In Downtown San Antonio, they make a point of putting up signs that clearly communicate the bicycles are sharing the lanes. No sharrows that I could see in the downtown, but the signs suggest that people on bicycles are to be treated as vehicles. A research question I have been mulling around in my head is whether it's better to clearly communicate to cyclists that you're a vehicle (thus we may need to modify our existing vehicle signal timing) or is it better to treat them as unique vehicles that would need specialized infrastructure. In Portland, we've been trying both and in some cases indicating to cyclists that they need to use the pedestrian signals. My theory on this is that it sends a message that people on bicycles can selectively choose what mode they are and they take advantage of this in some cases and it will make our designs more challenging. One thing that is clear to me is that we do need to test some of these devices (nearside signals that are smaller) to understand how our population will respond to their use. We have done some very early evaluations of the bicycle signals at the Interstate and Oregon location, but I am not sure how well publicized that information is. Bicycle signal compliance is important thing that needs to be more effectively communicated.
One of the inspirations for this post was the Bicycle Research Statement that I was editing for the Transportation Research Board. The references in that document includes the article by David Gibson entitled: Making Signal Systems Work for Cyclists.