Develop bridge image for the cloud
When we first started working on the bridge design, we were thinking to ourselves "we should get lots of bridges, so China can't block them all." That was wrong. Instead we should be thinking "we should make sure the rate of getting new bridge addresses exceeds the rate that China can block them." The key resource that bridges need is changeable IP addresses. So we should experiment with easy-to-set-up images for the cloud, so people can pop up a bridge, and then discard it once it gets blocked.
Step one is to set up a bridge on the cloud (say, Amazon) and run it for a little while, to make sure there aren't any stupid things making this harder than it sounds.
Step two is to learn more about the pricing structures: is baseline time cheap but bandwidth is expensive? Or CPU? Etc. How much money are we talking here, for a variety of bridge scenarios? Evaluate a variety of cloud providers.
Step three, investigate programmatic "get new IP address" cloud functions we can use. In the future (e.g. #1851 (moved) or others), bridges will be able to automatically discover that they need a new IP address. The crude approach would be to abandon the bridge image and start up another one next door. The better approach would be to teach Tor how to press the "new IP address please" button on its host.
Step four is to learn more about automation. What are the steps for making it so you can tell other people "just launch image Z and you'll be running a bridge"? Do these steps and make it so.
Step five is to write the howto for groups like RFA who want to ask people to run bridges for them. Make sure to resolve usability pieces like "should my bridges publish in bridgedb or not".