It looks like we've been affected by some major sea level rise here in Coastal VA/NC. 
I'm wondering if there is an issue with the coastal shapefiles.
anyway, great work. That website will be very useful.
The water shapefiles are directly from the Census Bureau' s 2009 shapefiles for coastal counties and counties with significant lakes and rivers, filtered to remove smaller water features. Areas like the Florida Keys and Cape Hatteras don't show up on the regional maps because the islands themselves are too small to show up on the large maps. They appear on MapWindow GIS when properly zoomed in.
I'm not sure if Census considers swamps as water features, which might explain why parts of Louisiana and North Carolina get swamped in Light Cyan.
It almost looks like there's some sort of buffer around it that makes the bays and inlets wider, leading to places with complicated inlet networks to appear to have less land area than in reality, and peninsulas ending up as tiny islands or disappearing.
There is a line around the water features. It is colored Light Cyan. Perhaps that is doing it. I'll try to set the line width to zero and add a transparent water layer, like I did to get rid of the precinct lines, but I think that just creates a line that's the same color as the mapped color, so things won't change.
I wish Dave's redistricting app had stuff for Oregon. 
You should ask user James Allen on Daily Kos about that. He has a data set he's been adapting to Oregon's DRA file for some time, though I don't think it's complete yet.
The problem with Oregon is not the lack of precinct-level data. Harvard's Election Database has it. It's the lack of a VTD or precinct map. Oregon outside of Multnomah County didn't participate in the VTD program. And Multnomah County's VTDs were so large as to be useless. Theoretically, results could be reported by block group, like in California and Rhode Island. But without precinct maps, it's hard to translate data down to the block group level.
Montana's issues are similar to Oregon's.
