We should user test what we have so far on the forget button, so that we are better informed about how people actually use it. Specifically, I'd like to check if the »close all windows« part causes any problems, but the testing shouldn't be limited to that.

Primary research questions: Discoverability of interface. Intelligibility of icon and vocabulary. Mental model of expected outcomes and proposed timeframes.

I assume that this work is done, so I'll close the bug. Feel free to reopen in case I'm wrong :)