Obviously, you are trying to avoid iterating/sorting all vertices to find recent questions. One way might be to build some sort of date structure into your schema like:
year --> month --> day --> question
If you have a sufficient number of "questions" in your you use case you might consider breaking time down further to hours, minutes, etc (or a higher level of aggregation....maybe you just need year and month). Index the edges between the year, month, day, and question using a reverse sort order of the time, so that you can just do:
g.V('year','2014').out.out.out[0..<10]
which would return the first 10 most recent questions. Note that Gremlin nicely compiles this to a vertex query to take advantage of your index:
gremlin> g.V.has('year','2014').out.out.out[0..<10].toString()
==>[GremlinStartPipe, GraphQueryPipe(has,vertex), IdentityPipe, VertexQueryPipe(out,[],vertex), VertexQueryPipe(out,[],vertex), VertexQueryPipe(out,[],range:[0,9],vertex), IdentityPipe]