Commoditizing Dog Food

It’s always refreshing to see someone take their own advice (aka eating their own dog food).

Last week Facebook announced their Open Graph initiative.  It’s their hugely ambitious attempt to take over the web.  In a nutshell, the idea is that you Facebook will be used to log you on to all websites and all your preferences (which bands you like/hate, etc.) will reside with them and in return they’ll personalize everything for you (both content of sites and ads).

It’s a great idea as long as you trust Facebook with every private aspect of your life. And you don’t mind if they share it with their development partners. And you trust them to delete your info if you like something you shouldn’t have. And you don’t want to use any competing service that might be improved based on your preferences (you can bet Facebook won’t share your data with them).  In short, it’s a closed universe and those work really well right up until they don’t – usually when you try to do something someone in the corporate Politburo didn’t want you to.

So, what to do? Well, a group of interesting folks have started OpenLike, an open source project to come up with a standard for ‘liking’ objects on the web (that’s the mechanism for capturing your preferences).

Are these folks doing it out of the goodness of their hearts?  Probably in part, but more likely they’re focused on “commoditizing the complement“.

The notion is simple: you want to drive down the cost of anything that augments your product/service.  Joel Spolsky wrote the definitive treatise on this.  While the notion is simple, identifying a complement isn’t always obvious and that’s what makes OpenLike so interesting.

If you look at the folks behind the initiative, one of them is Chris Dixon, who wrote the blog post about complements referenced above.  He runs Hunch, a site that gives you recommendations and gives you better results as you and others use it.

Hunch’s value lies in their algorithms: the more data they have about you – most notably your preferences about what you like/don’t like – the better their algorithms work.  In other words, preference data is the complement to their algorithms, so they’ve got a massive interest in making it as cheap as possible for people to give them their preferences.

Facebook is threatening to make it infinitely expensive (they’ll almost certainly never give Hunch that data), so they’ve got a fire under their butt to consider doing something else.  They could try buying the preference info from users, but that’s typically not a good way to run your business (ask Microsoft who is offering cash back on search and losing a fortune).

They’re focused on building a killer product that will offer enough “value” (I put it in quotes as it’s such an elusive term) to users for them to want to give them their preferences.  But that’s not enough when you’re going up against the web-wide distribution of an industry leader with over 400 million users.

Hence OpenLike.  If they can provide an alternative to Facebook’s platform they provide a way for themselves to thrive.  Publishers are happy as their content can get shared to more places on the web.  Hunch is happy as they can get more preferences.  And open web advocates are happy as the user is back in control of their preferences rather than one private company.  Plus it’s always nice to see people like Chris walking their talk.

On another note, this will also be a small win for Google, who will now have a search entry for the exact phrase “commoditizing dog food”: