The main reason for that is that there is an established reputation system for papers, namely journals, but there isn't one for content formats that aren't papers - data-sets, code, videos, images etc.

Part of what Academia.edu is doing is building a new reputation system in science based on audience metrics: how many people have read your papers, from which countries; how many followers you have etc. We are then attaching various kinds of scientific media to that reputation system in order to accelerate sharing.

The idea is that if you can build your reputation by sharing your data-sets, you're likely to consider doing it.