The last few years has seen many attempts by researchers to break free from the shackles of (commercial) journal publishers. There's the whole-sale exodus that produced the ACM Transactions on Algorithms, as well as other journals, and technology aided creation of open-access free journals like the Theory of Computing. There's also a movement to create an open access computational linguistics journal, spearheaded by my colleague Hal Daume here at the U. of Utah.
Joachim Gudmundsson and Pat Morin have been investigating the feasibility of making such a journal for Comp. Geom, motivated by costs, and copyright issues with current journals. Here are posts one, two and three on the topic.
They've worked out most of the logistical issues involved in creating such a journal, and are now trying to reach out to the community to see what kind of interest there is. After all, the main currency of a journal is its reputation, and that comes from community participation (and then perception). So if you have any opinion on the matter, hop over to Dense Outliers, take the poll and post a comment (don't post comments here).
My personal view: I think open access is a great idea in principle, but I'm not seeing a pressing need within CG itself for such a journal at this point in time. (Disclaimer: I'm involved with the International Journal for Comp. Geom and Applications).