Many readers of this blog doubtless have heard Chris Anderson preaching why Free is the Future of Business. His basic ideas are that the cost of creating digital goods continues to fall, while opportunities to make revenues from advertisers continue to rise. He further suggests that once these dynamics drive the price to zero, or so cheap that the price is negligible, users will start wasting the zero-priced-goods, and that this waste eventually leads to ideas that change the world.
I believe that one area where Anderson's ideas will become increasingly relevant is in the syndication of web services. Today, hundreds of sites expose their APIs for syndicating their content. Clearly, there's a cost associated with syndication: each API call costs hardware resources and bandwidth.
Now imagine someone using the search functionality in these APIs to display search results in real time: let's say site A is making these calls on an API provided by site B to display site B's results on site A. At first glance, this seems like a fairly expensive operation to perform, when you could instead index the contents of these sites. But consider that site A also displays snippets of results from these sites which induce users to click through to site B which gives site B an opportunity to monetize this traffic.
Using the API, instead of creating an index allows you to 1) access the underlying data of the site, instead of accessing just the web pages created using the data 2) access the metadata of the website 3) keep up to date with the latest data which's a difficult task if you're indexing. These advantages help provide a better user experience, and hence increase the usage of these sites.
Such a syndication deal is profitable if the average revenue per visit is greater than the the cost per visit. The cost per visit is the inverse ratio of click-through-rate times the cost of serving one request on the API.
If you apply Anderson-ian economic principles, you'll see that the average revenue per visitor is on the rise for almost every internet site, and the average cost of serving one request is on the decline. Which implies that web services will be syndicated much widely in the future.
What could happen in the extreme scenario where the syndication is totally free? Well, for one, anyone could build a search engine without building an index of the web: For every user query, you'd have the luxury of calling Google's API hundreds of times (in addition to Yahoo's and MSN's) to retrieve many different kinds of data from Google's index. You could then run algorithms on top of this data to pick the best search results, package them and present them to the end user. Or, if your algorithms are a little faster, you could directly call the APIs of thousands of websites in real time, and scan through them.
Pretty nifty, huh? Obviously, you still need to build ranking algorithms that are global, and having your own index helps you do that better than accessing external APIs and trying to rank disparate sources of data in real time.
Also, syndication will probably impose conditions on the syndicating site to provide some business benefit to the originating site: e.g., drive profitable traffic back to the site.
Obviously, this is an extreme scenario, and not likely to happen soon. But the trend is real, with Yahoo's BOSS being the significant attempt in this direction. I do believe that all internet companies need to watch out for this trend, both to defend themselves from the upcoming explosion of web service syndication, as well as to reap the opportunities in this new world.
Comments