For the last year, Sergey Brin has been campaigning at Google with the slogan "features not products" in an attempt to reduce the sprawl of the company's eccentric creativity.

This campaign may have gone too far, at least when it comes to Google's new OpenSocial initiative. Google's new programming interface that allows social networks to communicate with applications is a nifty add-on feature, but it's not a product. And thus it doesn't compete with Facebook, despite the dozens of blog posts that say it does.

I can't believe OpenSocial, in itself, can revive Google's Orkut social network, which is moribund most everywhere except Brazil. That said, if there are developers who write applications in Portuguese, Orkut could become even more popular and more widely used in Brazil than it already is.

Think about Facebook. If you like Facebook, it's still because of its core features — the way it lets you communicate and keep tabs on your friends. If you like any of the new applications, it is because on the margin they let you do what you already like doing a bit better. You can throw a sheep at someone instead of poking them. I haven't seen a Facebook application so compelling that you would join Facebook just to use it.

The frenzy about open platforms misses an essential truth: no one will go through an open door if there isn't something worthwhile on the other side. The best example of the sort of platform that works is Google Maps. It is a powerful, flexible way to display geographic information. When combined with interesting data, the results can be compelling. (Here is the Los Angeles Times's wildfire map. Here is a blog with many more examples.)

Social networks do have interesting data that could be used in a mashup application: Profiles of members, and links that define the relationships between members. But this data is only useful if the network has deep penetration of the people users care about. I can see why Marc Andreessen's Ning, which helps create custom social networks for little league teams and such, might find OpenSocial useful.

Open, moreover, isn't entirely a good thing when it comes to social networks, which must keep some information private and protect their members from harassment. You can bet that teams of spammers, phishers, and overeager Madison Avenue executives are already poring over the OpenSocial specs to see how they can ply their trades on social networks.

One of the strengths of Facebook is that its platform has a strict set of rules. The company controls what applications look like and where they put ads, and has very careful controls about how they get and use information about its users. And the site has kept tweaking these rules to react to complaints by members who find some applications too intrusive.

Google appears to understand this, and has proposed that OpenSocial enable, but not require, communication between applications and social networks. So LinkedIn has said it will allow some OpenSocial applications on its site, but it is firm that there will be no food fights. But controlling behavior of many players in a more distributed environment will be more difficult. E-mail is the most open social application, and it is filled with problems because there is no way to enforce rules on all of the participants.

LinkedIn has also said it may ask for a revenue share from those who create applications for its network. And haggling over money, no doubt, will slow down the deployment of many social applications.

My point here isn't that there is something wrong with OpenSocial. But it is a piece of raw technology. And like any technology, it can enable good products, when combined with good design, marketing, business relationships, and usually luck.