Thursday, December 06, 2007
Lets Build An API
Are you feeling a déjù vu? As always, we need to watch not only the picture but the full movie to understand what is going on.
As Jared Spool said, every market goes through four steps (in this order): Technology, Features, Experience and Integration.
We are now in the Experience phase. Web 2.0 is all about user experience, tell that to 37Signals. However, in the software industry there is an always subjacent war about platforms. Windows, Mac, Office, Google, SAP, Facebook, Data, Html, Browsers... everyone want to be the winner platform, and in this way, they have to open the doors for developers to empower their position, and so, they make a public API. Each day, someone else is entering in the API game, and this is good for all of us.
But the real interesting point are the assumptions in the background.
If you build an API and someone else can replicate your software... which is your competitive advantage? You get the database, the raids, the data center, and the 24x7 problem and the third party the advertisement? Actually, in my modest opinion, there are two very interesting assumptions: the URL fidelity and that data remains in your side. And you know, Facebook is living from its three-tables-database (Users, Friends and Applications) and the gift-card they received from Microsoft to send a message to Google. For this reason, you and many engineers in these companies may think: "we won't put this data available through the API because this is our key data".
The URL fidelity is a very interesting topic. 15 years ago nobody knew about urls and now you can't get a decent one. What will happen in 10 years with urls? Who cares? In 10 years we will be in Web 7.0.
However, the point about the data isn't so clear anymore. This was ok until someone wrote a robot capable of stealing all your data (slides), and now, you can steal and then store it somewhere (it would be very funny to take amazon data and save it back in the EC2, would you join the project?).
But here is the funniest part of the history: they don't need a public API. They are using HTML as your public API and taking all the data they need, to do whatever they want. You can also use this robot and be sure I will be doing it in the future. Hey! It should be illegal! Don't tell me, and what do you think Google has been doing from the beginning? Ok, they call it crawl (not steal) and don't save it in a relational database but in a home-made File System, but this is just tech-stuff.
You know my point of view, users should own the data, or at least, not just [Put Your Favorite Company Here]. I would love to hear your opinion...
Labels: Internet, Platform, Technology
