The first category that you identify may be defined too broadly. The term “bundling,” suggests the tying of non-zero marginal cost goods to zero-marginal cost goods either literally or technologically. For example, creating videogames coded to play only on a proprietary platform. This kind of activity is different in kind from simply laboring to produce something that people will feel motivated to pay for due to feelings of “charity, community, authenticity, and emotional security.”

The second category that you establish seems to suffer because it does not distinguish between creating a compelling distribution system and compelling people to use your distribution system. Updates that create strategic incompatibilities coerce the customer in a way that a simple offer to distribute content efficiently at a price does not.

The third and fourth category that you identify could probably be collapsed into one category. Every content provider that is in the advertising business has a strong incentive to mine data in order to increase the value of its advertisements. This fact begs the question of who really provides content with ads without collecting any user data.

Finally, I agree with your general point that data mining can create a desirable kind distributional efficiency if it is used by content distributors to reduce consumer search costs. I do, however, recoil in fear at the thought of Facebook being able to correctly suggest who I should “friend” and where I should go to meet my new friend for lunch in order to have a meal that we will both enjoy. Are you really willing to embrace such a future?

-- StephenClarke - 19 Nov 2009