,
30 tweets,
13 min read
Read on Twitter
Do network effects create winner-take-all markets in tech? The answer has huge implications for VC’s, entrepreneurs, and antitrust regulators. I made these “Network Effects Market Maps” (h/t @RyanLeeHagemann for designing) to help organize my thinking. 1/ niskanencenter.org/blog/you-cant-… 
First, what is a network effect? Telephones are the best example: the value of having a telephone increases with the number of other people who have telephones. So, in general, a network effect is when the demand for X depends on the number of people already using X. 2/
.@JamesCurrier at @NFXGuild traces the concept of network effects back to a 1908 annual report from AT&T Chairman Theodore Vail in which he said... 3/ nfx.com/post/network-e…
“A telephone — without a connection at the other end of the line — is not even a toy or a scientific instrument. It is one of the most useless things in the world. Its value depends on the connection with the other telephone— and increases with the number of connections.” 4/
Network effects can be direct or indirect. Telephones are an example of a direct network effect since the value of the network scales with the size of one user group (i.e., telephone users). 5/
An indirect network effect is when the value to a user in one group increases as the size of a *complementary* user group increases. Uber is a good example here: the value of Uber’s platform to *riders* increases with the number of *drivers* and vice versa. 6/
So what kinds of markets have significant network effects? In the analog era, we had communications networks (e.g., phones, faxes, mail), marketplaces (e.g., shopping malls, stock exchanges, nightclubs), and media (e.g., newspapers, magazines, radio, television). 7/
In the digital era, we still have communications networks (e.g., FB Messenger, WhatsApp, GroupMe, Skype), marketplaces (e.g., Amazon Marketplace, eBay, Uber/Lyft, Airbnb, GrubHub) and media — it’s just become social (e.g., FB, Instagram, YouTube Twitter, Snapchat). 8/
We also have platforms (e.g., Android, iOS, Windows, Xbox) which create ecosystems in which developers extend the platform itself by inventing new apps/games for users. Users and developers create both same-side and cross-side externalities for each other. 9/
And then there are data networks (e.g., Waze, Netflix, 23andMe, Google Search). @mattturck defines them as “effects [that] occur when your product, generally powered by machine learning, becomes smarter as it gets more data from your users.” 10/ mattturck.com/the-power-of-d…
Now, multi-sided platforms/networks/marketplaces are hard to get started because they suffer from a chicken-and-egg problem caused by the interdependence of demand between the customer groups. Who wants to join a two-sided market with no one on the other side? 11/
One way companies can bootstrap their networks is by having a “single-player tool,” aka some value proposition for users unrelated to network effects. @cdixon calls this the “come for the tool, stay for the network” strategy. 12/ cdixon.org/2015/01/31/com…
To take one example, Instagram started with cool photo filters that were useful on their own and later hit hypergrowth once enough people were sharing their photos with friends for network effects to ignite. 13/
A platform reaches “critical mass” when usage benefits exceed usage costs for a large # of potential users. At that point, the flywheel of network effects kicks in, and growth hits the knee of the S-curve. @trengriffin has a great post on this concept. 14/ a16z.com/2016/03/07/net…
The hope for VC’s and entrepreneurs (and the worry for antitrust regulators) is that once a platform hits critical mass, the market will “tip” and become winner-take-all. If the winner is a first-mover with an inferior product, that could be bad for long-run innovation. 15/
But there are many reasons to be skeptical that a particular market will actually become “winner-take-all.” @ce_tucker has the best paper summarizing why. But I’ll briefly list a few reasons here... 16/ sites.bu.edu/tpri/files/201…
Reason #1: In the desktop PC era, Windows/Intel really did have a lock on users and hardware OEM’s. But the proliferation of physical devices and digital platforms has dramatically lowered switching costs. 17/
In fact, in many of these markets users “multihome” (i.e., they use multiple competing services simultaneously). For example, the average internet user has about seven social media accounts. See excerpt from @David_Evans1 and @DickSchmalensee. 18/ object.cato.org/sites/cato.org…
Reason #2: Network effects are often local, not global. People care about their small network of friends and family much more than random strangers so network effects can diminish rapidly with size (see this great chart by @JamesCurrier). 19/ nfx.com/post/network-e…
Reason #3: Network effects can be negative, too. Congestion, spam, competition between sellers, and advertising load can all make the value to end-users decrease with the size of the network. 20/
Reason #4: Users also have an incentive to go off-platform for recurring or high-value transactions. 21/
But even if a market becomes “winner-take-all,” the winner might be an open and interoperable format (e.g., DVD, VHS, ethernet, x86 architecture) rather than a proprietary platform. 22/
Lastly, given all of these caveats (and the fact that FTC hearings begin today), I want to discuss antitrust implications. Regulators would be wise to take each market on a case-by-case basis and follow three general policy recommendations 23/
First, they should consider what type of network effects are present (positive? negative? increasing? diminishing?) and how they affect market power. 24/
Second, in this framework, the SC’s Amex decision would apply to marketplaces and platforms and not data networks or communications networks. Social media networks are an edge case. 25/ niskanencenter.org/wp-content/upl…
Third, stick with the Consumer Welfare Standard. It’s flexible and empirically grounded enough to handle platform business models. Regulators just need to think about how network effects might influence consumer diversion ratios. 26/ justice.gov/atr/horizontal…
Anywho, check out the full blog post. FIN/ niskanencenter.org/blog/you-cant-…
PS further reading: Almost everything I learned about marketplaces, I learned from the articles in this epic @andrewchen thread...
and this slide deck from @a16z and @anuhariharan was foundational in my learning. So much good stuff in there. a16z.com/2016/03/07/all…
And here's the European Commission report that informed a lot of the network effect magnitudes in the charts: ec.europa.eu/jrc/sites/jrcs…
