I think there are a lot of good conjectures out there that don't get as much publicity as they should.
I think it would be useful for the research community if there was a centralized repository for all open conjectures much like: http://www.openproblemgarden.org/ I think was intended to function.
The most difficult factor preventing this from happening in my opinion is getting users to actually submit their conjectures and getting readers to visit the site to explore conjectures.
MO is not purely intended for this (and some users believe it is off topic while others believe its on topic) but the point is that A LOT of open problems probably do get asked here and this is the closest humanity has had to a crowd sourced open problems repository.
If there existed an "open problem", "formerly-open-problem" tag that was appropriately added to questions containing a mathematical question that is not known to be solvable and have generated interest but not generated any resolution then it would be very easy to filter through all of MO and find the set of all open problems.
I think that there are people (well at least myself) who would be interested in further structuring and organizing this data and then making it publicly available so that the state of research in math becomes a least a little bit clearer.
Would MathOverflow be interested in something like this?