How do you authorize a new device in an "easy and secure" way without simply outsourcing the problem to an intermediary who is then in a position to attack you by authorizing its own devices?
This issue has quite concrete implications for the security and convenience of lots of existing security tools, from GPG to iMessage to Skype to Firefox. They've chosen different approaches but the underlying problem and associated tradeoffs apply to all of them.
On the bright side, there are now a lot of people exploring the space of possibilities for dealing with these tradeoffs.
