at a sports event where you can sell hot dogs for $8. You're busy that day, so you hire me to take the cash and hand out the hot dogs.

The combination of your machine and my work generates $5 of value for each hot dog sold. If you capture all $5 I have no incentive to work; if I capture all $5 you have no incentive to lend me the machine. So how much will each of us capture?

If you have time and you can find some third party willing to do the work for $0.50 per hot dog you'll have a strong negotiating position - either I accept $0.50 per hot dog or you have the other guy do it.

On the other hand, if it's at the last minute and you can't find anyone, and there's another guy offering me $4 per hot dog, I'm in a much stronger negotiating position.

The amount of value an employee creates represents an upper bound on how much value they can expect to capture - job market competition provides the lower bound.

> Why would an employee work for less than the value they are creating?

Because if the employee worked for greater than or equal to the value they created, the expected return on hiring them would be less than or equal to zero, and nobody would choose to hire them.

I understand why employers won't pay employees more than the value they create - my question was the opposite direction. Why (outside of a monopolistic or otherwise non competitive market) would an employee who is delivering a lot of value based solely on their contribution chose to work for significantly less than that value.

Clearly there needs to be some profit for their employer, but, using my example of a lawyer - there are thousands of law firms, and if a Lawyer is capable of driving $500K/of billable value, based solely on their effort - why would they chose to work for $50K when they could go to another law firm and work for $100K (or $150K, etc...)

I'm using an extreme example, but the same concept applies elsewhere.

The reasons I can think of why a person might not do this:

o Lack of information - the employee may not realize how much value their effort drives.

o Monopoly for the purchaser of their services - they may work in a niche field, in which their is only one purchaser for their service.

http://en.wikipedia.org/wiki/Surplus_labour

    This means you will likely get paid an amount
    less than the value you create for others

You are usually not creating more value then you get paid. Just because an author needs somebody to proof read his million-dollar-bestseller, the proofreader did not create millions of dollars. Just because a picker at Amazon picks stuff that makes Amazon tens of thousands in profit, he did not create that value.

To create value means to create value that otherwise would not exist.

You create some value, otherwise the company would not hire you. Of course the picker is not responsible for all the revenue accrued by the products s/he picks, just a part of it.

> I think the amount one gets paid is equal to the amount of value one creates for others.

I get what you mean, but thats not quite true, it is also supply and demand, intelligence, salesmanship, the ability to exploit others, and greed.

Of course a software dev can create more value than a picker at Amazon, but if what you said were true or if it were the only rule, there would be a correlation of performance and executive pay, which provably isn't the case.

Yes. To use ghshephard's terminology we would prefer to move people from "low-order" jobs to "high-order" jobs. This has happened somewhat in the past; as he points out with his linked picture.

My fear is that instead they are being moved to lower order jobs or are left with no job at all; while the benefits deriving from technological progress are accruing to the relatively few who own the means of production.

In the case of very low paying jobs, the employer has externalized some of the cost of the labor. They can get labor cheaper than it's true cost because someone else is paying for the food stamps and health care costs.

> I think the amount one gets paid is equal to the amount of value one creates for others.

What one gets paid is less than the value one creates. (Otherwise there would be no incentive to pay for that value.)

It's implied that the jobs might be of lower quality, for sure. However, this seems in contrast to their recent 'pay to quit' scheme: http://www.shrm.org/hrdisciplines/compensation/articles/page...

I don't believe it.

Amazon is known in Germany for abusing the system and paying as less as possible, while squeezing as much hours as possible to employers from their German sites.

Basically people at the job center get an offer that they cannot refuse, as they are unemployed, at the very minimum Amazon can get away with, without any kind of compensation for overtime.

http://www.huffingtonpost.com/tag/amazon-strike/

EDIT: Downvoting me every time I mention Amazon worker rights exploitation wont make them go away.

Is amazon committing illegal acts in Germany? One of the articles you linked to states that Amazon is already paying higher than average wages.

> Is amazon committing illegal acts in Germany?

No, but they are on the borderline of what is acceptable in German society.

> One of the articles you linked to states that Amazon is already paying higher than average wages.

I have not read them, as I was just trying to fetch some English related information for the HN audience.

The information I have, from memory, is from German newspapers.

The whole situation worked as follows:

- offer a job to people at the job center (quite nice)

- since they are at job center, the job cannot be refused unless for special reasons

- pay them the bare minimum, almost lower than the unemployment fund after taxes

- have them work extra-hours without compensation

- let them go, re-hire from the endless pool at job center

Also the guys doing deliveries, are at multiple levels of sub-contractors. Have a wage that barely pays the fuel and monthly expenses, with a variable bonus attached to the amount of successful deliveries per day.

Since they are not to blame for people not being at home, they try to deliver the packages to just about anyone in the neighbourhood, thus leaving a information to collect at place X, not always the post office.

There was an undercover story about this a couple of years ago, done by a famous German reporter, Günter Wallraff.

Maybe this situation today is different and has improved a lot, but this was how it used to be from the newspapers info.

Two of these are government-created problems, if the government would like to solve them, they can rewrite the rules

> since they are at job center, the job cannot be refused unless for special reasons

> pay them the bare minimum, almost lower than the unemployment fund after taxes

This one is illegal, with enough proof it should be easy to fix in courts

> have them work extra-hours without compensation

And two of these seem to be neutral on society

> offer a job to people at the job center (quite nice)

> let them go, re-hire from the endless pool at job center

> This one is illegal, with enough proof it should be easy to fix in courts

Since when? Do you live in Germany?

Do you remember the issues a few years ago with Hartz-IV and recruiting people for "Spargelstechen", as Hartz-IV was better than the salary being offered?

> - have them work extra-hours without compensation

I'm not familiar with German labor laws, but I know this is (typically) illegal in the US and I suspect that can't be different in other first world countries. Is it?

There is a limit under which it is allowed.

Hmm, I may have misunderstood. When you say "without compensation," do you mean they're still getting their normal hourly wage but without any additional compensation (overtime) for working the extra hours? Or $0 compensation at all for working that time? I was originally under the impression you meant $0 compensation at all, but I suspect I'm misunderstanding now.

Yes, that is what was being aired in the press. Courtesy of Google translate

https://translate.google.de/translate?sl=de&tl=en&js=y&prev=...

If you know German,

http://www.spiegel.de/wirtschaft/unternehmen/gesetzesluecke-...

It seems as if English is not your first language. You make some good points but do not assume that people are down-voting you because they don't agree with your opinions.

> I find this idea pretty interesting. That the amount of jobs stays the same even though robots do more and more of the stuff humans did in the past.

The point is that if the use of robots enables lowering the cost of online shopping, then more people can afford to shop online, and Amazon might even need to hire more people, to keep up with increased demand.

It's the simple case of an innovation increasing productivity (less work hours per unit of product produced), which means they can lower prices, and -- possibly -- gain customers.

If the increase in sales offsets the increase in productivity, then no workers need to be laid off. Of course, this isn't necessarily the case. An increase in productivity can lead to both layoffs, hiring and neither.

What he's avoiding saying is that while the need to hire more people isn't going away those human workers displaced by technology will, in all likelihood, be out of a job because there is now a higher-order job that they are unqualified to do. The savings from from automating the warehouse gets reinvested back into Kiva systems, for example. And it's not a 1:1 replacement either as it takes the elimination of a handful of minimum wage warehouse workers to let Amazon hire just one more engineer.

but that's the natural progression - if all you can do is move some packages from A to B, then you aren't worth very much, and when time comes that such jobs get automated, you _have_ to either skill up, or "die" off. Being able to make a good living isn't a right.

>I find this idea pretty interesting. That the amount of jobs stays the same even though robots do more and more of the stuff humans did in the past. So far it seems to hold true.

This is only true so long as they are growing. That growth comes from either growth in the end market or by taking market share from competitors.

I don't know the numbers, but I suspect a lot of Amazon's growth will be from taking market share due to lower costs because of those productivity gains.

If this is the case, then those robots are still displacing jobs, just not at Amazon.

Very cool, now the last step, automate the picker. So no humans are required to do this work. Saves a lot of money. And Amazon can be cheaper still.

This is what IT is all about. Automate the humans away. Now what are we going to do with all those humans that are not capable of doing work that cannot be automated away?

Amazon actually has a competition [0] to design a robot that can do the picking. The difficult part ATM is being able to grab hold of a large variety of objects, from keychains to footlong dildos. I'm sure that they'll introduce picking robots for the more uniformly sized products like books soon enough, or introduce a container system to provide a generic framework for the picking robots to use.

Still leaves humans to assert the quality of the products though, although with fully automated systems I'd imagine that products get damaged less while in the warehouses.

[0] http://amazonpickingchallenge.org/

I've always imagined that if amazon would have as much negotiating power as Walmart does, they could mandate suppliers to put items in standard boxing, so you could put a bunch of vending machines on top of kiva bots.

...or boxing that fits in standard sizes of storage containers that can be easily handled by the robots. Similar to how small sized by high-value items (video games, perfumes, etc) are often stored (and rattle around in) clear locked boxes on the shelves at department stores.

The manufacturers could be encouraged to reduce their box size (and, thus the amount of room it takes up in the warehouse) by being charged a fee for items that took up too much room and/or non-standard room.

For a while during Jobs' Keynotes they'd announce a new iPod, and then say how much smaller the packaging was. The end-user probably doesn't care, but for those in the supply change I guess it's a huge deal. Less airplanes and shipping containers from China, more stock on the same amount of shelves, etc.

Indeed.

Somewhat coincidentally, today is the 101st anniversary of the Ford Assembly Line, where more autonomous functions allowed the production of a car to be reduced from 22 hours to 1.5 hours. It wasn't popular though: to fill 100 vacancies over 900 people had to be hired, the majority promptly quitting the repetitive and relentless work.

and then the $5 day was implemented by Ford, and employee retention went from avg of 9 months to many years.

Turns out if you pay your workers they want to keep working for you.

Send them to Mars for the science

I am surprised that only 20% of the Amazon warehouses have the Kiva robot system. Maybe it is just the new construction warehouses.

From the article "In the US, Amazon now has a workforce of 15,000 Kiva robots deployed across 10 of its 50 domestic fulfillment centers."

So 80% still to get automated.

You can't just switch this stuff out overnight. Give it time and Amazon will probably run most of their FCs off Kiva robots, but there will also be some FCs that don't adopt them because their items are too large, oddly-shaped, or other reasons.

60 Minutes did a story on robotics in January of 2013 that showed Kiva robots in action. You can jump to 2:45 in the following link to see the demo: http://www.cbsnews.com/videos/are-robots-hurting-job-growth-...

Is there a weight limit for those machines? I take great pleasure in ordering silly heavy things off amazon with overnight shipping. Most recently a very heavy adjustable dumbell set.

Reply on news.ycombinator.com