If you imagine email being (as it is) a “network” with addresses, senders, receivers and the normal paraphenalia of a network (routers, etc.) then the question of location of the “spam-filtering” function should be guided by the E2E principle, ie, if a function can be performed at the end-points of the network, do it there.

Spam filtering can be done by individual users, thus E2E would recommend an end-point solution.

Latest statistic I’ve seen is that 11.7% of valid email is currently not delivered due to “ham-fisted” filtering outside of the control of the user. Certainly the user should be able to cede control of spam to another entity, but it should remain possible for them to do it themself, their way, part of the optionality value of such systems.

Spam is in the eye of the beholder, like a lot of things. The historical solution is centralised filtering, censorship for example, but as we move toward a future where decentralisation is an option, spam can be a “customer owned” problem, and solved collaboratively by multiple sets of like minded individuals.

In all these discussions about “spam” it’s worthwhile to substitute “porn,” “copyright material” or “criticism of the Government” as the subject, because it seems to me, the day you define and control “spam,” everything will be controllable with the same technology.

Our demonisation of spam might obscure the fact that it’s existence is part of the cost of participation in an untrammelled communication network.

Removing spam might cost us that.