James’s Musings

thoughts, photography, and geeky stuff
from an unrelentingly curious Silicon Valley entrepreneur

It’s the API, Stupid! How to Crowdsource Your App Ecosystem

APIs Are Cool.  Who Knew?

by James Beldock on May 6, 2012

In the ’90s, everyone knew that BizDev was the key to success in Silicon Valley. What will be this decade’s BizDev? The API. Twitter was propelled to early success by leveraging a huge community of developers virtually none of whom were actually employed at Twitter. How did they do that? They crowdsourced. Twitter built a solid API and evangelized that API throughout the red-hot San Francisco consumer and social media developer community. (For the uninitiated, an API, or Application Programming Interface, allows other software to interact with your software, without human intervention.) Virtually overnight, the API supported not just the majority of Twitter traffic, but within a year, fully 91% of Tweets came through the API. You would be justified in answering “What is Twitter?” with “It’s the API, Stupid.”

Here’s why APIs matter: you crowdsource your app ecosystem. Developers who like what you’re doing and have users who would benefit from mixing a little of you with a little of them will grab your API, mash it up with their app, and voilà, they’ve spawned new members of your ecosystem. Your success is now multiplied by their success. How multiplied? Well, in the case of Twitter it’s literally thousands of Twitter-enabled apps, including one or two you’ve no doubt heard of (the iPhone, for example: Twitter is built into iOS 5). If you’re my favorite tool for reading things when I have time, Instapaper, you have 140 Instapaper-enabled apps you can point to.

It’s not just that social networks have taken over the web. It’s that social networks have taken over software. Do it right, and as your platform goes viral, you become the single repository for your particular content and data, while simultaneously process of using your content and data is crowdsourced by your ecosystem. Your value increases not just as your user base increases, but exponentially as your product is integrated into those of others. (How much does it increase? I’m glad you asked. See The Math Part below.)

What happens when we take an app or social platform and give it an API that lets other apps or social platforms leverage that network? Nitrous. Each app that connects to your platform gives your platform access both to additional users and to additional data. Of course, these mathematics (see The Math Part) further underscore why tech has become such a winner takes all world: there is limited screen real estate in all those apps in the ecosystem. They will likely have room for only one preferred provider of whatever your platform is good at. That had better be you, or you miss out.

The winners have figured this out. Google offers 96 APIs; Microsoft has more than 30; Yahoo has over 50. And the old world companies? They’re getting it to. The New York Times offers 14; AT&T 9; Ericsson 16.1 Amazon.com’s Amazon Web Services takes their API supremely seriously, and there are entire companies, like Eucapyptus, betting on them to continue doing so. Foursquare innovates with their API regularly enough that you use the date you are writing your code to access their API, to ensure forward compatibility. Programmable Web thinks there will be 5,000 any day now:

Programmable Web API Growth Chart

So, if you’re starting a tech company, lead with your API. Treat it like a product. Release it regularly. Advertise it. Promote it. Even build your company around it. For it will exponentiate your value.

The Math Part

OK, you asked for math. There are three laws often used to describe the value of a network: Metcalfe’s Law, Reed’s Law, and the somewhat more complicated Beckstrom’s Law. 2 Metcalfe’s law is quite straight forward: it says that the value of a network of n nodes is simply n^2. Simple enough. One person with a fax machine? Useless. 10 people with fax machines? At least 100 times more useful. Reed takes Metcalf a step further and points out that it’s the connections among users that are scaling, and therefore that a measure more like 2^n is more appropriate, because the combinatorics permit so many permutations of connections (e.g. of the n users, two of the n might be in connection for one purpose, three of the n, etc., and then a different three of the n might be in connection for a different purpose). Beckstrom’s law is more complicated, so I’ll just quote it later and let you read his explanation.

How much value does this add? Certainly for a given application, that value is proportional to the size of the other application’s user base (let’s call that q for each other application). And let’s call the number of other applications that leverage your platform p. Depending on whose network value law you like, I propose that the value of the network increases by a factor associated with the number of users of each of those other apps that leverage your platform, which would be expressed as c\sum_{i=1}^pq_i, where c is some scaling constant that likely changes dramatically depending on whether your product is social in nature or not. Here’s my proposal for modifying the three laws:

Beldock’s Law
More properly, Beldock’s Corollary to the Network Value Laws
Network Value Law API Impact on Network Value
Metcalfe’s Law n^2 \displaystyle \left(n^2\right) \cdot c\sum_{i=1}^pq_i
Reed’s Law 2^n \displaystyle \left(2^n\right) \cdot c\sum_{i=1}^pq_i
Beckstrom’s Law \displaystyle \sum_{i=1}^n\sum_{k=1}^m\frac{B_{i,j,k}-C_{i,j,k}}{(1+r_k )^{t_k}} \displaystyle \left(\sum_{i=1}^n\sum_{k=1}^m\frac{B_{i,j,k}-C_{i,j,k}}{(1+r_k )^{t_k}}\right) \cdot c\sum_{i=1}^pq_i

So there you have it. Treat your API nicely. It’s worth a cool c\sum_{i=1}^pq_i!

Photo Credit: Thomas Hawk, Sept. 14. 2006, licensed under Creative Commons. Slightly modified.

  1. See the great 5,000 APIs post from Programmable Web. []
  2. Credit to friend, serial entrepreneur, RapLeaf CEO, and Founders Fund Venture Partner, Auren Hoffman, for first giving me a copy of The Starfish and the Spider, which Rod Beckstrom co-wrote, for first getting me to think about the mathematics of these phenomena. As usual, Auren was a step ahead in realizing how important these scale equations would become. []

Comments on this entry are closed.

Previous post:

Next post: