> This raises an interesting question: should email addresses be private?
Yes and no. Both of them. As any powerful tool, email is going to be abused, like any other alternative would be when it will come one day.
Those services allowing creation of dynamic email addresses do their job (until they're banned, that's why I'm not mentioning them), however using them isn't automatic and most people don't even know about their existence. What if we then did upgrade email protocols to reflect current needs wrt privacy and modified existing mail servers so that they could create dynamic addresses when asked by a simple flag?
Example: I want to subscribe to a service from company XYZ, however I'm not sure how much I can trust them, therefore, when writing an email or filling a web form I can activate the option to create a new address that is tied to the recipient I'll be writing to, and will work as a dedicated proxy for my real address, that is, every mail I send to the recipient using my real address will be actually sent from the new dynamic address, then all replies to the dynamic address will be routed to my real one, but a field in its headers will always contain either a memo by me (example: "signup with XYZ") or the original recipient (example: "info@xyz_trustuswerenotspammers_yeahsure.com"). This way one can immediately spot whoever sold their address to others and blacklist them.
As said, those services work well but not being built in into mail servers and clients their adoption is quite restricted. I don't see why that function shouldn't be embedded in a new upgraded email protocol as the modification would neither be that hard nor consume any serious resource. I would however expect heavy resistance against the adoption, of course.