The idea is to model a standard 'genome annotation format' in the wiki (called GFF) and access that data using a generic 'server' that implements a standard communication format for those annotations (called DAS).
In this way the mass of DAS compliant clients that visualise and analyse 'genome annotations' can access the data in the wiki.
Each annotation is associated with a URL (according to the DAS standard), so users can be sent back to the wiki to comment on or correct the annotations that they see.
What I'm calling 'genome annotations' here can be things like SNPs coming from analysis of personal genomics data (like in SNPedia), or other biomedically important things that the community (of biologists) should be collaborating on.
DAS paper (quite light): http://www.ncbi.nlm.nih.gov/pubmed/11667947