When delivery is free, will ownership survive?

This is a story how self-driving cars will, through cheap delivery, change how we interact with our world.
Let me say at the outset: I am acutely aware that predicting human behavior is the pinnacle of folly. My goal is not to be 100% right. My goal is to find a new way to think about the problem.
Put another way, my aim is to start the conversation, not finish it. Comments appreciated.

JD robotStill from a video about JD.com’s robot delivery vehicles, already launched on college campuses in Beijing.

In the late 1990s, my father became a thief.

He would stay up late into the night in his office upstairs, his desk perched where my bed used to be. The green glow of the CRT reflecting off his forehead in the dark, he would sit nearly motionless, hour after hour watching the progress bar fill as he pirated old radio shows, books on tape, and music from the golden age of rock and roll.

Back in the early 1990s, it was not obvious that the internet would launch a global army of greying men into the world’s most tedious crime spree. It was not obvious college students and housewives and kids obsessed with Jamiroquai would join them to take down the record industry. And journalism. And video stores, encyclopedias, and travel agents.

In retrospect, we misunderstood what the internet was. In the mid 1990s, we saw the internet as communications technology, a way to send notes for work or update family about the kids. But as the internet grew and download speeds increased, it turned into something far more profound: History’s cheapest distribution technology.

Self-driving cars will bring the next great disruption in distribution. Like the internet, they will restructure not just their intended market – transportation – but every single thing they touch.

This post describes how.

Distribution is hard to write about because it is deliberately hidden from sight, buried as pipes and quietly routed through the backs of buildings. But distribution systems bring us every shirt we wear, every meal we eat, every plank of lumber that supports the houses we live in. It was a change in the technology of distribution that allowed my father to engage in petty larceny from a bedroom. It was a change in distribution that brought the record industry to its knees.

Self-driving cars will launch in 3-4 years, and immediately drop the cost of distributing objects to just a fraction of today. Distribution-by-robot has already started in China, and has the potential to herald as profound a change in society as the introduction of the car, train, or airplane – changes not seen in living memory.

And although it has been over a century since the last disruption in transport, we have a light to guide our understanding of what’s coming: The deployment of the internet. The turmoil that followed the roll-out of digital distribution – the demise of print journalism, the growth of sharing, and the end of music ownership, to name a few[1] – gives us a roadmap for understanding what may soon unfold in the world of things.

In a very short time, cheap physical distribution could carry our newly-acquired digital habits to the physical world. We could share objects. We could subscribe to stuff. We could go through life accessing every thing we desire from a mechanized, Lysol-misted cloud.

Disruption from the internet blindsided us. But this time, we have a chance to learn from history, to have a conversation about what world we’d like to build next. Prediction carries risks – and predictions involving people are the riskiest of all. Yet there is much to gain from reflecting on the disruption we just experienced, so we are not surprised to see, for example, the very concept of owning things recede into the rear-view mirror.

But we’ll return to ownership. To begin, the best way to glimpse what our future may bring is to dial back the clock a couple decades, and consider what my father and his friends did to the world of music.


RIAAPaid subscriptions (peach) and satellite and internet radio (SoundExchange, pink) are now a bigger revenue stream than purchased music (purples and blue). Source: RIAA.

In the mid-1990s, when I was living in Pasadena, California, I would walk from my student apartment through neighborhoods of stately Tudor homes to buy music at Tower Records. Tower didn’t have the best prices, but it was well-lit, with friendly staff, and stocked a copy of almost anything I needed.

It is, of course, gone now.

We did not fully respect it at the time, but Tower’s reason for existence was its selection – it was built to hold inventory. And in a world where music distribution was expensive, this was a sensible and economic thing to do.

Inventory let a customer browse for music. Inventory assured that sales closed. Inventory served lovers of pop, classical, or punk rock with equal facility.

And this is the primary function of retail today: House product, and wait for customers to arrive. Implicit in this business model is the assumption that a good’s lowest cost path to the home is for a consumer to drive, park, fetch, and return.

The internet created a cheaper solution. Optical fiber was the instrument of Tower’s demise, but any cheaper and faster distribution path would have sufficed; Amazon, FedEx, and UPS have steadily whittled away at the cost of distribution, and in doing so have ever so slowly chipped away at brick and mortar’s dominance.

Retail salesOnline sales keep going: Source: Census.gov

A gallon of milk or a bag of frozen French fries can’t fit through a fiber, and even today, with Amazon’s average delivery cost of about $8,[2] makes little sense to ship from warehouse to door. But what happens when self-driving cars cut the cost of home delivery all at once to $2, or $0.50?

In a world where an electric self-driving vehicle costs less than $0.20 per mile to operate, a customer would be happy to step outside and pick produce from a market that pulls up to the door. At that price, a consumer could afford to take just a single apple, and consign the store to its next destination. Can retail survive that kind of discontinuity in distribution?

Should it?

When distribution costs fall, the advantages of inventory can be duplicated by multiple quick, small deliveries. There is no requirement to design retail around a loading bay; with cheap self-driving distribution, grocery stores could be made smaller, nestled deeper into neighborhoods (and closer to customers) without sacrificing selection. Retail could focus more on its other roles: providing a place to meet, a place to gawk (like the Apple Store), or a place for inspiration and community (like a farmer’s market). And for some goods, retail could shrink into the delivery truck itself; compress into a kiosk; or vanish entirely.

But most interestingly, the conversion to digital distribution in music didn’t just snuff out the inventory at the retail store: It dispersed the inventory at home as well.


We never really wanted to own music.

For my entire childhood, our basement held shelf after shelf of reel-to-reel tapes, recordings of a crooning Bobby Darin or Fats Domino to remind my dad of the thrill of his teenage years. But the tapes were artifacts; they were clutter that got in the way of the real goal.

What mattered was music. And when music arrived on demand, the shelves in the basement could move on.

Our apartments and houses are full of clutter, closets and cupboards full of goods sitting silently in the dark, waiting like my dad’s tapes for their shimmering moment of service. My pantry spills over with cans of tomatoes and boxes of trendy grains; yet ingredients provide no joy until I cook. My pantry exists to save myself future trips to the store. My pantry exists because distribution is expensive, and slow.

A pantry is inventory. In a world where distribution to my door is cheap, fast, and reliable, why would I need it? Why not shrink my own storage, and receive small shipments on demand instead?

It’s not just our pantries; our lives are choked with inventory. Find any collection of things, at home or at work — it almost certainly piles up because the cost of distribution is high.

A shed? Your garden tools are used only a few times per year, and if properly cared for will outlive you. Why do you store them?

Your closet? Your winter sweater is inventory, and earns no interest. Every suit or pair of shoes you own represents your savings, slowly dying.

When you follow this simple logic – when distribution costs fall, goods disperse – the conclusions surprise us, in exactly the same way that music surprised us.

The average renter has $20,000 to $30,000 in possessions. An affluent homeowner has likely spent hundreds of thousands of dollars[3] to stock his home with things which sit unused for 99.9% of their lives.

While a favorite T-shirt or a collection of shoes might inspire passion, most ownership provides no joy: We do not fondly reminisce about the extension cord in the garage, or develop an emotional connection to our colander. We carve out spaces in our homes – shelves and cabinets and closets – at a cost of over $1000 per square foot in a city like London or San Francisco, simply because there is no distribution system that could deliver a thing at the moment we need it.

So what would happen if we spent thousands of dollars not buying stuff, but grasping it only during use, and releasing it when we are done? What if distribution was cheap enough that the cost of summoning and returning things was economically sustainable? What if it was widespread enough to be quick and reliable?

The answers can be found in music. For this is, of course, the basis of Spotify’s business.


There are already startups seeking to make the closet obsolete, offering to ship you new clothing once a month, or more. There is no need to do laundry; dirty clothes are simply packed for return. This is, simply put, Spotify For Clothes.

It’s possible to build a Spotify to distribute garden tools, or wine glasses for parties, or single-purpose kitchen equipment. When we use a thing so sparingly why, exactly, do we own it? Why not subscribe to stuff instead, trying out a fun gizmo, but not make the commitment to house and care for it for the rest of our days?

Sushi bazookaIt’s a sushi bazooka. Admit it, you want to try it. Image: coolstuff.com

There are plenty of motivations to subscribe, storing stuff ‘in the cloud’, and summoning it only when required. Just as with Spotify, we would benefit from access to a greater variety of supply, the ability to try something new, and freedom from clutter. The logic of subscription does not end with bits; the benefits of sharing do not stop in kindergarten.

Ownership is not a law of nature. It’s an artifact of distribution.


We learned with music that you don’t have to own a thing to benefit from it. You could, for instance, share.

The sharing of music surprised everyone, because upper middle class people – by which I mean most everyone who works in technology – do precious little sharing (at least in the US). Our friends, the people we feel most comfortable asking for a favor, are scattered miles across town. And getting the stuff from them – distribution – is expensive. So we buy things ourselves, perhaps borrowing a serving platter from the neighbors when visitors arrive.

But when digital distribution became cheap, behaviors changed sharply. People swapped music with strangers. We gave up ownership (full disclosure: It took a year of lobbying by my children to convince me to let go). And we found three new ways to engage with this modern concept, music and information-on-demand, that could map directly to stuff if distribution gets cheap enough.

The first was to share.

Western society has produced at least an order of magnitude more things than it actually needs; we could sustain ourselves for years, if not decades, by temporarily checking out items and returning them when we are done. This is, in its extreme form, a radical socialism of stuff, powered by cheap technology.

Of course, unlike music, physical things chip, scratch, and wear; it is not clear whether it is in the best interest of ourselves or society to release our precious possessions to the wild. But we can do great service by distributing that which we treasure least. The things we exile to the top cabinet in the kitchen or to a table at Goodwill could instead be assimilated into the world’s largest thrift shop, invoked from the cloud. The death of the estate sale could be celebrated, not mourned.

The second path we followed was to subscribe, to pursue a corporate sponsor to ensure quality, ease of use, and reliability.

It is hard for us, as individuals, to make a guess as to how many times we will use an egg slicer or a guest room towel over the course of our lives. It is even harder to estimate how long a product will last. So a logical response is to buy cheap crap, cross your fingers, and hope it doesn’t break.

But a business can integrate use over thousands of customers; they can understand which zippers are built to withstand a tug from a toddler, which dishes will survive a teenager’s casual toss across the table. Companies invest for lifetime value, not sticker price; they see things as infrastructure rather than product. Back in the good old days, goods were built to last simply because we had not learned how to build them cheaply. With shared stuff, goods will last because building cheaply makes no sense.

If ownership wanes, we would have to rethink all of our stuff, designing not just for reliability but for all aspects of a good’s life. Clothing could be constructed to be folded by robot or washed with less energy. Quality and artisanship could be rewarded more effectively if people felt they were procuring an experience with every use, rather than acquiring stuff. It is only half in jest that I tell my sixteen-year-old daughter that I’d rather she work in design than computer science, because that is where the jobs will be.

The third path, for music and all of the internet (including this blog), was to contribute.

This is the Linux model of stuff, where a person could make a bowl in a pottery class and then donate it for the betterment of humanity. Another person may feel a calling to patch a torn quilt, or retap a stripped screw hole.  The low cost of digital distribution has enabled Patreon as well as Amazon, nurtured individual writers such as Tim Urban as well as conglomerates like Buzzfeed. We could build an artisanal cloud, erected on the same mix of altruism, reputation, and community as Wikipedia or StackExchange.

The cost of distribution is a barrier to giving as well as receiving. All of the unconventional business models of the internet could apply to stuff as well, if distribution becomes cheap enough.


Mercedes-Benz Vision VanConcept for a Mercedes delivery van that can be loaded robotically. And yes, that’s a delivery drone on the roof. From Fortune.

Are we there yet? Are we even close?

I am greatly aware that all of this sounds rather like science fiction: A choreography of robotic shelf-pickers and self-driving cars combine to form a clumsy Replicator. A modern-day Captain Picard commands “Alexa: Tea. Earl Grey. Hot.” and 15-20 minutes later, a steaming pot appears at the door.

It would be slow and clunky. It would not fit on a space ship. It won’t work to summon every thing, to every place, with equal facility.

To be fair, the internet did not sweep the world instantly either: It was eight years from the first www site to the peak revenue of the music industry; a decade after the launch of Spotify for music subscription to dominate purchases. And to be fair, we don’t yet know how fast the price of distribution will fall, or what levels it will need to reach to unlock new behaviors.

But to point out that the future will not unfold all at once should not blind us to the realization that, thing by thing, our world could be transformed. Pantries refurbish to social space. Cabinets make way for art. Closets turn to bedrooms. A world of cheap distribution could be a world of fewer, better-designed things, ones that consume fewer resources, and leave more for humans.

And not only will products and apartments change, we will as well.

Consider mealtime. In the middle of the last century, distribution by truck forged a national food network, with fresh fruits and vegetables changing the way we ate. Produce and grains were selected for their ability to withstand spoilage on the journey from farm to truck to aisle to table, farmers famously breeding out rapidly-oxidized omega 3 fats in the name of shelf stability.

And now our bodies lack those omega-3s. The last great change to distribution technology altered the composition of our cells. The next is likely to as well.

Technology defines who we are as much as we define our technology.

My father never got to subscribe to Spotify, but would have been thrilled by the experience. My teenage daughter, earbuds ever-present, can’t imagine a world structured another way. A drop in the cost of distribution took away the clutter of listening, leaving just the music, and the enjoyment.

Does it make sense to stop there? Or is it, as they say, time for things to change?

DadDad, resting while the family views an art installation
1 I am explicitly holding myself to the consumer space for this piece, to keep things tractable. But the reorganization of the software industry using new innovations like the API call can have physical correlates in manufacturing as well.
2 Amazon’s delivery costs amount to about 10% of total goods shipped, and the average on-line order is about $80, so the average delivery charge can be seen as about $8. It’s not an exact science, and the product mix at Amazon tilts very much towards lighter, higher value goods than what we bring home in groceries.
3 Homeowners insurance is usually sized to cover the replacement cost of possessions, at roughly 40% of the cost of the home itself. If you have a $300,000 home, then you should assume it would cost around $120,000 to replace your stuff.

18 thoughts on “When delivery is free, will ownership survive?

  1. I was thinking about the new businesses that send you clothes every month (and you send them back), and then you mentioned that. There is a “tool library” in West Philadelphia where you pay a nominal fee to be a member and then you can borrow various tools — makes much more sense than buying something you won’t use regularly. And I’d love to be brought food nearly instantly (and fresher) than my every-other-week shopping trips. (I like to cook, but like you, we accumulate a lot of food I just don’t get to as expected…) I think your thinking on this makes a lot of sense, as long as the distribution pipeline holds up. If the infrastructure for distribution fails — if we don’t invest in roads or bridges, or the air gets too crowded with drones to be safe, etc. — then maybe not. But it does seem as if we’re heading that way.

    1. I would love access to a tool library; we certainly have enough screwdrivers in this world to keep us going for generations. We have a maker space near me, but I don’t need bandsaws and 3D printers (though props to those who use them!) But I’d love to not have to go buy the 5-prong screwdriver to replace the hard drive on my laptop. It’s funny how it’s the little inconveniences that inspire, but that would be rad.

  2. Very interesting implications.

    If goods come to you instead of you going to the goods, what’s the effect on traffic patterns? If ownership and consumption patterns are static, then traffic doesn’t increase (it just reverses direction between retail point and consumer, basically). It could actually descrease because distribution could be made more efficient (multiple deliveries per vehicle, route optimisation, more flexibility in delivery times).

    But if we are talking about renting/borrowing most things, it’s not so clear. Would it require a magnitude more traffic, for example? If so, how much does this new distribution infrastructure (vehicles, roads) offset the gains made in reduced production of goods, both in economic and environmental terms? Reduced production of goods will of course change some other aspects of transportation where the new goods are moved into and through the distribution channels. As you mentioned, storage/inventory would also likely look quite different throughout the supply chain.

    Another thought: does this reduce the amount of food that has to be thrown away due to spoilage, both in the supermarkets and at home? I’d guess that it does. Increased efficiency of food distribution and consumption means that production can decrease, and land use can decrease correspondingly, which is a great outcome.

    1. There is an economic concept called elasticity of demand, which represents how much more demand is created when price falls. For transportation, the short term elasticity is about 0.3, which means that you’ll see 3X commerce if you drop prices by 10X. In the long term (accounting for the time it takes to build new infrastructure), the elasticity rises to 0.7. It’s hard to know if we should believe those numbers, because a small delivery bot seems very different from a car, but it’s a place to start.

      As for waste, I think you hit the nail on the head. I am very excited about that as well, because 40% of food is wasted, and that is a huge contributor to global warming. Right now “land use” accounts for 25% of greenhouse gas emission; if you are right, and we can decrease land use, it’s a big deal. It’s a story I’m trying to wrap my head around as a potential future post; if you run across anything on it I’d love to know!

      1. This paper suggests that food waste contributes 2% of total emissions and also has a breakdown of waste between distribution/retail/consumer & by food category: http://centmapress.ilb.uni-bonn.de/ojs/index.php/fsd/article/view/198/182

        This document is also interesting, although it focuses on transportation: http://www.cleanmetrics.com/pages/ch9_0923.pdf. It suggests that food production is by far the largest contributor to emissions when compared to transportation, cooking and waste disposal. Storage seems to be quite a large contributor (and can result in imported food producing fewer emissions!), which I thought was also pertinent to this discussion.

        More papers linked here: http://www.cleanmetrics.com/html/library.htm

        I’m by no means an expert so I don’t know if these are good sources. I was just looking it up because I was curious.

  3. The possibilities for sorely-needed restructuring of the economy of goods are worth exploring. So-called durable goods, which as you have pointed out elsewhere can be much more durable than they are made to be today, for economic reasons, become far lower in cost-per-use than perishable ones. This changes the definition of scarcity, and by extension the definition of wealth. This could be as disruptive as the transition, say, from feudalism to mercantilism. Will it have a correspondingly large and positive effect on human dignity and freedom? One might hope.

    1. I honestly see technology as a threat to capitalism, though coming from both directions: Sharing on one end, monopoly on the other. I am not sure how that one is going to go, but it’s one of the stories I’m thinking about tackling next. We can hope that dignity wins over feudalism; not ready to lay down my chips yet, but I can hope.

  4. You don’t mention 3-D molecular printers, which may be able to produce almost anything in-house. You won’t even need Amazon to deliver items…just print out what you need, from a diamond to a spoon, to a hamburger, to a shirt. Just fill the printer with the raw materials…probably powders or liquids providing various elements or small molecules, and load the software to control the printing of whatever you want.

  5. I’m a chemist by training, and back when I was in graduate school I was excited about the prospect for molecular building blocks. But man, it turned out to be hard – to give some context, the average time from lab to product in materials science is about 20 years. I’ve worked on stuff that finally hit the market 50 years after invention. It’s tough stuff.

    That said, I am *very* interested in 3D printing overall. The opportunity for customization is compelling; as much as I love the idea of sharing, it won’t work for arch supports and ergonomics. Innovation is coming from all directions, and it’s pretty cool.

  6. Brilliant Sir! Your article is actually a very practical window in terms of the enabling of the sharing economy. Given that UBI might come into the picture as well, we might have a situation which enables local self sufficient communes enabled by highly efficient supply chain. Companies may start seeing themselves as enablers to P2P commerce. Possibly the end of capitalism may be (as one of the comments above puts it) might be capitalism again.

  7. I have read your article and shared it extensively. It does provide a solution to peak oil potentially reducing the peak. And getting a fully- charged autonomous car would make our lives much richer. We already do JIT food with condiments because we rent our place in the winter when we snowbird for 5 months. This just makes the process much easier. Like I only have one sweater but sometimes need some variety even though I am in perpetual summer. I do rent tools after downsizing.

  8. I wonder if by changing the distribution model such that less stuff is needed, how that changes the producer business model. Is the cost of entry (R&D) too high if you don’t make and sell as many widgets, even if the per widget sales price is higher because it is shared? I’m curious how this balance would be struck.

    1. A good question, Jenny. I did not directly state in this post (though will in subsequent ones) that delivery is fairly undifferentiated. This means it tends towards economies of scale (there is little difference between companies other than price and on-time performance, and both get better as you get bigger), which means that monopolies or oligopolies will control distribution. If oligopolies control distribution and have some pricing power, then manufacturers end up with very little margin.

      I think that structure also tilts product design towards specialized experts who have a brand name that matters (e.g., Oxo); if people demand that brand, then the distributor will be forced to carry it. So brands can make money on one end, subscription/delivery will make money on another, and producers have razor thin margins, and will invest as little as possible in R&D.

      The best reference I can think of would be this post from Stratechery: https://stratechery.com/2017/intel-mobileye-and-smiling-curves/. Take a look, and see what you think.

Leave a Reply