Network effects are central to our investment thesis at USV. We believe that they provide one of the few (the only?) source of sustainable competitive advantage. We have also been outspoken about wanting to limit the power that these network effects provide to winning companies. One proposal that I have made is to shift more power to the edge of the network by giving humans the right to be represented by a software bot.

It is useful to think about how contestable a market with network effects is. The concept of contestable markets was first developed by Baumol. The idea is that even a market with relatively few companies (including possibly a single dominant company) could be quite competitive *if* there is a credible threat of new entrants. The three characteristics of perfectly contestable markets are

1. No entry or exit barriers

2. No sunk costs

3. No technology advantage for incumbent

It turns out that this increasingly describes many digital markets.

Consider messengers as an example. Even though messengers have a strong network effect, with cloud computing and rented hardware there are very few entry or exit barriers, sunk cost or technology advantages for incumbents. As a result it has been difficult to monetize messengers (with the proviso that if they achieve deep integrations such as WeChat, then there are barriers to entry).

Messengers have an extra challenge in that they (largely) deal with ephemeral content. Between having the graph in the phone book, no investment in past content and minimal learning curve, endusers are free to explore new services. Incidentally this could also turn out to be a challenge for Snapchat as there is no archival content (yet).

On demand car services, such as Uber, Lyft, etc. may turn out to be locally contestable. It is unclear how much of a cross city network effect exists (despite what we as frequent travelers might think – and even among frequent travelers many go to only a few cities, so can easily have a few apps installed). There are local players competing already and new entrants being started including Juno in New York city and the Waze Rider in Tel Aviv.

Protocols, such as say SMTP for email, make markets extremely contestable. Despite the dominance of a few email providers the services are free and even advertising monetization is minimal. This means that other existing network effects are susceptible to increased contestability should standards emerge. That’s why work on standards for messaging, transportation, payments, etc. is so important and new blockchain based protocols are worth exploring.

Network effects are central to our investment thesis at USV. We believe that they provide one of the few (the only?) source of sustainable competitive advantage. We have also been outspoken about wanting to limit the power that these network effects provide to winning companies. One proposal that I have made is to shift more power to the edge of the network by giving humans the right to be represented by a software bot.

It is useful to think about how contestable a market with network effects is. The concept of contestable markets was first developed by Baumol. The idea is that even a market with relatively few companies (including possibly a single dominant company) could be quite competitive *if* there is a credible threat of new entrants. The three characteristics of perfectly contestable markets are

1. No entry or exit barriers

2. No sunk costs

3. No technology advantage for incumbent

It turns out that this increasingly describes many digital markets.

Consider messengers as an example. Even though messengers have a strong network effect, with cloud computing and rented hardware there are very few entry or exit barriers, sunk cost or technology advantages for incumbents. As a result it has been difficult to monetize messengers (with the proviso that if they achieve deep integrations such as WeChat, then there are barriers to entry).

Messengers have an extra challenge in that they (largely) deal with ephemeral content. Between having the graph in the phone book, no investment in past content and minimal learning curve, endusers are free to explore new services. Incidentally this could also turn out to be a challenge for Snapchat as there is no archival content (yet).

On demand car services, such as Uber, Lyft, etc. may turn out to be locally contestable. It is unclear how much of a cross city network effect exists (despite what we as frequent travelers might think – and even among frequent travelers many go to only a few cities, so can easily have a few apps installed). There are local players competing already and new entrants being started including Juno in New York city and the Waze Rider in Tel Aviv.

Protocols, such as say SMTP for email, make markets extremely contestable. Despite the dominance of a few email providers the services are free and even advertising monetization is minimal. This means that other existing network effects are susceptible to increased contestability should standards emerge. That’s why work on standards for messaging, transportation, payments, etc. is so important and new blockchain based protocols are worth exploring.