Something that I just can't stop thinking about: Value and cost

I keep thinking about the value of information. Value is important for information because it often determines cost. And as we all know, costs for licensed databases are extremely high. How did they get to be that way? How does a journal article go from something that someone essentially gives away to being in a database that costs thousands of dollars? Authors of journal articles generally hand over their material and copyrights to the journal publishers. The publishers then repackage the material into journals, both print and online. Those journals are then put into databases. So a question that I have is if this is the general process for scholarly article publication, then why are some databases so much more expensive than others?

For instance, many chemistry and physical science journals are notoriously expensive. What is so special about these journals, and the databases in which they are distributed? Now, I understand that publishers can make some claims that they add value to the journals. Yes, they might lay out the article and perform some copyediting (maybe). But what is so special about chemistry databases, over say nursing or library science databases (which are generally much cheaper)? Could journal publishers/database providers really be adding that much value to chemistry journals but not to nursing journals? It's really hard to believe.

I also keep pondering the notions of excludability and rivalry. Usually information goods often exhibit characteristics of non-excludability (that is, you don't exclude others from using the good) and non-rivalry (that is, you can share the good and it doesn't get used up). Licensed databases really push information goods into the realm of excludability: if you don't pay for this information, you can't have access to it. The information generally will not lose value if more people know it (although, that really depends on the situation, and would not be applicable to trade secrets for instance.) So we can still share this information (i.e., scholarly journals) without all the knowledge being used up, but now we are strictly delimiting who has access to that knowledge. (This is starting to remind me a lot about that Open Access presentation I did last year...)

All in all, these thoughts bring me back to this idea of value and cost. Information can be free and easily shared. So if that is the case, why are these databases so expensive?