Winner Takes All

By Douglas Rushkoff. Published in Market Leader on 1 June 2016

Those of us who thought the digital marketplace was going to follow the distributed nature of the net, with its decentralised connectivity and ad hoc social activity leading to an equally distributed marketplace, got it wrong–at least, in the short term. We thought that instead of buying everything at Walmart and watching our personal and community wealth being extracted by a highly centralised corporation, we would now enter a new phase of peer-to-peer commerce. The values of the bazaar would be revived by the internet. A new, digitally enabled, people-driven economy would dominate as industrialism’s extractive growth mandate needed.


In a nutshell, we implemented our digital business plans without any real awareness of the deeper principles we were espousing, much less the ones we were challenging. For too many, the internet was supposed to take care of all this, as a matter of course. If we just built the platforms for internet commerce, they would be inherently decentralising, empowering, and good for humankind, as if by their very nature. We made the same old mistake of underestimating the importance of our roles as purposeful human programmers in determining how these platforms would impact the existing marketplace.


With a few notable exceptions–such as eBay or Etsy–we didn’t really get a return of the many-to-many marketplace or digital bazaar. No, in online commerce, it’s mostly a few companies selling to many, and many people selling to the very few–if anyone at all.

Take music. The best part of an online music catalogue catalogue is that it is unlimited in size. The local record store can only hold so many albums in its bins. A website can list everything and anything, however obscure, at virtually no additional cost. A surfer in New Zealand can purchase an album by a lute player in Norway. Wired editor and economist Chris Anderson called this the ‘long tail’ of widespread digital access, which would support many more times the artists, writers and innovators than could be supported through traditional distribution channels.

His theory was that the low cost of reaching customers online would enable thousands of hitherto unpopular titles to become popular. Since the marginal cost of selling different music files was negligible, Anderson argued, the online merchants would now make a profit by “selling less of more”.1 The marketplace would become more diverse and support more creators as the long tail of former losers fattened.

Yet it turns out that’s not what’s happening. Instead, according to Nielsen SoundScan, a few blockbuster hits make up a greater percentage of all the music sold than ever before. In the days of physical albums and CDs, the industry rule was that about 80% of sales came from the top 20% of products on offer at that moment. That means that the bottom 80% still accounted for 20% of all sales. On iTunes today, the bottom 94% sell Iess than 100 copies each. Just 0.00001% of tracks sold accounted for a sixth of all sales. And these figures are roughly the same for every creative industry, from books to smartphone apps.

However we slice it, digital selling platforms exacerbate the extremes between superstars and those who sell nothing. This is because of a phenomenon called ‘power-law dynamics’. Normally, the popularity of everything from people to products is fairly evenly distributed along a bell curve. That means just a few members of the population are extremely popular, most are moderately popular, and a few are utterly unpopular. .Just like school or a bookstore, most kids and books are in the hump in the middle of the bell curve.

Internet business experts expected digital platforms to flatten out that curve, giving less popular members the opportunity to connect with new friends, audiences, or consumers, while also taking away some of the advantages enjoyed by the entrenched stars of radio and television. But instead of a new, fatter, longer tail for formerly obscure products to to thrive, we got an extraordinarily ‘hit-heavy, skinny tail’.2 Instead of a flatter bell curve with a big “middle class” of participants, it maps out like a steep slope upward, from losers with nothing straight to winners with everything.

This is what’s meant by a power-law distribution–basically, a winner-takes-all disparity, like the infamous 1 per cent.

For some reason, the original industrial-age mandate for extractive, monopolised growth was not only still in force, but getting worse. Net economists were quick to blame human nature. “This has nothing to do with moral weakness, selling out or any other psychological explanation,” explained Clay Shirky in 2003. “The very act of choosing, spread widely enough and freely enough, creates a power-law distribution.”3 Others went on to use these naturally occurring power-law dynamics to rationalise the injustices of capitalism and increasing wealth inequality. It’s just a function of increased choice–a product of human freedom itself.

What we were missing is that once again, and in spite of how social they may appear, these platforms remove living human beings from the process of selection. We are looking at numerical ratings, online reviews and top sellers. We chat to customer service ‘bots’ that are programmed to stoke our purchases, or people reading from automated scripts, who may as well be bots. Like most online platforms, selling sites remove the sort of human buffers and intervention that often slow things down and replace them with frictionless digital cycles.


For instance, when a bricks-and-mortar record store plays a particular song on its audio system, the tune will sell more copies. But that’s just one store. What happens when a subtle cue takes the form of an online recommendation on a website? The recommendation leads to more sales, which then become a new data point that is fed back into the automated system through which more recommendations of the same or a similar sort are made. This is called positive reinforcement, but in the digital realm it’s more of a feedback loop, instantaneously reinforcing itself again and again, growing and spreading.

The overwhelming variety of possibilities leads us to gravitate to machine-winnowed lists, if for no other reason than to make the selection process more manageable. In tum, the more we depend on an item’s popularity for discovery and selection, the more we reinforce that item’s popularity, and the more of a winner-takes-all landscape we created.

So these extreme divides between winners and losers are not simply an expression of human nature. They are an expression of one aspect of human nature, amplified by machines at the expense of all the others. In fact, it’s by trying to imitate human, social reality that the biggest distribution platforms, from Amazon to Netflix, create and promote the distortions that lead to power-law distributions.

But the biggest and most unacknowledged factor contributing to the net’s winner-takes-all effect is that these platforms are highly centralised. iTunes sells the music, Netflix sells the movies, and Amazon sells the books (and almost everything else). Everyone passes through the same digital turnstiles, sees the same lists and recommendations and is subjected to the same algorithms. These monopolistic commerce platforms are not true peer-to-peer systems and they are anything but freeing.

They are growth machines–digital department stores, where the many purchase items from the few. We are all buying from the same few places and people. Continuing to do so only reinforces their position at the top, leading to more of that same centralised growth invented by the aristocracy to disempower everyone else.

Compare an Amazon, Netflix, or iTunes to a peer-to-peer platform such as eBay. Whatever one might think of eBay’s corporate ambitions, its basic business model is anti-industrial, at least in that it connects buyers directly to sellers. Most seIIers are real human beings with used items to unload. Although ratings matter, they are less about driving consumers to particular product choices than reassuring them of the integrity of the amateur seller whose product they have already found. The platform does take its cut–which adds up to something tremendous–but it creates a new opportunity in return. Instead of removing humans from the marketplace, eBay enhances their capacity to create and exchange value. Instead of monopolising value, or limiting value creation to just a few players, peer-to-peer platforms distribute the ability to create and exchange it.


These principles can be applied intentionally by any online marketplace. For instance, Bandcamp, a music streaming and download service much like iTunes or Spotify, distinguishes itself by intentionally working against power-law dynamics. It caters to less-established underground and alternative artists, charging less than half the sales commission of its competitors. Unlike the ‘Top 40’ emphasis of most music sites, Bandcamp eschews download counts and leaderboards in favour of a “discover” button. Users wade through their favorite genres the way they might have once flipped through the stacks at a record store.4

Current digital marketing dogma dictates that by promoting aimless surfing and sampling of music, a site like Bandcamp will only generate a “tyranny of choice”, overwhelming users with so many options that they won’t be able to pick anything at all.5 But were people overwhelmed back in the day when they had to walk into a big record store and browse through the bins? I’m not so sure. And even if the net has trained people to accept the two or three choices at the top of popularity lists, I’m fairly confident we can regain the ability to shop. The real panic about such sites is that they emphasise human unpredictability and work against the industrial-age logic of removing people from every link in the value chain. They distribute the ability to grow.

At the very least, Bandcamp and eBay (as well as Maker’s Row, Etsy, Indiegogo, and many other consciously programmed sites) prove that digital platforms don’t have to lead to power-law distributions. They only end up that way because we are programming them to support our inherited strategies for mass production. We optimise for more directed consumer choice, less human intervention, volume sales, and monopoly control of a given marketplace.

When the expectation of free labour from consumers dovetails with the winner-takes-all lottery of the new mass marketplace, we end up with a whole lot of people working for nothing. After all, the power-law distribution writ large is really just another way of saying income disparity.

1 Chris Anderson, The Long Tail: Why the Future of Business Is Selling Less of More, Hyperion, New York, 2006.
2 Will Page and Eric Garland, ‘The Long Tail of P2P, ,Economic Insight No 14 2014.
3 Clay Shirky, Power Laws, Weblogs and Inequality,, 8 February 2003.
4 Denise Lu, Spotify vs. Bandcamp: Which Is More Band-Friendly?,, 19 November 2013.
5 David Bollier, The Tyranny of Choice: You Choose,, 18 December, 2010.