Facebook's business model is based on violating trust. They collect information behind your back so they can sell it to their customers (advertisers). The current kerfuffle with Google over contact information is an excellent example of this: Facebook wants to extract your address book from Google, but it won't let you export your Facebook address book to someplace else (including Google) that you might find useful. That's because they don't care about serving you; they care about selling you. Their attitude is that they own everything you do. That's why I doubt I'll get a Facebook email account -- somewhere in the fine print will be something that says "we own all your email and can do whatever we want with it" (If I do get one, I'll be walking on eggshells every time I use it).
In The Decline of Facebook, I didn't mean to say that Facebook was currently in decline but rather that it will decline because it won't be able to change its business model. The first law of the Internet (which I am just now making up) is: "It's all about trust." Companies that violate trust might succeed in the short term, but the second law of the Internet is: "The Internet routes around trust violations."
Let's do a thought experiment to discover what Facebook's successor might look like, using for a first approximation upside-down thinking. Basically, whatever Facebook does, this experiment will go to the opposite extreme. In addition, this experiment will use what I talked about here: the design will only be about what serves the customer. If that means this company will be lean and low-income like Wikipedia, so be it.
I'll suggest a code name for the product: Bubbles. That's just what it feels like to me. The plurality suggests that you can easily make many of them (Facebook wants you in a single, neat box, packaged nicely to serve up to advertisers).
1. Facebook is centralized
A big company like Google can fight Facebook on their own turf. But to create something new and different that auto-scales, let's go to the other end of the spectrum and make it peer-to-peer. That way, no servers (or perhaps minimal servers) will be required as the thing grows. The third law of the Internet might be: "Size doesn't matter" and so, to conform to that, a product's success should not depend on whether it has dozens of users or millions; peer-to-peer allows seamless and un-bumpy growth that ignores size.
2. Facebook controls your experience
Facebook decides what kind of stuff you can put up, how you can post, etc. While it's very helpful to have fundamental structure and basic building blocks (and a default layout that you get unless you ask for something different), in the end the user should be able to control what they put up and how it looks.
People who create new designs can put them up for others to adopt -- again, this decentralizes the design and allows better designs to emerge organically rather than being centrally controlled (thus, the cost of experimentation -- and failure -- is cheap, which makes creativity easy and is essential for rapid innovation).
3. Facebook owns your stuff
Of course you should own your stuff. This is a no-brainer, and one of the big reasons that people might move from Facebook to Bubbles. When you post, you would be encouraged to use Creative Commons, and only retain copyright if you really want to.
Now we see an interesting possibility. Suppose this system gains traction, and people begin moving over in large enough quantities to cause Facebook to notice. Will they actually say: "you can't move your stuff over there because we own it?" Seems to me that would be a huge faux pas, and cause a much bigger exodus. But Facebook put that language in your contract in the first place, so I would be totally unsurprised at such a blunder.
4. Facebook wants you to be one thing
You have many facets, and many different kinds of relationships with people. Facebook defines a single kind of relationship: "friend"/"not friend." I can't believe they haven't thought of allowing you to categorize people into groups, so I can only guess it doesn't serve their business model. Bubbles would allow you to create a new Bubble for each group of people, to allow them to sign up by Bubble, for you to put them in one or more of your Bubbles, and for you to create Bubbles in which you (the owner of the Bubble) could be anonymous. Also, a Bubble could have multiple owners and, if necessary, different levels of control for different owners (somewhat like Unix permissions). Note the possibility for easy creation of discussion groups and the like. Again, the point is to do what you want, not what Facebook wants.
4. Open source, naturally
This is part of the trust issue. It doesn't depend on the success of a single company, bugs are far more likely to get fixed, and no one can get a vice-grip on the direction of the product.
The architecture of this system will have to be nothing short of brilliant -- every installation is both a P2P server and (at least to some degree) an application framework (but the right framework; flexible enough and at the same time focused enough). The design is definitely a good use of crowdsourcing.
Feel free to add your own ideas at to the discussion group, or add them over at Artima if you don't want to join the group.