It's not so obvious. My (naive?) take on that is that if control over data is something people want, you give it to them. You make it hard for them to leave your service by excelling over the alternatives. That's the sane default anyway; I understand there may be cases where it makes business sense to sell your soul.
But note that barriers to exit are barriers to entry too. Data lock-in is the main reason why I don't consider desktop or web apps for things such as calendar and to-do lists. I use plain text (or YAML) files, which I maintain with a combination of hand editing and Python scripts. This also makes it easy to keep this data under version control, and back it up.
Since Facebook et al are not targeted for a technical audience, maybe they don't get enough requests for data export tools.
You're right about the barriers to entry part. But barriers to entry are low enough for relative first movers like Facebook/Ning. When the market matures and there are several clones of these services, Facebook/Ning will have to rely on the lock-in to stay ahead of the competition.
There is a general trend in users' demanding openness in data, and maybe this will have to be a core feature of any new web-service/desktop-app to succeed in the future. But the general rule is that a business won't open the data if it doesn't have to.
But note that barriers to exit are barriers to entry too. Data lock-in is the main reason why I don't consider desktop or web apps for things such as calendar and to-do lists. I use plain text (or YAML) files, which I maintain with a combination of hand editing and Python scripts. This also makes it easy to keep this data under version control, and back it up.
Since Facebook et al are not targeted for a technical audience, maybe they don't get enough requests for data export tools.